VMWARE MANAGE VSPHERE ESXI STROAGE USING VCENTER SERVER PART 3

SERIES

Watch VMWARE MANAGE VSPHERE ESXI STROAGE USING VCENTER SERVER PART 3 - SERIES on Dailymotion

Transcript

00:00:00And these aren't necessarily enumerated steps. Just to let you know, I always do this step first, okay?

00:00:08I don't think there's really too much point in setting up the adapter until you've done the networking,

00:00:12but VMware always presents them in this order, so I sort of stuck with that.

00:00:17The order is sort of irrelevant. In order to activate the adapter, you have to configure the adapter as well.

00:00:23And to configure it, you sort of have to get in there and have the network already set up.

00:00:30So, sort of either order here, but if you want it to work, you have to set up the networking anyway.

00:00:37Optionally, you can also enable jumbo frames.

00:00:41This, of course, we know if you're into networking, Ethernet defaults to 1,500 bytes for an MTU.

00:00:48VMware supports up to 9,000 bytes for its MTU.

00:00:53What we usually call jumbo frames.

00:00:56Of course, the advantage here, well, there's a couple, right?

00:00:59The more data that you put into a frame, the less frames you need, the less packets you need.

00:01:06And, of course, the less packets means less headers.

00:01:11So, overall, your communication gets more efficient, okay?

00:01:16So, if there's fewer headers, that means less wasted traffic, right?

00:01:24Headers are great.

00:01:25We need headers to get traffic from point A to point B.

00:01:28But at the end of the day, the headers just get thrown away.

00:01:31So, it's a necessary waste, but headers are a waste of traffic, right?

00:01:39So, the fewer headers we can have for the same amount of traffic, the better.

00:01:44Not to mention, of course, you have to calculate CRCs and all that kind of fun stuff.

00:01:50So, it can also save CPU overhead to have jumbo frames as well.

00:01:56Now, they are completely optional, okay?

00:01:59Keep in mind that jumbo frames would, of course, have to be supported end-to-end.

00:02:04So, the underlying switches have to support jumbo frames.

00:02:10And we'll talk about this a little bit more on the networking side in the networking course.

00:02:16But you have to enable this for the switch.

00:02:18You have to enable it for the VM kernel adapter, the kernel port that we created above here on step two.

00:02:26We have to enable it on there as well.

00:02:28So, as we go through and we set this up, we'll see where to do it.

00:02:31Again, I'm not going to enable it today in our lab simply because, you know, we're not too worried about performance.

00:02:38And we're not going to be doing throughput tests or anything like that.

00:02:42But normally, I would encourage you, if you're going to use iSCSI, to enable jumbo frames.

00:02:49Keep in mind, it means your storage array has to support it as well.

00:02:53So, again, it has to be the whole way through, end-to-end, okay?

00:02:56Now, it's not going to hurt if you enable your infrastructure to support it, but you haven't turned on jumbo frames yet.

00:03:04But, you know, so what I'm saying is, like, underlying switches.

00:03:07You could enable jumbo frames on your switches and still not actually be using them yet.

00:03:11That won't harm anything, but it's generally not just, like, you know, a step towards it.

00:03:17Step towards it.

00:03:18Okay.

00:03:19Your next step is on the adapter.

00:03:20You have to tell it how to learn about its iSCSI targets.

00:03:27ESXi is going to be acting as the initiator, which in server terms, that would be, you know, client-server.

00:03:36The initiator is the client.

00:03:39The target is the server.

00:03:42So our SAN is the target.

00:03:44ESXi is the initiator.

00:03:46We have to tell the initiator how to learn about the target.

00:03:53So we can either statically configure that, which, of course, would involve us knowing all that information, or we can dynamically configure it.

00:04:03And the cool thing is, like, if you're building all this, like, for the first time, and you know, okay, I've only got one LUN so far, and that's, you know, my LUN for this.

00:04:15You can tell it to dynamically discover, and then make that dynamic discovery a static map.

00:04:22In fact, the kernel will do that for you.

00:04:25And then take the dynamic discovery back out.

00:04:28So, you know, if you don't want to leave dynamic discovery on for whatever reason, generally speaking, I just leave it on dynamic on most of the stuff I set up.

00:04:39Of course, the fear, danger, whatever you want to call it, is with dynamic discovery on, it could go out and grab a LUN that it's not supposed to have.

00:04:53Because dynamic just means, you know, it's going to send what's called an iSCSI get targets, and the SAN is going to reply with every single LUN that that host is allowed to get to.

00:05:06So, that may be LUNs you're not supposed to have.

00:05:12That's the fear, danger, whatever you want to call it.

00:05:16The flip side of that, of course, would be make sure that the array is not presenting LUNs that it's not supposed to.

00:05:23I prefer to go that method, but that's, of course, also assuming that you're in charge of both sides.

00:05:30So, if you're only in charge of the VMware side, you know, maybe you want to do static just to make sure that you're not connecting to something you're not supposed to.

00:05:40Because the last thing we'd want is for one host to try to connect to somebody else's boot LUN.

00:05:46If this is shared storage, it shouldn't be a problem, but if it's a boot LUN, that's bad.

00:05:53Okay, something else with iSCSI, one of our security features is that we can do CHAP.

00:06:04Okay, so you can do CHAP for authentication.

00:06:07You have a couple options for this.

00:06:09I'll take a look at the options as we're setting it up.

00:06:12We're not going to use any today, but there's a couple choices.

00:06:17VMware sort of has that approach of just to put the choices plainly without seeing them just yet.

00:06:26But basically, I can say, don't use authentication.

00:06:30Use authentication only if the other guy wants it.

00:06:34Always use authentication and, you know, don't connect if the other guy doesn't allow it or, you know, don't allow it at all.

00:06:42We'll see the choices as we go.

00:06:43But you have a couple choices with that.

00:06:47The other thing you can do is you can do mutual authentication.

00:06:52So, typically, with CHAP authentication with iSCSI, it would be the array or the target authenticating the initiator or ESXi.

00:07:08That's great.

00:07:09I mean, that, you know, protects the storage.

00:07:12That makes sure that the storage isn't being connected to by somebody who is not authorized.

00:07:18But the flip side of that is nothing's protecting ESXi from being inadvertently directed to the incorrect storage.

00:07:29So, mutual authentication is, of course, both authenticating both.

00:07:36Pretty straightforward.

00:07:37The big thing with that, it's a little bit newer.

00:07:41I mean, I don't want to say it's new.

00:07:42It's been around for a while.

00:07:43But it is newer.

00:07:45However, some of your arrays may or may not support having the client or the initiator challenge them for authentication.

00:07:57They may not know how to handle that.

00:07:59And if you force that, it could...

00:08:02I'm just throwing out there.

00:08:03It could potentially break things.

00:08:07You have to make sure that your storage supports that option if that's what you want to do.

00:08:13It's a good option.

00:08:14It adds, of course, a whole other layer of security, which is always a good thing.

00:08:18But you have to make sure that it's fully supported on both sides.

00:08:23iSCSI, of course, has to run over a network.

00:08:25This is running over TCPIP.

00:08:29And, you know, some of the advantages to this, just before we move on, because a lot of people look at this and go,

00:08:34okay, so, you know, we talked about Fiber Channel.

00:08:36We're talking about iSCSI now.

00:08:39When should I use one versus the other?

00:08:42And, okay, so just, you know, real quickly here.

00:08:45Fiber Channel has advantages.

00:08:48And it's well-known, well-supported.

00:08:54It's more efficient from a packet efficiency standpoint.

00:08:59In other words, the Fiber Channel headers, if you're running native Fiber Channel, very, very thin, very lightweight protocol.

00:09:10I mean, essentially, there's no Layer 3 protocol.

00:09:13So Fiber Channel is pretty much a Layer 2 protocol that you're sending the data right over.

00:09:18Which makes it very efficient.

00:09:20The downside, of course, is that means everything pretty much has to be locally connected.

00:09:25We can set up vSANs and we can do some inter-vSAN routing if you want to get fancy with Fiber Channel.

00:09:31But it's pretty much a local solution.

00:09:37iSCSI technically can be routed.

00:09:41You can have mixed results based on throughput, bandwidth, and so on.

00:09:45I would certainly, certainly not recommend this.

00:09:49But just to throw it out there, I have, for a small customer of mine, I have run iSCSI through a VPN over the Internet.

00:09:59And it works.

00:10:01Now, obviously, you have very mixed results as far as throughput and so on and so forth.

00:10:07But it works.

00:10:09And that was over about 1,200 miles or so.

00:10:13So it can certainly be done.

00:10:16iSCSI is running over TCP.

00:10:17So the other advantage for iSCSI is not only is it routable, but it also means that it's no specialized hardware.

00:10:28You're running it over standard Ethernet equipment, standard routing equipment.

00:10:32There's no special purchases or special hardware or anything else.

00:10:37You can just run it right over anything at all.

00:10:40So if it supports TCPIP networking, you can run iSCSI networking across it.

00:10:48So to do that, you need a VM kernel port or VMK port.

00:10:54A VM kernel port is essentially a host's layer 3 presence on a network.

00:11:03You can look at it almost like an SVI on a Cisco switch.

00:11:07Same sort of idea.

00:11:08You're supposed to, and this is best practice here, one VM kernel port per each physical adapter that you want to use for iSCSI.

00:11:23So if you're going to use three iSCSI connections through three network cards, you're supposed to create three VM kernel ports, a separate switch with one NIC and one VMK in each separate V switch.

00:11:42You could technically put them in the same V switch and separate them using network policies such as port groups.

00:11:48You can change the teaming and failover of the VMK, and we'll see this a little bit again as we get into configuration.

00:11:55But you can change the teaming and failover of the VMK to use only a specific NIC if that's what you want to do.

00:12:04Okay, I'll show you how to do that here in just a minute.

00:12:06All right.

00:12:06And then this last choice, VMware doesn't really throw this out there as a solution, so I sort of put it in here.

00:12:16It's not good.

00:12:17I wouldn't do it.

00:12:19I would go with their best practice up above, but just under the category of it can be done.

00:12:24And you can also use link aggregation or what VMware calls NIC teaming.

00:12:31You could do that as well, but you can't load balance iSCSI traffic with that.

00:12:38And in a nutshell, you wouldn't be able to anyway.

00:12:41It's not supported, which is the reason I'm saying you can't do it.

00:12:45But really, it wouldn't work anyway.

00:12:49If you think it through from a networking standpoint for a minute, if you created a link aggregation group

00:12:57and connected multiple network cards to the same VMware switch and then created one VM kernel port,

00:13:08and that would be the idea.

00:13:09Otherwise, you would be doing the top section, right, which is a separate network card per switch.

00:13:15So I'm saying about not doing that down here.

00:13:18If we just had one VM kernel port in one switch doing NIC teaming,

00:13:24I mean, basically, let me do this here real quick so you can see what I'm talking about.

00:13:28So basically, what we'd be saying is you would have one VM NIC that connects

00:13:39to one V switch.

00:13:46Remember the VM NIC up here?

00:13:50And that should really be, you know what, let me change that.

00:13:57Because that's not really, I was going to write it on the side.

00:14:01I should really just put it in here.

00:14:02That's really the VM kernel interface.

00:14:07So one VM kernel interface connecting to one V switch going out through two uplinks

00:14:13that are teamed into a lag.

00:14:18Here's the thing.

00:14:19Again, if the VM kernel port is what carries the IP address,

00:14:24and this goes out to your physical switch,

00:14:30which goes out

00:14:36to your storage.

00:14:41And I don't really care about the part going on down there at the bottom.

00:14:46See, here's what I'm trying to get at is if you use a lag,

00:14:49the link aggregation group,

00:14:51you're not going to get load sharing,

00:14:55even if VMware supported it,

00:14:58because what load sharing method

00:15:01could you possibly use

00:15:03on that Ether channel or lag

00:15:06to actually load share traffic

00:15:10going from one IP address,

00:15:13which is the VM kernel address,

00:15:15to one IP address of the storage.

00:15:18That's what I'm trying to explain.

00:15:20Is,

00:15:21even if I did Ether channel,

00:15:23and even if I said, you know,

00:15:25use source destination IP hash,

00:15:29or even if I said source destination port hash,

00:15:32it's not going to matter.

00:15:34This is one IP address

00:15:35talking to one other IP address,

00:15:38one port talking to one other port.

00:15:41So there's going to be no differentiation

00:15:43with the iSCSI traffic.

00:15:45It's just the host accessing storage.

00:15:48So could this handle failover

00:15:51if this link went down?

00:15:54Sure.

00:15:56It could absolutely do that.

00:15:59Okay?

00:16:00It could absolutely handle that.

00:16:02Okay?

00:16:03Um,

00:16:04however,

00:16:05like I said,

00:16:06because of the way

00:16:07that lag does load sharing,

00:16:09it's not going to make

00:16:11any differentiation.

00:16:13All of the different choices

00:16:14we have for load sharing,

00:16:16none of them would send

00:16:17some traffic over one link

00:16:19and some over the other.

00:16:21So I have a question here.

00:16:22Is it like static pinning?

00:16:24Um,

00:16:24to an extent,

00:16:25it is.

00:16:26Um,

00:16:27I mean,

00:16:27it's not,

00:16:27of course,

00:16:27exactly the same thing,

00:16:29but conceptually,

00:16:31it sort of is

00:16:32because,

00:16:33you know,

00:16:33we're putting four links in here,

00:16:35however,

00:16:35and it's only going to use one of them

00:16:37because all the traffic

00:16:39is going from the same source IP address

00:16:42to the same destination IP address.

00:16:43Now,

00:16:44what might happen,

00:16:46depending on the physical switch,

00:16:48uh,

00:16:48what could potentially happen

00:16:50is maybe traffic going to storage

00:16:52could maybe go over one link

00:16:55and traffic coming from storage,

00:16:58so in other words,

00:16:58reads,

00:16:59could use the other link,

00:17:01potentially.

00:17:02Um,

00:17:03but again,

00:17:04it's really not the recommended way

00:17:06to do this.

00:17:07Um,

00:17:08the recommended way

00:17:09would be to use,

00:17:11uh,

00:17:11separate VLANs,

00:17:12separate IP addresses

00:17:14with two separate VM kernel ports

00:17:15and let it load share,

00:17:17uh,

00:17:18that way.

00:17:19Okay?

00:17:20And again,

00:17:21we'll get back into a little bit more

00:17:22multi-pathing here

00:17:23in a little bit too.

00:17:24Okay?

00:17:25Now,

00:17:25another thing you can do,

00:17:27and this is,

00:17:29um,

00:17:29the next section,

00:17:30so let me clear this out

00:17:31so it's not a mess for next time.

00:17:35And this has to do

00:17:37with port binding.

00:17:39And again,

00:17:39we'll see that,

00:17:40we'll see all of this,

00:17:41uh,

00:17:42in the config.

00:17:44Okay,

00:17:44we have another question here

00:17:45real quickly.

00:17:46Um,

00:17:47is this V switch

00:17:49a separate switch

00:17:50from the V switch

00:17:52for the VM NIC?

00:17:54Okay,

00:17:54so let's make sure we're,

00:17:56uh,

00:17:57I'm not quite sure

00:17:59if you mean the VM NIC

00:18:01or the V NIC.

00:18:02The VM NICs

00:18:04are your uplinks.

00:18:06So those switches

00:18:07are going to go through

00:18:08VM NICs.

00:18:09I'm guessing you probably

00:18:11mean the V NIC,

00:18:12which is what faces

00:18:12the virtual machine.

00:18:14Yeah,

00:18:14okay.

00:18:15Um,

00:18:16so yeah,

00:18:17the V NIC,

00:18:18generally speaking,

00:18:20uh,

00:18:21I would say yes.

00:18:22Okay.

00:18:24Um,

00:18:25does it have to be?

00:18:27No.

00:18:28when you're talking

00:18:30about,

00:18:31and we'll get into this

00:18:32more,

00:18:33uh,

00:18:33if you guys are going

00:18:34to be attending

00:18:35the networking,

00:18:35uh,

00:18:36section as well,

00:18:38but generally speaking,

00:18:41what it comes down to

00:18:42is this,

00:18:42what traffic do you want

00:18:44going over which

00:18:46uplinks?

00:18:47Okay.

00:18:48Whether you should create

00:18:50separate V switches

00:18:51or different port groups

00:18:53on the same V switches,

00:18:55really honestly comes down

00:18:58to which uplinks

00:19:00do you want to carry

00:19:01the traffic?

00:19:02So it sort of comes down

00:19:04to network design.

00:19:06Technically,

00:19:07there would be absolutely

00:19:08nothing stopping me

00:19:09from having one gigabit link

00:19:14going to a host,

00:19:16trunking that link,

00:19:19and sending all of my

00:19:21management traffic,

00:19:24NFS traffic,

00:19:25iSCSI traffic,

00:19:27vMotion traffic,

00:19:29fault tolerant traffic,

00:19:31and all the traffic

00:19:32for all my virtual machines

00:19:34over that same link.

00:19:37Nothing to stop that.

00:19:39I'm guessing you would

00:19:41probably have quite a bit

00:19:42of a throughput problem,

00:19:44right?

00:19:45So,

00:19:45again,

00:19:46it comes to scalability,

00:19:47it comes to design.

00:19:50The,

00:19:50the real,

00:19:51the real factor

00:19:53is going to come into

00:19:54a physical nick

00:19:57in a host,

00:19:58which is your uplink,

00:19:59that's your VM mix,

00:20:01can only belong

00:20:03to one V switch.

00:20:04the question is,

00:20:08is the V switch,

00:20:09is this V switch separate?

00:20:11And unfortunately,

00:20:11my answer is,

00:20:13it depends on design,

00:20:15but if I could make a

00:20:17really strong recommendation

00:20:18to make it separate,

00:20:20then,

00:20:21yes,

00:20:21I would make it separate.

00:20:23Okay?

00:20:23But it could be either way.

00:20:26In fact,

00:20:27with the design we have today,

00:20:29I can show it to you both ways.

00:20:31That it could be separate

00:20:32or it could be the same.

00:20:34We'll do that,

00:20:35just for fun.

00:20:37Okay?

00:20:38We'll show it,

00:20:38we'll show it both ways.

00:20:41But,

00:20:41but I would strongly recommend

00:20:42that you keep it separate.

00:20:45It really comes down to

00:20:46how many network cards

00:20:47are in the host.

00:20:49That's what it comes down to.

00:20:50And again,

00:20:50it's network design.

00:20:52If you have plenty

00:20:53of network cards,

00:20:56then separate everything.

00:20:58If you don't,

00:20:59then you've got to start

00:21:00teaming things up.

00:21:02You know,

00:21:03maybe I put,

00:21:04maybe I have to put

00:21:05management traffic

00:21:07together with

00:21:08V motion traffic.

00:21:10Okay?

00:21:10But,

00:21:12when it really comes down to it,

00:21:15if you're using iSCSI,

00:21:17which,

00:21:18since we're in this section,

00:21:19we'll assume we are,

00:21:20since we're having this

00:21:20discussion,

00:21:22then,

00:21:23yes,

00:21:23I would really,

00:21:24really,

00:21:24really strongly recommend

00:21:25that you keep this separate.

00:21:28You know,

00:21:28if you had to put

00:21:28everything else together

00:21:30and separate just one thing,

00:21:32then it would be

00:21:34iSCSI networking.

00:21:36Pull this out separate.

00:21:37Because this is going to

00:21:38create a lot of traffic.

00:21:39This is the traffic

00:21:39to your storage.

00:21:41And,

00:21:41ultimately,

00:21:42you know,

00:21:42this is your virtual machines

00:21:44talking to the hard drives.

00:21:46You know,

00:21:49so,

00:21:49you're talking about

00:21:51how fast is the throughput

00:21:52between

00:21:53a

00:21:54operating system

00:21:56and its hard drive.

00:21:58So,

00:21:58the faster that can be,

00:22:00absolutely the better.

00:22:02If you have a,

00:22:03question here,

00:22:04if you have a 10 gig NIC,

00:22:07can you put all

00:22:08iSCSI network traffic

00:22:09on the same

00:22:10VM NIC switch?

00:22:12Or do you recommend

00:22:13buying a separate NIC

00:22:14just dedicated to iSCSI?

00:22:16At 10 gigs,

00:22:17you should be fine.

00:22:19At 10 gigs,

00:22:21honestly,

00:22:21you're really no different

00:22:22than doing

00:22:23fiber channel over Ethernet

00:22:24or FCOE

00:22:25where you're sending

00:22:27your data traffic

00:22:28and your storage traffic

00:22:30across the exact same circuit.

00:22:34Now,

00:22:35obviously,

00:22:36there might be

00:22:36a little bit of

00:22:37redundancy concern

00:22:39coming into that

00:22:40if you only have

00:22:40one 10 gig NIC.

00:22:43You know,

00:22:44if it fails,

00:22:44you're sort of

00:22:45out of luck.

00:22:46But,

00:22:47just from a

00:22:48throughput standpoint,

00:22:50I would think

00:22:51that would probably

00:22:51be fine

00:22:52under most circumstances.

00:22:53Again,

00:22:55that's essentially,

00:22:58well,

00:22:58it's not essentially,

00:22:59it's exactly

00:22:59what FCOE does.

00:23:01It's the storage

00:23:02and the network traffic

00:23:04or data traffic

00:23:05over the same NIC.

00:23:07So,

00:23:08I think you'd be

00:23:08absolutely fine with that.

00:23:11Okay.

00:23:13So,

00:23:13port binding here.

00:23:15And,

00:23:16we'll see this choice

00:23:17as we go into

00:23:19setting up

00:23:21the iSCSI networking.

00:23:23which is

00:23:25for the port binding.

00:23:29You can read

00:23:30what it says here.

00:23:31If you're doing

00:23:32this exact same

00:23:34config that I had

00:23:37just drawn up there.

00:23:39Well,

00:23:40actually,

00:23:40let's rephrase

00:23:41a little bit.

00:23:42You have

00:23:42two

00:23:43VM kernel mix

00:23:46that you're going

00:23:47to put on

00:23:48the same subnet

00:23:49and your storage

00:23:51is on the same

00:23:52subnet

00:23:53with a single

00:23:54IP address.

00:23:56You can use

00:23:57port bindings.

00:24:00Okay.

00:24:00So,

00:24:01this is assuming

00:24:02that there's no routing.

00:24:03This is assuming

00:24:04that you're not using

00:24:06LACP

00:24:07on your

00:24:07uplinks.

00:24:09This is

00:24:09single ports.

00:24:11Okay.

00:24:14And,

00:24:15that they're on

00:24:16the same

00:24:17V switch.

00:24:19Then,

00:24:19you can use

00:24:20port bindings

00:24:21to tie it in.

00:24:22It's just a way

00:24:23of doing load sharing

00:24:24on the same

00:24:25subnet.

00:24:27Okay.

00:24:27Without going in

00:24:28and doing

00:24:30the stuff

00:24:30we just looked at.

00:24:32As you can see

00:24:33down here at the bottom,

00:24:34when you should not

00:24:34use port bindings,

00:24:36again,

00:24:37port binding

00:24:38does not support

00:24:38routing.

00:24:40So,

00:24:40if you're routing

00:24:41to your iSCSI

00:24:42array,

00:24:43then you can't

00:24:44use port bindings

00:24:45if they're on

00:24:46different subnets

00:24:47or different

00:24:49V switches.

00:24:50Under any of those

00:24:51circumstances,

00:24:52you do not do

00:24:53port bindings.

00:24:55So,

00:24:55we'll see if,

00:24:56I didn't actually

00:24:57test doing port

00:24:58bindings in our

00:24:58config here today.

00:24:59We'll give it a shot

00:25:00when we get there.

00:25:01See if we can get

00:25:02it to work.

00:25:03Okay.

00:25:04Shouldn't be a problem.

00:25:05But,

00:25:06let's take a look

00:25:06at it when we get

00:25:07there.

00:25:08Okay.

00:25:09Some other things

00:25:10you probably want to do

00:25:11to help increase

00:25:11performance.

00:25:13We already mentioned

00:25:13this.

00:25:15My formatting

00:25:16got off a little bit

00:25:16on that.

00:25:17Sorry about that.

00:25:22Alright.

00:25:22No sense fixing it

00:25:23right this second.

00:25:24We already mentioned

00:25:26this though

00:25:27is the jumbo frames.

00:25:28Okay.

00:25:29A really great

00:25:30way to get a little

00:25:31bit of increased

00:25:32performance here.

00:25:33Okay.

00:25:33We already said

00:25:34most of this.

00:25:35Right.

00:25:36We support up to

00:25:379,000 bytes.

00:25:39Right.

00:25:40Jumbo frames

00:25:40have to be supported

00:25:41end to end.

00:25:42So,

00:25:42we pretty much

00:25:43already said all this

00:25:44and I am going to

00:25:45show you

00:25:45as we set this up

00:25:47how we configure it.

00:25:49So,

00:25:50not a real big deal

00:25:51there.

00:25:51Right.

00:25:53We're going to do

00:25:54dynamic

00:25:54just because it's

00:25:55real easy.

00:25:57Array

00:25:57sends a list

00:25:59of all the targets.

00:26:00So,

00:26:00the initiator,

00:26:01our host,

00:26:03does a

00:26:03send targets request.

00:26:06The array comes back

00:26:08with a list of the offers

00:26:09and of course,

00:26:10I mentioned this

00:26:12earlier as well,

00:26:13that it adds that

00:26:14to the static discovery.

00:26:15We'll see that

00:26:15in a minute as well.

00:26:17And,

00:26:17once you put it in,

00:26:18you can't change it.

00:26:20Just sort of one of those

00:26:21interesting little caveats

00:26:22that

00:26:25if you do need to change

00:26:27the port number

00:26:28or IP address,

00:26:29you just have to remove

00:26:30the dynamic discovery

00:26:32and just put it back in again.

00:26:34So,

00:26:35I mean,

00:26:35the IP address

00:26:36and the port number,

00:26:37it's pretty much

00:26:38what defines

00:26:38the target.

00:26:40So,

00:26:41you know,

00:26:42just take it out

00:26:42and put it back in.

00:26:44Okay.

00:26:44The other choice,

00:26:45of course,

00:26:45is static,

00:26:46which,

00:26:47you know,

00:26:47I just couldn't resist

00:26:47putting it in here.

00:26:49VMware lists it as

00:26:50the two types of discovery,

00:26:52one of them being static.

00:26:54So,

00:26:55you

00:26:55statically discovered

00:26:57nothing.

00:26:59You're

00:27:00typing it in manually.

00:27:02So,

00:27:02okay,

00:27:04so we have another question here.

00:27:05Let's see.

00:27:06If you have two

00:27:07iSCSI

00:27:08Nicks

00:27:09for redundancy

00:27:09and multipathing,

00:27:11should each

00:27:12iSCSI

00:27:13Nick live

00:27:14in a separate subnet

00:27:15or they can both

00:27:18live on the same subnet,

00:27:22the storage array

00:27:23is on another

00:27:24routed subnet?

00:27:25Okay.

00:27:27If the storage array,

00:27:29let me actually go back

00:27:30here one second

00:27:31or try to.

00:27:34Here we go.

00:27:34So I just want to show you

00:27:36the restriction here.

00:27:39Okay.

00:27:43Which is simply

00:27:44this.

00:27:48Whoops.

00:27:49No,

00:27:50go back.

00:27:50Sorry about that.

00:27:52Right there's your problem.

00:27:55Okay.

00:27:55So basically

00:27:59there's two setups.

00:28:01Let's just do it this way.

00:28:01There's two setups

00:28:02that are supported.

00:28:04Okay.

00:28:04You can create

00:28:06two

00:28:07VM kernel ports

00:28:08connected to the same

00:28:11switch,

00:28:13give them two

00:28:14different IP addresses

00:28:16on the same subnet,

00:28:19and then

00:28:20use port binding.

00:28:22Or

00:28:25you can create

00:28:27two VM kernel ports

00:28:29on two separate subnets

00:28:32hooked to two separate

00:28:35V switches.

00:28:41The second one

00:28:42is what you have to do

00:28:44if you're then going

00:28:45to route

00:28:45to your storage.

00:28:48Because

00:28:48the first option,

00:28:50okay,

00:28:52again,

00:28:52VM kernel ports

00:28:53on the same subnet,

00:28:55on the same switch,

00:28:57is

00:28:58using port binding,

00:29:00and port binding

00:29:01doesn't support

00:29:02the

00:29:04array

00:29:06being on a different subnet.

00:29:08Look at the first bullet point

00:29:09there under the do not.

00:29:11Right?

00:29:12Do not use it

00:29:13when it's on a different subnet.

00:29:16So

00:29:16if you're routing,

00:29:18then you want to go

00:29:19with the option

00:29:20of

00:29:20two different

00:29:22NICs

00:29:22on two different switches

00:29:23on different subnets,

00:29:25and then route

00:29:27to the iSCSI storage.

00:29:29And then you just

00:29:30bind both adapters

00:29:31to your

00:29:32iSCSI adapter.

00:29:35Again,

00:29:36we're not going to be doing

00:29:37any today.

00:29:39Okay?

00:29:39But

00:29:40if you look at the different choices

00:29:42for what supports

00:29:43CHAP

00:29:43and what doesn't,

00:29:44iSCSI

00:29:47supports,

00:29:48of course,

00:29:49all three types of adapters

00:29:51cannot do authentication,

00:29:53which is the first choice there.

00:29:54Right?

00:29:57Unidirectional

00:29:57can be done

00:29:58on the software

00:29:59and the dependent.

00:30:00Remember,

00:30:01the dependent hardware type

00:30:02gets its config

00:30:04from ESX.

00:30:05Okay?

00:30:06Unidirectional

00:30:06can be done

00:30:08on all of them.

00:30:09So there's a choice

00:30:10for unidirectional

00:30:11unless prohibited.

00:30:12So in other words,

00:30:12that's the

00:30:13host saying

00:30:15I want to do CHAP,

00:30:17I will try to do CHAP,

00:30:19but if the array says no,

00:30:22oh, well,

00:30:22that's fine,

00:30:23I'll connect anyway.

00:30:25Okay?

00:30:25So just so we're clear,

00:30:27you know,

00:30:27the first choice is

00:30:29we're not doing CHAP.

00:30:30The second choice,

00:30:33unidirectional

00:30:34if required,

00:30:35basically,

00:30:36that's the host's

00:30:37attitude of

00:30:38well,

00:30:39I don't really

00:30:40want to do CHAP,

00:30:41but,

00:30:42you know,

00:30:42if they insist on it,

00:30:44then I will.

00:30:45If the array says

00:30:46I must do CHAP,

00:30:47then I'll do CHAP.

00:30:48Otherwise,

00:30:49no.

00:30:51Unidirectional

00:30:52unless prohibited

00:30:53says I want to do CHAP,

00:30:55I will try to do CHAP,

00:30:57but if the array

00:30:58refuses CHAP,

00:31:00then I won't.

00:31:02But I'll still connect.

00:31:03That's the thing.

00:31:04I'll still connect,

00:31:05I just won't use CHAP.

00:31:07Unidirectional,

00:31:09again,

00:31:09supported on all

00:31:10of the adapter types

00:31:13is required.

00:31:16In other words,

00:31:17he's saying,

00:31:18you know,

00:31:18I must do authentication.

00:31:20If the array

00:31:21doesn't support authentication,

00:31:23then I guess

00:31:24I just don't connect.

00:31:25And that's that.

00:31:27Bidirectional

00:31:28is supported

00:31:30in both the software

00:31:31independent

00:31:32ISCSI types.

00:31:36However,

00:31:37like I mentioned earlier,

00:31:39you just got to make sure

00:31:40that the array supports it.

00:31:42Because not all

00:31:43of the arrays

00:31:44will actually accept

00:31:46the client

00:31:47or the initiator

00:31:49requesting authentication

00:31:51from them.

00:31:53They're used to

00:31:53authenticating

00:31:54the initiator,

00:31:55but not necessarily

00:31:56the other way around.

00:31:58You have a couple choices here.

00:32:00Basically,

00:32:01you can either do it

00:32:02from an independent

00:32:04hardware.

00:32:06Now,

00:32:06like we said before,

00:32:07an independent hardware

00:32:08is basically

00:32:09a network card

00:32:11with the full

00:32:12ISCSI implementation

00:32:13in the firmware.

00:32:15So you're going

00:32:16to go into its BIOS

00:32:18and you're going

00:32:18to configure it

00:32:19just like you would

00:32:20configure your

00:32:21fiber channel HBA.

00:32:23You're going to say,

00:32:24go connect to this

00:32:25ISCSI storage array,

00:32:27this LUN,

00:32:28this is what you boot from.

00:32:31And again,

00:32:31for the independent,

00:32:34this really,

00:32:35again,

00:32:35has nothing to do

00:32:36with VMware.

00:32:38Okay?

00:32:38It's all on the network card

00:32:39and the storage array.

00:32:41The only thing

00:32:42that's going to come

00:32:43into it is,

00:32:45again,

00:32:47just like with

00:32:48the fiber channel,

00:32:49when you install ESX,

00:32:51just simply installing

00:32:52it onto the correct LUN.

00:32:54That's really it.

00:32:56Okay?

00:32:57Now,

00:32:57for the IBFT,

00:32:59that's where

00:33:01it's done

00:33:01separately.

00:33:03Okay?

00:33:04So basically,

00:33:06just to wrap

00:33:07this up here,

00:33:08the IBFT

00:33:09is configured

00:33:11on the HBA

00:33:12and then,

00:33:13of course,

00:33:13all these parameters

00:33:14are passed

00:33:16from the VM kernel.

00:33:17So,

00:33:18sort of like we said

00:33:19with the

00:33:21FCOE,

00:33:24the software FCOE,

00:33:26the card

00:33:26has to support

00:33:27IBFT.

00:33:29Okay?

00:33:29We have it written up

00:33:30on the top there for you.

00:33:31It's the ISCSI boot

00:33:32firmware table,

00:33:33very similar

00:33:34to the whole concept

00:33:36with booting

00:33:37from FCOE.

00:33:39As long as

00:33:40the network card

00:33:40supports it,

00:33:42basically what happens

00:33:43is the config

00:33:43is saved

00:33:44in the firmware

00:33:45of the card

00:33:46while ESXi

00:33:48is up and running

00:33:49and when it reboots

00:33:51and we have to boot

00:33:53from that LUN,

00:33:55the IBFT

00:33:55is read

00:33:56on the network card,

00:33:58the network card

00:33:58connects to the LUN

00:33:59and begins

00:34:01to boot VMware.

00:34:02As VMware boots

00:34:04and the kernel loads,

00:34:05the VMware kernel

00:34:06will then of course

00:34:07load its own

00:34:09ISCSI drivers,

00:34:12its own ISCSI stack.

00:34:13It connects

00:34:14to the storage

00:34:15and then just continues

00:34:16to boot

00:34:16going through

00:34:18the VMware kernel

00:34:19at that point.

00:34:21So let's take a look here.

00:34:23That's pretty much

00:34:24all of the intro

00:34:25and definition stuff

00:34:28here for ISCSI.

00:34:30So let's go in,

00:34:30let's jump in

00:34:31and take a look

00:34:31at getting this set up

00:34:33and actually getting it running.

00:34:34So let's switch

00:34:36back over here

00:34:36to our client

00:34:38and for this one

00:34:40I'm not going to use

00:34:41at least,

00:34:42well at least

00:34:43not to begin with,

00:34:44I'm not going to use

00:34:44my first host here,

00:34:46I'll use my other two.

00:34:48Now as you can see

00:34:49on these other two hosts,

00:34:50I've got some warnings

00:34:51on them.

00:34:52If I click on this

00:34:53and I go to summary

00:34:55to see the warnings,

00:34:59okay,

00:35:00first warning is

00:35:01no data stores

00:35:02have been configured.

00:35:03Now,

00:35:03like I mentioned before,

00:35:05it's up and running,

00:35:06right?

00:35:07I mean,

00:35:07the host is powered on,

00:35:08it's booted up,

00:35:09it's running ESX,

00:35:11so obviously it has storage.

00:35:14But again,

00:35:15it's just a four gig

00:35:16locally connected drive

00:35:19that could be a USB thumb drive,

00:35:21you know,

00:35:22really it could be anything.

00:35:24It's just that

00:35:25it's small enough

00:35:26that ESX looked at that

00:35:27on the install

00:35:28and said,

00:35:29yeah,

00:35:30I don't have room here

00:35:31for a data store,

00:35:34so I'll install to it

00:35:36and it'll partition it

00:35:37and ESXI can run from it,

00:35:40but it does not have

00:35:41any persistent data stores

00:35:43at the moment.

00:35:45If you look right down here

00:35:46under storage,

00:35:48zero data stores.

00:35:50So he has no place

00:35:51to put anything

00:35:52at the moment.

00:35:54So I can't create

00:35:55a virtual machine

00:35:55or anything on this host

00:35:57because he has

00:35:58zero storage.

00:36:00So what we're going to do

00:36:01is we're going to connect him

00:36:02to our

00:36:03iSCSI storage array.

00:36:06And we'll start

00:36:07with doing

00:36:07number two here.

00:36:09Now,

00:36:09as I mentioned before,

00:36:11my general attack

00:36:12to this

00:36:13is to go

00:36:13for the networking first.

00:36:15All right?

00:36:16So like I said,

00:36:18you don't have to,

00:36:19but I usually do.

00:36:21So to do this,

00:36:22we go to manage

00:36:22and we're going to go

00:36:24to networking

00:36:25to start with

00:36:26and we'll go

00:36:29to virtual switches.

00:36:31Okay?

00:36:31Now,

00:36:32for those of you

00:36:33who may not be

00:36:34familiar

00:36:34or that familiar

00:36:36at all

00:36:36with VMware,

00:36:39the switches

00:36:40are used

00:36:40because,

00:36:41you know,

00:36:41we have a finite

00:36:42number

00:36:43or a limited number

00:36:44of physical uplinks

00:36:46out of any given host.

00:36:48And we're going to have

00:36:50all sorts of virtual machines

00:36:51and all sorts of stuff

00:36:52that need to go out of here.

00:36:53So we're going

00:36:55to have switches.

00:36:57Okay?

00:36:57Now,

00:36:58there can be

00:36:58a combination

00:36:59of distributed switches

00:37:01or there can be,

00:37:04sorry,

00:37:07distributed virtual switches,

00:37:09regular switches,

00:37:10third-party

00:37:10distributed virtual switches.

00:37:13And these are all things

00:37:14that we'll cover

00:37:15in the networking course

00:37:17for VMware.

00:37:19We're going to start off

00:37:21with just using

00:37:22a standard V switch here.

00:37:24And what I have

00:37:25hooked to ESX2 right now,

00:37:29and actually,

00:37:30in looking at this,

00:37:31I don't actually want

00:37:32this in here right now.

00:37:33So I'm going to get rid

00:37:34of this for a minute.

00:37:38We'll do more of this

00:37:39in the networking side.

00:37:43And it's being used.

00:37:44That's fun.

00:37:46Okay,

00:37:46well,

00:37:46on this one,

00:37:47we'll just connect

00:37:47the normal way then.

00:37:49That's fine.

00:37:50I'll worry about

00:37:50cleaning that up later.

00:37:52I don't want to go

00:37:52hunt down what's using

00:37:53it right now.

00:37:54So we'll just stick

00:37:55with this standard switch.

00:37:57This standard switch,

00:37:58as of right now,

00:37:58has two uplinks,

00:38:00VMNIC0

00:38:00and VMNIC1,

00:38:03okay,

00:38:03which are connected out

00:38:04to the rest

00:38:06of the network.

00:38:07I already have,

00:38:08now,

00:38:09there's always going

00:38:09to be

00:38:10vSwitch0

00:38:11is going to be

00:38:12created by default.

00:38:14It's always going

00:38:15to be there.

00:38:15I already have deployed

00:38:18the Cisco

00:38:20nexus1000v,

00:38:22which is actually,

00:38:22when I look at that,

00:38:23that's what's using that.

00:38:26So,

00:38:27it has more ports

00:38:28than I wanted it to

00:38:29at the moment.

00:38:30But,

00:38:31we can stick with

00:38:32the vSwitch0

00:38:34for now.

00:38:35It's fine.

00:38:36And,

00:38:36because I'm going to,

00:38:37I have more on this other guy

00:38:38when we get to him.

00:38:39So,

00:38:40all we need to do

00:38:41is give him a way

00:38:42to communicate

00:38:43out to

00:38:45the storage network.

00:38:47Now,

00:38:48again,

00:38:48there's a couple ways

00:38:49I can do this

00:38:49as we mentioned

00:38:50in some of our discussions

00:38:51a minute ago.

00:38:53Easiest way to do this,

00:38:55if these two uplinks,

00:38:57if these two physical adapters

00:38:58over here

00:38:59are trunks

00:39:01and can carry

00:39:02all of the traffic,

00:39:04then,

00:39:05bare minimum,

00:39:06all I have to do

00:39:07is create a VM kernel port

00:39:09to allow him

00:39:11to communicate

00:39:11to storage.

00:39:13So,

00:39:14you know,

00:39:14to get back

00:39:14to some of the questions

00:39:15we had earlier,

00:39:17you know,

00:39:17does it have to be separate?

00:39:18Can it be together?

00:39:18All those kinds of questions.

00:39:21Just to start off with here,

00:39:23we can very simply

00:39:24go to this switch,

00:39:27edit the existing switch.

00:39:28Now,

00:39:28I didn't mean to do that.

00:39:29Sorry.

00:39:30Highlight the switch itself.

00:39:32So,

00:39:32on the vSwitch up here,

00:39:34we hit edit.

00:39:35By the way,

00:39:36you want to turn on

00:39:37jumbo frames?

00:39:39Right there

00:39:39is one of the places

00:39:40that we would have to do it.

00:39:42We would have to crank this

00:39:43up to 9,000.

00:39:45These other things

00:39:45we'll cover

00:39:46in the networking course.

00:39:48Okay?

00:39:48But that's where

00:39:49you would turn on jumbo frames.

00:39:52And what we really

00:39:53want to do here

00:39:53is simply add

00:39:55host networking.

00:39:57Okay?

00:39:57We need a VM kernel port

00:39:59to talk to

00:40:00our iSCSI storage.

00:40:02So,

00:40:03we'll say

00:40:04it's a VM kernel adapter.

00:40:05Hit next.

00:40:07Okay?

00:40:07And here it's going to say

00:40:08should I use

00:40:09an existing port group,

00:40:11an existing standard switch,

00:40:12create a new switch.

00:40:14So,

00:40:15if I had uplinks,

00:40:16and we'll do this

00:40:17on the other host,

00:40:18but if I had uplinks

00:40:19that I wanted to create

00:40:20a whole new switch with,

00:40:22go back to some

00:40:22of those questions,

00:40:23I could do it

00:40:24right from here.

00:40:25I don't even have to

00:40:25create it ahead of time.

00:40:26I could do it

00:40:27right out of this config,

00:40:28but we're going to use

00:40:29this same switch for now.

00:40:31We're going to hit next.

00:40:34Give it a label.

00:40:36And we'll call this

00:40:37IP storage.

00:40:44What VLAN

00:40:44are we using

00:40:46for our storage?

00:40:47Now,

00:40:47you guys won't know this,

00:40:48but I'm using a thousand.

00:40:51Okay?

00:40:52For storage on my network.

00:40:55We're going to use

00:40:55IP version 4,

00:40:57default stack,

00:40:58and I don't need

00:40:59to use any of these

00:41:00things with it.

00:41:01It's not for

00:41:02vMotion traffic.

00:41:03It's not for

00:41:03fault tolerance.

00:41:04It's not for management,

00:41:06and it's not for

00:41:06virtual SAN.

00:41:08So,

00:41:08I'm not going to check

00:41:09any of these.

00:41:10It's just standard

00:41:11IP traffic.

00:41:13So,

00:41:13I hit next.

00:41:17Give it an IP address,

00:41:19which,

00:41:19of course,

00:41:20would be on that

00:41:21same VLAN

00:41:22and or being routed

00:41:24through another

00:41:25VLAN,

00:41:26if that was your choice.

00:41:28If that's how you

00:41:29were setting up

00:41:29your network,

00:41:30in our case,

00:41:31it's not.

00:41:38So,

00:41:38that's just an IP address

00:41:40on the same network,

00:41:43because I just don't have

00:41:44routing set up right now

00:41:45for the storage network.

00:41:46So,

00:41:47they're just connected.

00:41:53And it'll simply create

00:41:55a new VM kernel port here.

00:41:58So,

00:41:58here we go,

00:41:59IP storage

00:42:00with that VLAN ID

00:42:02and that IP address.

00:42:05And that's the same VLAN

00:42:07that my storage array is on.

00:42:09Okay?

00:42:11So,

00:42:12from the host perspective,

00:42:13that's all it needs.

00:42:14That's the VM kernel port

00:42:15and

00:42:17the networking.

00:42:20That's all it needs.

00:42:22Now,

00:42:22again,

00:42:22we can get,

00:42:23we'll get fancier with it

00:42:24on the other host,

00:42:25but I'm just trying to start off

00:42:26just real simple here.

00:42:28Okay?

00:42:29So,

00:42:30that's really the bare minimums.

00:42:32Now,

00:42:32again,

00:42:33if we wanted

00:42:34to do jumbo frames,

00:42:36just to finish it up here,

00:42:37you would highlight this.

00:42:40Hello.

00:42:41Here we go.

00:42:42Highlight,

00:42:43edit,

00:42:44and turn up the MTU

00:42:51right here

00:42:52on the network card

00:42:53as well

00:42:53on the VM kernel interface.

00:42:55So,

00:42:56make sure you turn on

00:42:57jumbo frames

00:42:59both places.

00:43:01Okay?

00:43:02So,

00:43:03under the VM kernel interface

00:43:05right here

00:43:05and

00:43:07under

00:43:08the properties

00:43:09of the switch

00:43:10that this traffic

00:43:11is going to be going out through.

00:43:13Be that,

00:43:14like I said,

00:43:14in this case,

00:43:16and this would not,

00:43:16I would not recommend this

00:43:18as best practice.

00:43:19I'm sending my management traffic

00:43:21and who knows what else

00:43:22out through this same

00:43:23NIC team

00:43:25Okay?

00:43:27Now,

00:43:27in this case,

00:43:28if we take a look at this,

00:43:29let me just show you real quick.

00:43:30If we look at the properties

00:43:31of the standard switch

00:43:32and I look at

00:43:35Oh,

00:43:40I'm on the same,

00:43:41hit the wrong thing again.

00:43:42I'm sorry.

00:43:42On the switch up here,

00:43:44we go to the properties

00:43:45and we go to teaming

00:43:48and failover.

00:43:49Okay?

00:43:50I have both of these network cards

00:43:52set to active.

00:43:53And again,

00:43:55we're getting a little bit

00:43:56into the networking side

00:43:57of it here,

00:43:57but both of these

00:43:59are set to active

00:44:00and they're set to route

00:44:01based on originating

00:44:03virtual port.

00:44:04So,

00:44:05this is essentially

00:44:06doing NIC teaming

00:44:07as it is,

00:44:09but

00:44:09it's routing based

00:44:12on originating

00:44:13virtual port.

00:44:14Well,

00:44:14for all of this traffic,

00:44:16the originating virtual port

00:44:18is going to be

00:44:19VMK1 down here.

00:44:20So,

00:44:21this is what I was

00:44:22talking about earlier.

00:44:23You're not going to get

00:44:24load sharing

00:44:25in this case

00:44:26because,

00:44:29yeah,

00:44:29it's technically

00:44:30load sharing,

00:44:31but everything's going

00:44:32to be coming

00:44:32from that same

00:44:33source port.

00:44:35And even if I change

00:44:36this to route

00:44:36based on IP hash,

00:44:38again,

00:44:39what I was saying

00:44:39earlier

00:44:39is,

00:44:42you know,

00:44:42this is not going

00:44:43to cut it.

00:44:45Okay?

00:44:46Because it's all

00:44:46coming from the same

00:44:47IP address

00:44:48going to the same

00:44:49destination IP address

00:44:50which is going

00:44:51to be our storage.

00:44:53Okay?

00:44:53So,

00:44:54just a very simple

00:44:55single path.

00:44:56Now,

00:44:56this does,

00:44:57again,

00:44:57we're offering

00:44:58a little bit of

00:44:58failover here,

00:45:00but it's certainly

00:45:01not the best

00:45:01as far as traffic goes

00:45:03having all of this

00:45:03stuff mixed

00:45:04on one switch

00:45:05going through

00:45:05the same uplinks

00:45:06as my management

00:45:08traffic

00:45:08and everything else

00:45:09right now.

00:45:10But just to show you

00:45:10it can be done.

00:45:12Okay?

00:45:12Let's go over here

00:45:13to storage

00:45:14and we have to go

00:45:16to storage adapters

00:45:17which is our

00:45:18default option

00:45:20that came up here

00:45:20and we're going

00:45:21to hit add.

00:45:23Okay?

00:45:23Now,

00:45:24again,

00:45:24this is where you

00:45:25add either

00:45:25the software

00:45:26FCOE

00:45:27or the software

00:45:27iSCSI.

00:45:29Okay?

00:45:30And we're just

00:45:30going to say

00:45:31software iSCSI.

00:45:32Bam.

00:45:34And it's going

00:45:34to say it's

00:45:34going to add it.

00:45:36Okay?

00:45:36And we're going

00:45:36to have to go

00:45:37to properties

00:45:37to finish it.

00:45:38Well,

00:45:38we already know

00:45:39that.

00:45:39At least I hope

00:45:40we do.

00:45:40We will in a

00:45:41minute anyway.

00:45:42So,

00:45:43it's going

00:45:43to create it.

00:45:44It's going

00:45:44to put it

00:45:44on this list

00:45:45up here

00:45:45at the top.

00:45:54Okay?

00:45:54So,

00:45:55here it is.

00:45:55So,

00:45:58VHBA33

00:45:59online.

00:46:01And there,

00:46:02by the way,

00:46:02you can see

00:46:02that wonderful

00:46:03name that it

00:46:03created for us.

00:46:04Right?

00:46:05So,

00:46:06our identifier

00:46:06is IQN.

00:46:07Well,

00:46:08I'm not going

00:46:08to read it.

00:46:08You can see it.

00:46:09But,

00:46:10that big,

00:46:11fun,

00:46:11long IQN name.

00:46:13Okay?

00:46:14And then,

00:46:15down here at the

00:46:15bottom,

00:46:16we go to properties.

00:46:18Once you've

00:46:19turned it on,

00:46:20it's going to go

00:46:21to enabled.

00:46:23Now,

00:46:23one thing that's

00:46:24not,

00:46:24I don't think,

00:46:25like,

00:46:25incredibly intuitive

00:46:26is,

00:46:28what if I don't

00:46:29want this anymore?

00:46:31What if you,

00:46:32and again,

00:46:34we're going to spend

00:46:34most of class

00:46:35setting things up,

00:46:36right?

00:46:36Not tearing things

00:46:37down.

00:46:37But,

00:46:38you know,

00:46:39what if you

00:46:39clicked on that

00:46:40by accident,

00:46:41and you went,

00:46:41oh,

00:46:42oh,

00:46:42no,

00:46:42wait,

00:46:42I didn't,

00:46:43I'm not doing

00:46:43that.

00:46:43I don't want

00:46:44that.

00:46:45What's not

00:46:46incredibly intuitive

00:46:47is how you get

00:46:47rid of it,

00:46:49which is,

00:46:50you have to say

00:46:50disable.

00:46:52Disable is the

00:46:53equivalent of

00:46:54delete it.

00:46:55Okay?

00:46:57Just so you know,

00:46:58though,

00:46:58and I'm not going

00:46:59to do it right now,

00:46:59maybe we'll do it

00:47:00at the end of the

00:47:00day once we're done,

00:47:02but if you disable

00:47:04it,

00:47:04and it removes it,

00:47:06it actually has to

00:47:07reload the host.

00:47:08So,

00:47:09to stick up for

00:47:11VMware here,

00:47:12because I do like

00:47:12it,

00:47:13you know,

00:47:13it's a great

00:47:13product,

00:47:14very,

00:47:15very few things

00:47:16actually require

00:47:17that you reload

00:47:18the host,

00:47:20but that is one

00:47:21of them.

00:47:22Okay?

00:47:22I mean,

00:47:23obviously,

00:47:23you know,

00:47:24upgrading the

00:47:24kernel and stuff

00:47:25like that do as

00:47:26well,

00:47:26and this is why

00:47:27we have things

00:47:28like vMotion,

00:47:28right?

00:47:29That way we can

00:47:29vMotion all of

00:47:30our machines,

00:47:31all of the VMs

00:47:32off of the host,

00:47:33and then reboot

00:47:36it,

00:47:36but you can add

00:47:37it on the fly

00:47:38like we just did,

00:47:39no problem,

00:47:39but to remove

00:47:41it,

00:47:41you got to

00:47:42reload the host.

00:47:43Okay,

00:47:44so once it's

00:47:44enabled,

00:47:45the big thing

00:47:46we're going to

00:47:46have to do

00:47:46is come down

00:47:47here,

00:47:47and under

00:47:48general is

00:47:49we say

00:47:50edit.

00:47:52Sorry,

00:47:52not for that

00:47:53part,

00:47:53but if you

00:47:55wanted to

00:47:55change that

00:47:56name,

00:47:56you could,

00:47:57but I

00:47:57apologize,

00:47:58that's not

00:47:58where we

00:47:58actually need

00:47:59to go,

00:47:59I have to

00:47:59change that

00:48:00part,

00:48:00I want to

00:48:00go to

00:48:00devices.

00:48:02This stuff's

00:48:03going to be

00:48:03empty,

00:48:04because we

00:48:04don't have

00:48:05anything running

00:48:06yet at all.

00:48:07If we

00:48:09wanted to

00:48:10do network

00:48:10port binding,

00:48:11we'd come

00:48:11over here.

00:48:12I'm going

00:48:13to try

00:48:13and do

00:48:13this next.

00:48:14If we

00:48:14hit plus,

00:48:15it should

00:48:15give us,

00:48:18see it's

00:48:19not even

00:48:19going to

00:48:19give us

00:48:19a choice

00:48:19to set

00:48:20it up

00:48:20right now,

00:48:22because it

00:48:23tells you

00:48:24only VM

00:48:25adapters

00:48:26compatible with

00:48:27iSCSI port

00:48:28binding

00:48:28requirements are

00:48:29available,

00:48:30and we

00:48:31have other

00:48:32things attached

00:48:33to our

00:48:33switch other

00:48:34than just

00:48:34iSCSI,

00:48:35so it's not

00:48:35going to show

00:48:36up here,

00:48:37but we'll

00:48:37come back

00:48:38and mess

00:48:38with this

00:48:39a little

00:48:39after we

00:48:40get it

00:48:40working.

00:48:42The big

00:48:42thing we

00:48:42want to

00:48:43do is

00:48:43go to

00:48:43targets,

00:48:45and again,

00:48:47this is where

00:48:47you can do

00:48:47either dynamic

00:48:48or static.

00:48:50Static isn't

00:48:51really discovery,

00:48:52sort of fun,

00:48:53but you could

00:48:53add and put

00:48:54in, if you

00:48:55want,

00:48:55manually.

00:48:56We'll go

00:48:57for dynamic,

00:48:58we'll hit

00:48:58add,

00:48:59and we'll

00:49:00put in

00:49:00either the

00:49:01DNS name

00:49:02or the

00:49:04IP address

00:49:05of our

00:49:06SAN.

00:49:07In our

00:49:07case,

00:49:08I'll just

00:49:08put in

00:49:08the address.

00:49:10If you

00:49:11are running,

00:49:12remember this

00:49:12runs over

00:49:13TCP,

00:49:13so if you

00:49:14are running

00:49:14this on a

00:49:15non-standard

00:49:16port,

00:49:17that's where

00:49:17you would

00:49:18go to

00:49:18configure it.

00:49:20You just

00:49:20change it

00:49:21right there.

00:49:23Authentication,

00:49:24it's going to

00:49:24inherit from

00:49:24the overall

00:49:25policy unless

00:49:26you want to

00:49:27change it

00:49:27here,

00:49:28but again,

00:49:28we're not

00:49:29doing

00:49:29authentication

00:49:29right now,

00:49:30so we

00:49:30can just

00:49:30hit OK.

00:49:32and it's

00:49:36going to

00:49:36give you

00:49:37this little

00:49:37message pop

00:49:38up here

00:49:38telling you

00:49:39that,

00:49:40hey,

00:49:40you just

00:49:41made some

00:49:41configuration

00:49:42changes

00:49:42that require

00:49:45that I go

00:49:46out and I

00:49:46re-scan

00:49:47all my

00:49:48storage.

00:49:50OK,

00:49:50so anything

00:49:51like this.

00:49:57One second,

00:49:59we got a

00:49:59question here.

00:50:01It looks

00:50:01like the

00:50:01SCSI

00:50:02ID,

00:50:03iSCSI

00:50:03ID is

00:50:04automatically

00:50:05generated

00:50:06when we

00:50:06change it

00:50:06to something

00:50:07intuitive.

00:50:09If you

00:50:09go back

00:50:09to

00:50:10properties,

00:50:11I assume

00:50:12you're

00:50:12referring to

00:50:13the iSCSI

00:50:13name here.

00:50:16You can.

00:50:18You can

00:50:18hit edit

00:50:19and you

00:50:19can change

00:50:20it right

00:50:20here,

00:50:21but it's

00:50:22still

00:50:22recommended

00:50:22that you

00:50:23follow

00:50:23the

00:50:24IQN.

00:50:26OK,

00:50:27to be

00:50:27honest with

00:50:28you,

00:50:28you're not

00:50:29really going

00:50:29to see

00:50:29this much

00:50:30anyway,

00:50:31because this

00:50:32is the

00:50:32initiator

00:50:33side.

00:50:34So if

00:50:35you want

00:50:35to change

00:50:35it to

00:50:35something

00:50:36intuitive,

00:50:36it's usually

00:50:37more important

00:50:38to do it

00:50:39on the

00:50:39storage

00:50:40side.

00:50:42It's actually

00:50:43sort of

00:50:43interesting.

00:50:44We'll see

00:50:44what mine

00:50:45looks like

00:50:45when it

00:50:46comes up,

00:50:46because I

00:50:47have changed

00:50:48mine to

00:50:49something

00:50:49intuitive,

00:50:50so to

00:50:51speak,

00:50:51but I've

00:50:52had it

00:50:53not work.

00:50:53If you

00:50:55want to

00:50:55use a

00:50:55friendly

00:50:56name,

00:50:57what's

00:50:57really

00:50:58more

00:50:58recommended

00:50:58is to

00:50:59change

00:50:59the

00:51:00iSCSI

00:51:00alias

00:51:01down

00:51:01here

00:51:01at

00:51:01the

00:51:01bottom.

00:51:02That's

00:51:03more

00:51:03what's

00:51:03meant

00:51:03to be

00:51:03the

00:51:04friendly

00:51:04name

00:51:04that

00:51:05you

00:51:05can

00:51:05use

00:51:06to

00:51:06show

00:51:07that.

00:51:08So it

00:51:08displays

00:51:09a little

00:51:09nicer.

00:51:13So,

00:51:14yeah,

00:51:14since we

00:51:14made this

00:51:15change,

00:51:15we do

00:51:15have to

00:51:16rescan.

00:51:17And you

00:51:18got to

00:51:18be real

00:51:18careful.

00:51:19Up here

00:51:19at the

00:51:19top,

00:51:20there's

00:51:22this

00:51:22button

00:51:23here,

00:51:23which

00:51:24I

00:51:24have

00:51:24highlighted,

00:51:25which

00:51:26says to

00:51:27refresh

00:51:28the host's

00:51:29storage

00:51:29information.

00:51:30Okay,

00:51:31so you

00:51:32click on

00:51:32that,

00:51:33you hit

00:51:33refresh,

00:51:34and pretty

00:51:35much all

00:51:36that does

00:51:36is refresh

00:51:38this

00:51:38display.

00:51:40That's

00:51:40pretty much

00:51:41all it

00:51:41does.

00:51:43The one

00:51:43we're looking

00:51:44for,

00:51:45if you

00:51:45go one

00:51:45more

00:51:46over,

00:51:47rescans

00:51:48all storage

00:51:49adapters to

00:51:50discover newly

00:51:51added storage

00:51:52devices and

00:51:53or volumes.

00:51:55That's the

00:51:56one we're

00:51:56actually looking

00:51:57for.

00:51:58So,

00:51:58we'll click

00:51:59on that

00:51:59one,

00:52:00and it's

00:52:00going to

00:52:00say,

00:52:01what do

00:52:01you want

00:52:01to scan

00:52:01for?

00:52:02Do you

00:52:03want to

00:52:03scan for

00:52:03new storage

00:52:04devices,

00:52:06or actually

00:52:07new VMFS

00:52:09volumes as

00:52:10well?

00:52:11I usually

00:52:12just leave the

00:52:12defaults on

00:52:13this.

00:52:13It's

00:52:14usually fine.

00:52:16So,

00:52:16I'll just

00:52:17hit okay.

00:52:19What we're

00:52:19looking for

00:52:20is these numbers

00:52:20right here

00:52:21to go

00:52:21up.

00:52:23That's

00:52:23what we're

00:52:23looking for.

00:52:28And there

00:52:29we go.

00:52:30So,

00:52:30what we've

00:52:31discovered

00:52:31is one

00:52:32target,

00:52:33so in

00:52:33other words,

00:52:33one SAN,

00:52:36okay,

00:52:37offering

00:52:37three LUNs,

00:52:39and of

00:52:40course,

00:52:40three paths

00:52:41because there's

00:52:42three paths,

00:52:43one to each

00:52:44LUN,

00:52:44through the

00:52:45one target.

00:52:46so that's

00:52:48the paths

00:52:48right now.

00:52:50If we come

00:52:51down here to

00:52:52the bottom,

00:52:52we go to

00:52:52devices,

00:52:53you can see

00:52:54three LUNs

00:52:54being presented.

00:52:56So,

00:52:56one 480 gig

00:52:58LUN,

00:52:58and then an

00:52:5811 and a

00:52:5912 gig LUN,

00:53:00those are just

00:53:00to have fun

00:53:01with them a

00:53:01little bit

00:53:01later.

00:53:03Okay?

00:53:04You can see

00:53:04these wonderfully

00:53:05friendly iSCSI

00:53:07disk paths

00:53:08that it creates.

00:53:09yeah,

00:53:12as far as

00:53:13changing those

00:53:13to something

00:53:14intuitive,

00:53:15not that I've

00:53:16found.

00:53:16No.

00:53:17We just

00:53:18want to live

00:53:18with it.

00:53:20Okay?

00:53:20Go to

00:53:21paths,

00:53:22and this will

00:53:22show you the

00:53:23paths it has

00:53:24to get there.

00:53:25Again,

00:53:25at this point,

00:53:26we only have

00:53:26one active

00:53:27path to get

00:53:28there.

00:53:30Okay?

00:53:31That's our

00:53:31target,

00:53:32and it should

00:53:33have put it,

00:53:34there you go,

00:53:35moves it over

00:53:36into,

00:53:37uh,

00:53:38the static

00:53:40as well,

00:53:41and what it

00:53:42should have

00:53:43discovered is

00:53:44a data store,

00:53:46and it may

00:53:47or may not

00:53:48have.

00:53:49Um,

00:53:50when,

00:53:50when you're

00:53:50going back

00:53:51and adding

00:53:52storage to

00:53:53the device,

00:53:53and best,

00:53:55probably quickest,

00:53:56easiest way to

00:53:57see this is you

00:53:57go back to

00:53:58summary here.

00:53:59Um,

00:53:59notice it still

00:54:00says no data

00:54:01stores have been

00:54:01configured,

00:54:02and data stores

00:54:03still say zero.

00:54:05So,

00:54:06this is a little

00:54:07bit of an

00:54:07interesting one.

00:54:08Um,

00:54:09there is a

00:54:11VMFS data

00:54:12store on

00:54:13that LUN

00:54:13that I

00:54:14dismounted.

00:54:16At least,

00:54:16um,

00:54:18at least there

00:54:18better be,

00:54:19because that's,

00:54:21the LUN we

00:54:21dismounted is the

00:54:23one that I'm

00:54:23pretty much going

00:54:24to use for the

00:54:24rest of the

00:54:24class.

00:54:25Uh,

00:54:26it's already

00:54:26got some

00:54:26virtual machines

00:54:27on it and a

00:54:28bunch of other

00:54:28stuff we need,

00:54:29so I'm really

00:54:31hoping that the

00:54:32LUN is still

00:54:33there,

00:54:34but no,

00:54:34it should be.

00:54:35But the host

00:54:37won't always

00:54:37reconnect

00:54:38to a

00:54:40VMware,

00:54:41or to a

00:54:41VMFS storage

00:54:42that's already

00:54:44configured like

00:54:44that.

00:54:45So,

00:54:46what we may have

00:54:47to do is

00:54:47actually right

00:54:48click here

00:54:48and say

00:54:50new data

00:54:50store,

00:54:50and it should

00:54:54come up,

00:54:55and you're

00:54:55going to say,

00:54:55of course,

00:54:55we're going to

00:54:56say VMFS

00:54:56like before,

00:54:58and what it

00:54:59should do is

00:55:00list the

00:55:01LUNs from

00:55:02iSCSI,

00:55:03and again,

00:55:04you notice this

00:55:04is almost exactly

00:55:05like we did

00:55:06in fiber channel,

00:55:08right?

00:55:10And it's

00:55:10certainly taking

00:55:11its time.

00:55:12Here we go.

00:55:14And the LUN

00:55:15is LUN 0,

00:55:17okay,

00:55:18now if you

00:55:18notice,

00:55:19let me see

00:55:20if I can,

00:55:23these windows

00:55:24are all in

00:55:24flash,

00:55:25which makes

00:55:25them a little

00:55:26interesting to

00:55:27try and resize.

00:55:29There we go.

00:55:31Okay,

00:55:31notice that it

00:55:31actually knows

00:55:32that it has

00:55:34a volume on

00:55:35there called

00:55:35shared one,

00:55:37which is

00:55:38good.

00:55:39That's correct.

00:55:40That's the

00:55:40volume we're

00:55:41trying to

00:55:41mount,

00:55:42but notice

00:55:42that when we

00:55:43told it to

00:55:43scan it,

00:55:44it didn't

00:55:45just latch

00:55:47onto that

00:55:48and bring

00:55:49it in here.

00:55:50And part of

00:55:50that is because

00:55:51obviously I

00:55:53created this

00:55:54data store

00:55:54previously,

00:55:55and then I

00:55:56had to

00:55:57unmount it

00:55:58from all of

00:55:59these devices

00:55:59so we could

00:56:01go back through

00:56:01and put it

00:56:02back in later.

00:56:03Obviously,

00:56:04this had to

00:56:05happen or else

00:56:06there wouldn't

00:56:06even be a

00:56:07data store

00:56:07there.

00:56:08So when

00:56:10you're adding

00:56:12storage in

00:56:13again that

00:56:14existed before,

00:56:16it doesn't

00:56:17just grab it

00:56:18automatically.

00:56:18We're going

00:56:19to have to

00:56:19tell it to.

00:56:20So we

00:56:21select this

00:56:22drive,

00:56:22that's fine,

00:56:23and we

00:56:23hit next,

00:56:23and it's

00:56:24going to

00:56:24ask us

00:56:24another

00:56:24question here.

00:56:25It's going

00:56:25to say,

00:56:25hey, we

00:56:28have an

00:56:28unresolved

00:56:29VMFS volume

00:56:31with this

00:56:32signature on

00:56:33this disk.

00:56:34So basically,

00:56:36VMware is sort

00:56:37of yelling at

00:56:37me here in

00:56:38a way,

00:56:39saying,

00:56:40there's

00:56:41already data

00:56:43here,

00:56:43there's already

00:56:44a data store

00:56:45on this,

00:56:46and it

00:56:47presents me

00:56:47with three

00:56:48choices.

00:56:49The first

00:56:49one is I

00:56:50could just

00:56:50give it a

00:56:51whole new

00:56:51signature.

00:56:53The data is

00:56:54going to be

00:56:54retained,

00:56:56the new

00:56:56signature gets

00:56:57assigned to

00:56:57the data store,

00:56:58and any

00:56:58references

00:56:59would be

00:57:01updated.

00:57:03And they'll

00:57:03still come back

00:57:04with their

00:57:04original name.

00:57:05What this

00:57:06would cause,

00:57:07though,

00:57:07is any

00:57:08other hosts

00:57:09that were

00:57:11already connected

00:57:11to this

00:57:12data store

00:57:12would have

00:57:13a problem.

00:57:15So if

00:57:16this was

00:57:17literally a

00:57:18data store

00:57:18that you

00:57:19had transplanted

00:57:19from somewhere

00:57:20else,

00:57:21and you

00:57:22were just

00:57:22trying to

00:57:22rebuild it

00:57:24into this

00:57:24new environment,

00:57:25that might

00:57:26be a reason

00:57:26to use

00:57:27that.

00:57:29Generally

00:57:29speaking,

00:57:30though,

00:57:30basically what

00:57:32we're trying

00:57:32to do is

00:57:32import a

00:57:34VMFS volume.

00:57:34That's

00:57:35essentially

00:57:35what we're

00:57:36accomplishing.

00:57:36We want

00:57:37to import

00:57:37this.

00:57:38We don't

00:57:39need to

00:57:39change it.

00:57:40It doesn't

00:57:40need a new

00:57:40signature.

00:57:41I'm not

00:57:42trying to

00:57:42prevent

00:57:43previous hosts

00:57:44that may

00:57:44have been

00:57:45connected to

00:57:45this before

00:57:46from getting

00:57:46to it.

00:57:48So all

00:57:49of those

00:57:49things being

00:57:50the case,

00:57:51I'll just

00:57:51keep the

00:57:52signature and

00:57:53keep all

00:57:54the data.

00:57:55That's what

00:57:55I'm going to

00:57:55do.

00:57:56Of course,

00:57:57the bottom

00:57:57choice here

00:57:58would be

00:57:59the whole,

00:58:01what do you

00:58:01mean there's a

00:58:02volume there?

00:58:02I don't want

00:58:03anything on

00:58:03there.

00:58:04This was

00:58:04supposed to

00:58:04be new.

00:58:05Okay,

00:58:05format it.

00:58:07You know,

00:58:07the whole,

00:58:08obviously I'm

00:58:08not going to

00:58:09do that one

00:58:10because as I

00:58:10mentioned earlier,

00:58:12I have stuff

00:58:13on here for

00:58:14other sections

00:58:15of the

00:58:15VMware course.

00:58:17So I

00:58:18certainly don't

00:58:18want to get

00:58:19rid of it.

00:58:20So I'm going

00:58:21to go with

00:58:21this option.

00:58:22I'm going

00:58:22to hit next

00:58:22and this

00:58:23usually takes

00:58:23a couple

00:58:24seconds.

00:58:25Okay,

00:58:26because it's

00:58:27going to go

00:58:27through and

00:58:28look at it.

00:58:29This resolving

00:58:30VMFS volumes

00:58:31usually takes

00:58:32a couple

00:58:32seconds.

00:58:33Let's give

00:58:33it a minute

00:58:34and this

00:58:36should come

00:58:36up and

00:58:37connect.

00:58:39And then

00:58:39storage here

00:58:40should go

00:58:40to one

00:58:40data store.

00:58:46Now you

00:58:46have to

00:58:46refresh it

00:58:47here.

00:58:47look detached

00:59:01using iSCSI.

00:59:02Now let's

00:59:14see if you

00:59:15can actually

00:59:15so it

00:59:20has

00:59:21definitely

00:59:21mounted

00:59:21it.

00:59:22Here it

00:59:23is and

00:59:23there are

00:59:24the files

00:59:24on it.

00:59:25And since

00:59:25we've only

00:59:26connected to

00:59:26that one

00:59:26host,

00:59:28okay,

00:59:29it's fairly

00:59:30obvious that

00:59:31that's the

00:59:31host that

00:59:31it's reading

00:59:32it through.

00:59:35For some

00:59:36reason,

00:59:36the summary

00:59:37page isn't

00:59:38updating

00:59:38here.

00:59:41Wouldn't be

00:59:41the first

00:59:42time.

00:59:42There we

00:59:42go.

00:59:43Data store

00:59:43is one.

00:59:44Okay,

00:59:44just took

00:59:44a minute.

00:59:45Give it

00:59:45a second

00:59:45to refresh.

00:59:48So there's

00:59:51an iSCSI

00:59:51data store

00:59:52mounted.

00:59:53Just

00:59:53basics

00:59:54getting

00:59:54started.

00:59:56Can

00:59:57only one,

00:59:58we have a

00:59:58question here,

00:59:59can only one

00:59:59LUN be

01:00:01accessed or

01:00:02used by

01:00:05multiple

01:00:06initiator.

01:00:10No.

01:00:13No,

01:00:13we can have

01:00:14as many as

01:00:14we want to

01:00:15access that

01:00:15LUN.

01:00:16In fact,

01:00:16we're about

01:00:17to have a

01:00:17whole other

01:00:18initiator and

01:00:18a whole other

01:00:19host access

01:00:20that LUN.

01:00:22Unless I'm

01:00:22misunderstanding

01:00:23your question.

01:00:25We're talking

01:00:25about multiple

01:00:26hosts

01:00:27connecting to

01:00:29the LUN.

01:00:31Absolutely,

01:00:31or else this

01:00:32wouldn't really

01:00:33be considered

01:00:33shared storage

01:00:34then.

01:00:35So we'll go

01:00:38ahead and do

01:00:38that next.

01:00:39If that's not

01:00:39exactly what you

01:00:40meant,

01:00:40maybe if you

01:00:40could reword it

01:00:42there if that's

01:00:42not what you

01:00:43were talking

01:00:43about.

01:00:44So let's go

01:00:45to host

01:00:46three here

01:00:47because we

01:00:47already have

01:00:48host two

01:00:49there is

01:00:50already talking

01:00:51to it.

01:00:53So on this

01:00:54guy, let's

01:00:55go to

01:00:56manage

01:00:57networking.

01:01:00And let's

01:01:01see what I

01:01:01have set up

01:01:02on this

01:01:02guy.

01:01:03So, see,

01:01:04his is

01:01:04right.

01:01:05He has

01:01:05an extra

01:01:06NIC

01:01:06that's

01:01:06connected

01:01:07to that

01:01:07virtual

01:01:07switch

01:01:08and this

01:01:08guy

01:01:08should

01:01:08have

01:01:09four

01:01:09of

01:01:09them.

01:01:11So that's

01:01:12good.

01:01:13So what we

01:01:14have here,

01:01:15but he still

01:01:16has this.

01:01:17How'd that

01:01:17not get rid of?

01:01:19He still has

01:01:20a storage port.

01:01:20He shouldn't

01:01:21have that.

01:01:22Sorry about

01:01:23that.

01:01:23It shouldn't be

01:01:24here.

01:01:24It needs to

01:01:24go away.

01:01:25I'm going to

01:01:25delete this.

01:01:26That's

01:01:32set up

01:01:32exactly the

01:01:33same way

01:01:33as host

01:01:33two was.

01:01:36We don't

01:01:36want to do

01:01:36that.

01:01:37That's no

01:01:37fun.

01:01:40Okay.

01:01:41So some

01:01:41of the

01:01:41questions we

01:01:42had before,

01:01:42let's see if

01:01:43we can do

01:01:43some different

01:01:43combinations

01:01:44here real

01:01:45quick.

01:01:45So if

01:01:47we were

01:01:48to say

01:01:48now one

01:01:53that I

01:01:54can't,

01:01:54the only

01:01:55one I

01:01:55can't really

01:01:56show because

01:01:56I don't

01:01:56have routing

01:01:58set up

01:01:58right now

01:01:59for the

01:02:00storage

01:02:01network.

01:02:04Not a

01:02:04real quick

01:02:05easy way to

01:02:05do it either

01:02:06really.

01:02:09Let's do

01:02:10this first.

01:02:10So let's

01:02:11do a VM

01:02:11kernel adapter

01:02:12just like

01:02:13before.

01:02:14This time

01:02:14though we're

01:02:15going to

01:02:15say a

01:02:15new

01:02:15standard

01:02:16switch

01:02:16okay

01:02:18and I'll

01:02:20give this

01:02:20one

01:02:20he has

01:02:23no

01:02:23available

01:02:23adapters.

01:02:24That would

01:02:25probably

01:02:25help.

01:02:26So I'm

01:02:27going to

01:02:27cancel

01:02:27this

01:02:28and for

01:02:28this

01:02:29switch

01:02:29I'm

01:02:30going to

01:02:30edit

01:02:30the

01:02:30existing

01:02:31switch

01:02:31first.

01:02:32We've

01:02:32got to

01:02:32get

01:02:32some

01:02:32adapters

01:02:33available.

01:02:35Try

01:02:35to cancel.

01:02:38This is

01:02:38the

01:02:38manage

01:02:39physical

01:02:39adapters

01:02:40for the

01:02:40switch.

01:02:42So I'm

01:02:43going to

01:02:43take out

01:02:44you know

01:02:45we're

01:02:45going to

01:02:45take

01:02:45out

01:02:45all

01:02:45of

01:02:45these.

01:02:47So

01:02:48nope.

01:02:49Nope.

01:02:52Nope.

01:02:53And there

01:02:54we go.

01:02:55So we're

01:02:56going to

01:02:56drop our

01:02:57I can't

01:02:58kill all

01:02:58of them

01:02:58because this

01:02:59is of

01:02:59course how

01:03:00we're

01:03:00managing

01:03:01this host

01:03:01right now.

01:03:02So if I

01:03:03kill all

01:03:03of them

01:03:04we'll

01:03:05be

01:03:05taking

01:03:07lunch

01:03:07so I

01:03:07could

01:03:07fix it.

01:03:08okay so

01:03:11our

01:03:12regular

01:03:12management

01:03:13switch

01:03:14that we're

01:03:14using to

01:03:14manage this

01:03:15host right

01:03:15now is

01:03:16just reduced

01:03:16to two

01:03:18network

01:03:18cards

01:03:19and now

01:03:20we'll

01:03:21add

01:03:21for

01:03:22the other

01:03:23one.

01:03:23So let's

01:03:23add

01:03:24VM kernel

01:03:26adapter

01:03:26for a new

01:03:29switch

01:03:29and we'll

01:03:32add

01:03:33VMNIC2

01:03:36and of

01:03:37course this

01:03:37would

01:03:37of course

01:03:38be one

01:03:38of your

01:03:39physical

01:03:39network

01:03:40cards

01:03:41in your

01:03:41host

01:03:42right that's

01:03:42what these

01:03:43VMNICs

01:03:44are.

01:03:45We'll just

01:03:46give him

01:03:46one for

01:03:46now

01:03:47and we'll

01:03:48say next.

01:03:52Some other

01:03:52things you

01:03:53could change

01:03:53by the way

01:03:53just real

01:03:54quickly

01:03:54under

01:03:57properties

01:03:57if it

01:04:00wants to

01:04:00load there

01:04:01for me

01:04:01and

01:04:02anyway

01:04:03basically

01:04:04just the

01:04:05CDP

01:04:05and the

01:04:05LDP

01:04:06okay

01:04:07it's just

01:04:09showing you

01:04:09what it's

01:04:09doing.

01:04:10We will

01:04:11call this

01:04:12IP

01:04:16storage

01:04:16one

01:04:17VLAN

01:04:19of course

01:04:20has to be

01:04:20my storage

01:04:21VLAN

01:04:21since I

01:04:21do not

01:04:22have

01:04:22routing

01:04:22going on

01:04:24for the

01:04:25storage

01:04:25network

01:04:25right now.

01:04:28Next

01:04:28and

01:04:31is going

01:04:34to be

01:04:3410

01:04:35dot

01:04:35250

01:04:35dot

01:04:36100

01:04:36dot

01:04:36whoops

01:04:37yes I

01:04:37know

01:04:38dot

01:04:40I mean

01:04:43this is

01:04:433 so

01:04:44I'll do

01:04:44233

01:04:46and so

01:05:03this is an

01:05:04example

01:05:04where the

01:05:05VM kernel

01:05:05port is

01:05:06going to

01:05:07go through

01:05:07a separate

01:05:08V switch

01:05:08okay so

01:05:11notice up

01:05:11here at the

01:05:11top it

01:05:13created a

01:05:13whole new

01:05:14V switch

01:05:15V switch

01:05:15one now

01:05:16and now

01:05:17this is still

01:05:18a standard

01:05:18virtual switch

01:05:19we're not

01:05:19going to get

01:05:19into distributed

01:05:20switches or

01:05:21anything else

01:05:21today but

01:05:22those could

01:05:23be used

01:05:24just as

01:05:25easily no

01:05:25no difference

01:05:26from that

01:05:26perspective

01:05:27okay so

01:05:28here's a

01:05:28one-to-one

01:05:29relationship

01:05:29between our

01:05:31VM kernel

01:05:32port and

01:05:34our physical

01:05:35adapter

01:05:35okay so

01:05:38let's stick

01:05:39with just

01:05:39this for a

01:05:40minute and

01:05:41let's go

01:05:42over to

01:05:42storage

01:05:43and let's

01:05:46add our

01:05:46iSCSI

01:05:47adapter

01:05:48and let's

01:06:05go to

01:06:06network port

01:06:06bindings and

01:06:07this one

01:06:07should

01:06:08notice it's

01:06:10showing up on

01:06:10this one now

01:06:11for the port

01:06:11binding

01:06:12okay

01:06:13the reason for

01:06:15this of course

01:06:16is because it

01:06:17is connected

01:06:19all by itself

01:06:20on its own

01:06:22switch

01:06:23okay

01:06:24now if you

01:06:25take a look at

01:06:26this down here

01:06:26you can set the

01:06:27port groups

01:06:27that it's going

01:06:28to use

01:06:29and of course

01:06:29this is using

01:06:30the IP

01:06:30storage

01:06:31label

01:06:32let's see

01:06:33make this a

01:06:33little bigger

01:06:34here

01:06:34okay

01:06:37which switch

01:06:38it's exiting

01:06:39through

01:06:39and its

01:06:40properties

01:06:41which VM

01:06:42kernel adapter

01:06:43it's using

01:06:45okay

01:06:46and the

01:06:48physical adapter

01:06:48okay now what

01:06:55we should be

01:06:56able to do

01:06:56here

01:06:56again let's

01:07:02get it

01:07:03working

01:07:03first

01:07:04so let's

01:07:06go to

01:07:06targets

01:07:06and we'll

01:07:09add our

01:07:10target again

01:07:11let it

01:07:25finish

01:07:25and then

01:07:27we'll

01:07:27re-scan

01:07:28and again

01:07:31hopefully it

01:07:32should see

01:07:32the storage

01:07:33here

01:07:33so we

01:07:37should see

01:07:37our targets

01:07:37go to

01:07:38one

01:07:38devices

01:07:38to three

01:07:39and three

01:07:39and there

01:07:39we go

01:07:40okay

01:07:41so there

01:07:42it is

01:07:42working

01:07:42through a

01:07:43dedicated

01:07:43switch

01:07:43with a

01:07:44dedicated

01:07:44NIC

01:07:44okay

01:07:46now

01:07:47again

01:07:48from

01:07:49a

01:07:50multi-pathing

01:07:51with routing

01:07:51it would

01:07:54simply be

01:07:54another

01:07:54VM kernel

01:07:55port

01:07:56on a

01:07:57separate

01:07:57subnet

01:07:58and then

01:07:59the routing

01:08:00takes care

01:08:00of that

01:08:01okay

01:08:02but and

01:08:04I'll be honest

01:08:05here I have not

01:08:05tried this in this

01:08:06particular design

01:08:08but what we should be able

01:08:09to do here

01:08:10let's give it a shot

01:08:11let's go back to

01:08:11networking here

01:08:12and if we go back

01:08:17just to just to have it

01:08:18as a reference

01:08:20let me change this here

01:08:23real quick

01:08:24this is what we're going

01:08:26to be going for here

01:08:27we're going to go back

01:08:27just a second

01:08:28port bindings

01:08:31right

01:08:31so the idea is

01:08:34okay

01:08:35they're not

01:08:36they're not

01:08:37on different

01:08:38subnets

01:08:38they're not

01:08:40on different

01:08:40switches

01:08:41okay

01:08:42and not

01:08:44the array

01:08:45on a different

01:08:45subnet

01:08:46so we're not

01:08:46going to break

01:08:47any of these

01:08:48rules

01:08:48with the

01:08:49port binding

01:08:50what we're going to do

01:08:51is we're going to

01:08:51come back over here

01:08:52and we're going to

01:08:55create

01:08:55new adapter

01:08:58a new

01:08:59VM kernel

01:08:59network adapter

01:09:00it's going to go to

01:09:05an existing standard

01:09:06switch

01:09:06and switch one

01:09:08is already selected

01:09:09I don't think we have

01:09:09to actually do that

01:09:10but let's just make sure

01:09:11because that was the

01:09:12default anyway

01:09:13hit next

01:09:14and we'll call this

01:09:17IP storage 2

01:09:21VLAN is 1000

01:09:27we'll make this 234

01:09:38there we go

01:10:00okay

01:10:01so now we have

01:10:02two

01:10:02VM kernel ports

01:10:04on the same switch

01:10:05going out

01:10:07to the same

01:10:07physical adapter

01:10:08if we go back

01:10:11to storage

01:10:12yeah

01:10:22let's

01:10:23take that out

01:10:24and there we go

01:10:32now we've got

01:10:32two bindings

01:10:33okay

01:10:37I actually didn't have

01:10:38to take out

01:10:38the other one

01:10:38I guess

01:10:39didn't really need

01:10:40to do that

01:10:40but this allows

01:10:43us to use

01:10:44both

01:10:46VM kernel ports

01:10:47as sources

01:10:52and now what this

01:11:07would do for us

01:11:08if we go back to

01:11:15networking

01:11:16manage physical mix

01:11:21see here's

01:11:39see here's the advantage

01:11:39okay

01:11:41is once we add

01:11:43multiple physical adapters

01:11:44and then we go to the properties

01:11:48properties for the switch

01:11:49and we go to teaming and failover

01:11:51route based on originating virtual port

01:11:54that's going to mean something now

01:11:57because now we have two separate originating virtual ports

01:12:02so what this does is it allows the VMware kernel to initiate traffic going to storage

01:12:10from two separate source IP addresses

01:12:14therefore allowing your nick teaming to actually do its job

01:12:21to actually be able to share between these two ports

01:12:25so if we come back over here

01:12:27so if we come back over here to storage

01:12:28uh-oh

01:12:30and now it doesn't like that

01:12:32yeah it's not going to let us do it that way

01:12:40that's right

01:12:41he's afraid of that

01:12:42might have to

01:12:45might have to

01:12:45because what we do

01:12:59i might have to

01:13:00your

01:13:02over here

01:13:03iMAX

01:13:04it's not just

01:13:05like that

01:13:06iMAX

01:13:07iMAX

01:13:08Best

01:13:09like that

01:13:09in your

01:13:10lets

01:13:11do you

01:13:12to

01:13:12Yeah, it's not liking those now, so it's not going to let us do the teaming like that.

01:13:25So it's using it, but it's using it from the other teaming.

01:13:32So yeah, it's not going to let us do it that way.

01:13:45Let's put it back.

01:14:06There we go.

01:14:11Again, all it's really letting you do is source it from multiple sources.

01:14:18All right, so it still sees it as six paths.

01:14:24So it's multi-pathing, going out to the two targets.

01:14:30So that's all working.

01:14:33And if we go back here, no data stores.

01:14:40So we're going to have to add it.

01:14:45Yeah, hold on.

01:14:47It's yelling that network redundancy was lost.

01:14:49Yeah, we know.

01:14:50Reset it.

01:14:51Not worried about that.

01:14:53Because it should be back anyway.

01:14:56New data store.

01:15:00And 11, zero.

01:15:12Keep existing signature because I don't want to kick the other guy off.

01:15:19So it's resolving.

01:15:32Oh.

01:15:33Okay.

01:15:34So he should just reconnect to it then.

01:15:39So he should just reconnect to it then.

01:15:58Let's have him rescan.

01:16:21He's telling me it's already got the VMFS data store, but he's not connecting to it yet.

01:16:36He's not connecting to it yet.

01:17:01Nice.

01:17:08Actually, my blood is reconnected from over here.

01:17:26Let's see here.

01:17:44And my other host isn't finding it at the moment.

01:17:49So I might have to reload that host to get him to rescan it.

01:17:58So the important thing for right now, because we're going to deal with data centers after

01:18:04break.

01:18:05The important thing for right now, and I'm going to add the other host at break as well.

01:18:11Let's see.

01:18:12We had a question here.

01:18:14We now have two iSCSI adapters mapped to the same target.

01:18:20Two separate iSCSI adapters from two different hosts.

01:18:24Yes.

01:18:25They are redundant, but not load sharing.

01:18:32Well, you have two different VM kernel.

01:18:35Actually, reading that, I think you're talking about the same host.

01:18:38We have two different VM kernel ports, but that's all.

01:18:45Okay.

01:18:46So we're going to go ahead and take our 30 minute break at this point.

01:18:53And I'm going to, while we're at break, I'll get these other two hosts connected to the

01:19:00data store, like they should be.

01:19:03Because I'm not sure we're going to use host one, but just in case, I'll get it connected.

01:19:09So that's all good to go.

01:19:11So we'll start back up in half an hour.

01:19:15And we'll start with data stores, working with them and all that kind of fun stuff.

01:19:20Well, Amen.

01:19:25,

01:19:30.

01:19:39So we almost can go ahead and turn into a hybrid.

01:19:42And you can see a solid life of human spectrum.

01:19:44And then you'll see them again.

01:19:46All right.

Category

Transcript

Be the first to comment

Recommended