Network Topologies and Types

0:00<v ->Hello and welcome.</v>

0:01My name is Keith Barker, and when it comes to networking,

0:04the concept of one-size-fits-all does not apply

0:08because we have lots of different needs

0:09and uses that we have networks for.

0:11So we're gonna have different topologies,

0:13both physical and virtualized topologies,

0:15as well as different types of networks.

0:17And so it's those concepts that I like to discuss

0:20with you in this set of videos.

0:21Welcome aboard.

0:00<v ->In this video, I'd like to chat with you about underlay</v>

0:02and overlay networks.

0:03And if you're thinking, what the heck is that,

0:05you're in the right place.

0:06Let's begin.

0:07So let's begin with this topology.

0:08Here we have some computers that are connected to a switch,

0:12and those switches are interconnected together,

0:14and their switches are connected to a router,

0:16and that router goes to a firewall

0:17and then it goes to another router,

0:18and then it goes off to a service provider router.

0:21And then in the service provider,

0:22they probably got lots of routers as well.

0:24And then that service provider talks

0:25to another service provider with lots of routers.

0:28So if we had a server here at the branch office

0:31that we are trying to access from computer one,

0:33the literal path may go something like this.

0:35Into the switch down to this switch up to the router,

0:39through the firewall to this router,

0:41and then through one or more of these routers

0:44till it's finally routed to the branch office,

0:46and then it goes to the server.

0:48So that would be the actual path.

0:49Now all these connections

0:51and the routers and devices that make up that path,

0:53that's considered to be the underlay network.

0:55Think of the underlay network

0:56like the real networking devices, and cables,

0:59and connectivity used to deliver a packet, for example,

1:02from computer one to a remote server.

1:05And as part of that underlay,

1:06we're gonna have a bunch of physical connections.

1:09And those physical connections could be leading to

1:11like copper cabling, or fiber cabling, or wireless,

1:17depending on what we're using in our infrastructure.

1:19And for this connection right here between router two

1:22and our service provider that leads off to the internet,

1:24let's take a look at an example of

1:26what some of the options are

1:27right there for that connectivity.

1:29It could be using cable modem technology,

1:31which is delivering the signals over coax cable,

1:34at which point we'd have either built into the router

1:36or right next to the router, a cable modem device

1:39that then translates those signals into ethernet

1:42so the router could use it.

1:43Another option to deliver high speed internet access

1:46is to use DSL, which is using the older telephone lines.

1:49So I'll just make a note there that it's phone cable

1:52with the appropriate device right here

1:53to interpret those signals

1:54and then convert them over to ethernet

1:56so that our router could use them.

1:58Or we may have a router that has DSL capability

2:00built into it as one of its interfaces.

2:02Another option for this connection could be a leased line.

2:06And then we do the appropriate technology

2:07to support that lease line connection.

2:09And at the point at our location where we connect our stuff

2:12to the service provider stuff,

2:14that's referred to as a demarcation point.

2:16And the provider at that point,

2:18they may just provide us a jack that we just plug into.

2:20And that's often called a smart jack.

2:22And effectively the demarcation point is where our stuff,

2:25everything to this side of that connector

2:27is our responsibility,

2:29and everything to the right hand side

2:30is the service provider's responsibility.

2:32Other options for delivering connectivity

2:35from a service provider would be an optical connection

2:39using some type of fiber.

2:41And these all represent physical connectivity,

2:43but we could also get service via wireless.

2:45So we're regarding wireless, here are some of our options.

2:49We could use cell technology like 5G,

2:52or we could have satellite,

2:53or any other wireless technology

2:55that's being provided by the service provider.

2:58So regardless of the type of media that we're using,

3:00the path from this computer going out, for example,

3:03to the server, that's all part of the underlay network.

3:06So when I think of underlay,

3:07I think of the real networking devices

3:09and the interfaces that are used

3:11to forward packets through the network.

3:13And that leads us to our next conversation about overlay.

3:16What the heck is an overlay?

3:17Well overlay is taking our infrastructure,

3:20the underlying network,

3:21and then logically placing on top of that

3:24a logical network or a logical path.

3:27Here's an example.

3:27Let's imagine that we wanted to build

3:29a virtual private network between the headquarters site,

3:32that's everything over here, and the branch office.

3:36What we could do is we could configure a logical tunnel

3:40between those two sites,

3:41headquarter site and the branch office.

3:43And even though the actual packets

3:45would go through our service provider,

3:46and then through various routers here,

3:48and various routers here,

3:49and then over here logically with this overlay

3:52and overlaying a tunnel between those two sites,

3:55logically, those two sites that aren't directly connected

3:58with this one logical network.

4:00And that's an example of an overlay.

4:02So think of an underlay as the real interfaces

4:05and the real path,

4:06and an overlay being made up networking

4:09that you're putting on top of your existing infrastructure.

4:12And here's an example of an overlay technology

4:14where we could create a tunnel logically between two sites.

4:17And that would be using GRE.

4:20That's an acronym for generic router encapsulation.

4:22And there's several benefits of setting up a logical tunnel.

4:25If we set up a logical tunnel between, for example,

4:27the headquarter site and the branch office,

4:29they could dynamically communicate with each other

4:32over this very simple GRE tunnel.

4:34Now, in measurable terms,

4:36the traffic really is going over the underlay network

4:38and being routed and forwarded,

4:40but logically, the traffic is being forwarded

4:42through this GRE tunnel.

4:44Now, one of the problems with a GRE tunnel

4:46going over an untrusted network like the internet,

4:48is that any service provider who's in the path,

4:51they have an opportunity to see the contents of that packet,

4:55including the payload and what protocols are being used.

4:58So in the case of something like HTTP, or Telenet,

5:01or some other protocol that doesn't have built-in protection

5:04and encryption, that traffic is exposed.

5:06So oftentimes we'll use GRE tunnels for the logical overlay,

5:10but we'll also protect it

5:12with a security protocol called IPSec,

5:14which takes each and every packet,

5:16and it encrypts it before it ships it over.

5:18And another benefit of doing tunneling

5:20and trickery with overlay networks

5:22is that we can move traffic

5:24that normally wouldn't be forwarded on the internet.

5:26Case in point, let's imagine

5:28that we are running IP version six here on our network

5:31and also IP version four.

5:33So we're running both versions of IP.

5:35However, perhaps not every device on the internet

5:37in our path is able to support IPV6.

5:40So what we could do is build a tunnel

5:42between the headquarter site and the remote site,

5:45a GRE tunnel,

5:46and then we could hide inside of that our IPV6 traffic.

5:50So the internet would see GRE traffic, for example,

5:53with IPV4, but its actual payload is IPV6.

5:58So here's the big picture regarding underlay and overlay.

6:00Underlay represents the actual interfaces

6:03and paths that are being used

6:05as that traffic is being forwarded across networks.

6:08And that logical tunneling that we're doing on top of it

6:11is called the overlay.

6:12And to help reinforce this, let me show you a packet capture

6:15of a GRE tunnel, which is using IPV4

6:18to connect the two sites together.

6:19However, inside of that GRE tunnel,

6:22it's actually carrying IPV6 traffic.

6:24So this is Wireshark.

6:25It's a protocol analyzer.

6:27I captured some traffic

6:28that was being sent over a GRE tunnel.

6:30So from the networking perspective,

6:32the outer protocol is IP version four

6:34coming from this IP address, going to that IP address.

6:37Instead of having TCP or UDP, instead GRE,

6:41which is yet another layer for transport protocol.

6:45And then inside the GRE header,

6:46it then indicates what it's carrying in the payload.

6:49Into the contents of this IPV6, it is a ping reply.

6:52So the reason I brought that protocol analyzer

6:54was just to reinforce the idea

6:56that we can logically build these tunnels,

6:58and then in those tunnels we can send

7:00other types of traffic.

7:02In the example we just looked at,

7:03we have a GRE, generic routing encapsulation tunnel

7:06between two routers.

7:07It was router one and router two.

7:09And so from the outside world,

7:10it would just looked like IPV4 traffic being forwarded back

7:13and forth between the two piers, the two ends of the tunnel.

7:16But as we looked at the payload,

7:17it was actually carrying IPV6 traffic.

7:20So that's an example of an overlay network

7:22where we're logically placing on top of our infrastructure

7:25based on a business need.

7:27So in the case of this example here,

7:29maybe this network didn't support IPV6.

7:31So we set up an IPV4 GRE tunnel,

7:33and then we tunneled the actual IPV6 traffic.

7:35Another example of this would be create a GRE tunnel,

7:38and then use IIPSec to protect that traffic

7:41as it goes back and forth between the headquarter site

7:44and the branch office.

7:45And the idea or the concept of a topology,

7:48that idea can apply to the actual real network

7:51or the overlay network.

7:52So in the next video, let's take a look

7:54at some of the topologies that we're likely to see

7:56in both the underlay and the overlay.

7:58I'll see you there in just a moment.

0:00<v ->In the world of networking,</v>

0:01we have the concept of a topology,

0:03which basically means it's like how things are laid out.

0:05Like what is the topology for your house?

0:07Where are the rooms, where are the things?

0:09And in networks, we have typologies

0:11that we can apply to the underlay network,

0:13that's like the real network

0:14with the routers and the switches,

0:15and the connections,

0:17and the path that goes through the network.

0:18And we also have various typologies

0:19that we could use for the overlay network,

0:21like the GRE tunnels,

0:23or the IPSec tunnels

0:24that we can place logically on top of that underlay network.

0:27So in this video,

0:28you and I get to take a closer look at topologies

0:31for both the underlay and the overlay.

0:33So let me take you back in time for a little bit of this,

0:35it'd be a lot of fun.

0:36And that is back when I first got into networking

0:39a long, long time ago,

0:40we used coaxial cable

0:43for the networking of computer devices.

0:46So we'd have a piece of coax on each end,

0:48we'd have a little terminator

0:49that would stop any signals

0:50from being bounced back onto the wire,

0:52and then we'd have little taps

0:54depending on the type of ethernet we are using

0:56and that would tap in for the computer.

0:58So there's computer A,

0:59and there's computer B,

1:01and there's computer C and so forth.

1:04And this was referred to as 10base2.

1:07And the 10 refers to 10 megabits per second.

1:09The B is baseband,

1:11meaning there's only one signal,

1:12one frequency present at a time

1:14on this shared network segment

1:16and the two represented fairly close, not exactly,

1:18but close to 200 meters,

1:20which referred to how long this network could be.

1:22And regarding what topology would we call this?

1:25We call it old, but the actual concept is called a bus.

1:28It's a shared bus,

1:29everybody's connected to that same medium

1:32and because they're all sharing this network,

1:34only one device,

1:35think of it like a one lane road,

1:37only one device can speak at a time,

1:39and if two or three devices try to talk

1:41at the same exact moment,

1:42there's gonna be a collision.

1:44So in this bus topology

1:45where everybody's sharing the same network,

1:47it's referred to as one collision domain,

1:51which effectively means that only one device

1:52can talk at a time.

1:54This also can be referred as one broadcast domain

1:56because if one device speaks on the network

1:59and sends a broadcast packet into the network,

2:02everybody else on that network segment is going to see it.

2:05Now one of the big challenges with this bus topology

2:07back in the day was that

2:08if there was a break anywhere in the cable,

2:11or somebody just, you know,

2:12opened the connection here,

2:13basically the entire network would go down

2:15if we had a single fault between any of these devices

2:18on this bus.

2:19So about a couple decades ago,

2:20they came up with a device called a hub.

2:23And a hub is simply a multi-port repeater.

2:26Effectively, we could take all of our devices

2:28and plug them into this hub.

2:29So we have device A, and B, and, C, and D.

2:33This little hub is a layer one device.

2:35So any signals that are sent in on this port

2:38that device A is connected to,

2:40it would just be repeated and forwarded out

2:42all the other ports.

2:44So in this topology,

2:45this is a logical bus

2:46because they're still sharing logically

2:48a same similar network across the board here.

2:51And with a layer one hub,

2:53only one device can talk at a time.

2:54So there's still one collision domain,

2:57and because they're all in the same network there,

2:59there's also one broadcast domain.

3:00But the benefit is these cables here

3:02used unshielded twisted pair

3:05versus coax cable,

3:06which is used in this 10base2 scenario.

3:09So here this is called 10baseT,

3:12as in twisted pair with this example using a hub.

3:16So so far we have a physical bus,

3:19here we have a logical bus.

3:20But if we look at the hub as a center point here,

3:23it's actually wired as a physical star

3:25with a physical hub being the center of the star

3:27and then the actual hosts

3:29that are connected with the unshielded twisted pair

3:31making up the rest of the star.

3:33So if somebody just said,

3:34'What is this network topology?"

3:36You'd have to say,

3:37"Well, it's physically a star,

3:38"but it's logically a bus."

3:40So you can see how it can get a little bit tricky.

3:42So at this topology,

3:43we could say that this is a physical star,

3:45but still a logical bus using a hub at the center.

3:49Now one of the big challenges with a hub back in the day

3:51was that only one device on this network

3:54could communicate at a time

3:55because we had one giant collision domain.

3:57So they came up with a new device,

4:00and that new device is a layer two switch

4:03and it operates at layer two

4:04and it learns and memorizes the layer two addresses

4:07for everybody that's connected to it.

4:08So once again, if we have four devices that are connected,

4:12and once again with unshielded twisted pair,

4:14same cabling that we had with the hub,

4:17and here we have host A, and host B,

4:20and host C and host D.

4:21Now with the switch,

4:22because the switch is aware of the layer two addresses,

4:24if host A wants to communicate with host D,

4:27the traffic goes into the switch

4:29and the switch on its back plane

4:30just forwards it over to the port that's needed.

4:32So host B and host C don't have to see it.

4:35So effectively every port on a switch,

4:38on a layer two switch,

4:39is its own collision domain.

4:40So if this is a four port switch,

4:42I'll put four port switch.

4:44Effectively we have four collision domains.

4:47If we have a 28 port switch,

4:49we have 28 collision domains.

4:51Every single port on its own

4:53is basically a dedicated highway

4:55from the device to get to the switch.

4:56And then once traffic gets there,

4:58it's the switch's responsibility

4:59to forward it appropriately.

5:01Now, in the case of an unmanaged switch,

5:03which is the layer two switch,

5:04but we haven't carved it out

5:05or done anything too special with it,

5:07we still, with this switch, we have four ports,

5:09we still have one broadcast domain.

5:12Now what that means is that if this computer

5:14sends a broadcast packet or a broadcast frame

5:17into the switch,

5:18the switch says,

5:19"Oh my goodness, it's a broadcast

5:20"so everybody must need it.'

5:22So it forwards it out to all the other ports

5:25connected to that switch.

5:26And it's this type of a layer two switch today

5:29that most devices,

5:30unless they're wireless,

5:31are gonna be connected in.

5:32So in an office space,

5:34we have a computer,

5:35it plugs into a wall jack,

5:36that wall jack leads off to a wiring closet,

5:39and that wiring closet there is a patch panel

5:41that then leads to a switch.

5:42So effectively, all of our end devices these days

5:45that are wired

5:46are connecting into a layer two switch.

5:48So from a physical perspective,

5:50regarding the topology,

5:51this switch, this layer two switch,

5:52is physically wired as a star,

5:55but logically it's also operating as a bus.

5:57However, with the benefits of giving each device

6:00that it's connected to their own dedicated freeway,

6:03their own separate collision domain.

6:05So as I mentioned, a four port switch

6:06with four devices connected

6:08would be representing four individual collision domains,

6:11one for each port.

6:12Alright, and one other topology

6:14I'd like to chat with you about right here

6:16regarding local area networking

6:18is the concept of a ring.

6:20So let's imagine we have those four hosts, again,

6:22A, B, C, and D,

6:26but this time we're using some technology

6:28that connects them all in a ring fashion.

6:32So in this case that'd be a example of a physical ring.

6:35And with the ring topology,

6:36we could have a little token that's being sent to device A

6:39that says, "Hey, do you wanna talk?'

6:41And A says, "No, I'm good."

6:42And then that token goes to B.

6:43It's like a talking stick, you know,

6:45giving each device a chance to communicate.

6:47And then for that traffic around the ring.

6:48So this would be an example of a physical ring topology

6:51and also a logical ring.

6:52However, when I was first getting my teeth cut

6:55with networking back in the 80s,

6:57we had something called token ring.

6:59And token ring, we'd have a MAU,

7:01a multi station access unit, I think it was,

7:03think of it like a hub for token ring.

7:05And then you'd have,

7:06for example, four devices connected.

7:07So it was A, B, C, and D.

7:10And you might think,

7:11well, that's a physical star,

7:12and you would be right.

7:13So physically it's wired as a star,

7:15just like a layer two switch or a hub,

7:17physically it looks like a star.

7:18However, logically behind the scenes

7:20what happens with token ring

7:23is we have that little token

7:24that's sent to each of the devices.

7:25So host A, do you have anything to say?

7:27Nope, and then it goes to host B,

7:29do you have anything to say?

7:29And then C and D,

7:30and then it flips all the way over to A again.

7:32So with token ring,

7:34that's an example of a physical star,

7:36but it's an example also a logical ring.

7:40So in the back plane of this multi station access unit,

7:43it actually loops around when it gets to the end

7:45and then forwards it to the first device

7:47in the path once again.

7:48So the great news today regarding networks is that

7:50we aren't using token ring anymore

7:52and most of the technology we're gonna see

7:54in local area networks at our companies

7:56is gonna be layer two switches

7:57connected to our end stations,

7:59and we'll have separate sets of videos

8:00regarding the interconnection of switches and trunking

8:03and all that good stuff.

8:04So it's this one right here on the local area networks

8:07that we're gonna see most of the time

8:08if we're using wired technologies.

8:11So now that we've taken a look at some options

8:13regarding topologies for local area networks,

8:16let's take a look at some topologies

8:17we might see and work with

8:19with wide area networks.

8:21And when we see the term wide area networks or WAN,

8:24it just refers to, you know,

8:26sites that are not geographically close to each other,

8:29they're not in the same building,

8:30they're not in the same city, perhaps,

8:32maybe they're geographically dispersed,

8:34maybe one on the west coast,

8:35one on the east coast,

8:36one in another country, et cetera.

8:37So let's imagine that we have a headquarter site

8:40and we'll call that site number one.

8:43So it's got some computers and networks

8:45and then it has a router or some other device

8:47that's connecting to a cloud

8:49and that cloud,

8:50and the reason we draw clouds a lot is because

8:52we don't have to identify explicitly what's in that cloud.

8:54It means like stuff.

8:55So we could have a service provider here,

8:57or it could be the internet.

8:58And then let's imagine that we have site two

9:01and they've got a network

9:04with some hosts and servers connected.

9:06They've got a router or some other device

9:08that's connecting them.

9:09And when I say some other device,

9:10it could be a firewall that's doing routing services,

9:12but they've got some devices connecting them

9:14to the service provider cloud.

9:16And again, that could be the internet

9:18or a private service provider wide area network services.

9:21So the key with wide area networks

9:22is that site one and site two

9:24are not in the same building.

9:26They're geographically separate

9:27and they need some WAN connectivity,

9:29wide area network connectivity to talk with each other.

9:32And let's also bring in another site,

9:34let's go ahead and bring in site three.

9:36Site three has a network,

9:38some devices on that network,

9:40and they've got a router or firewall that's connected

9:42providing wide area network connections.

9:44So at each of the sites,

9:46it's very likely that they're using star topologies

9:48with switches and ethernet connections

9:51going to their workstations.

9:53They could also be using wireless,

9:54we'll have some separate videos on wireless as well.

9:56Another question comes in

9:58regarding how do we wanna connect the sites together?

10:01Do we wanna have site two connect to headquarters

10:04and site three connect to headquarters?

10:06And if we do, that'd be an example of hub and spoke.

10:11With the headquarter site being the hub,

10:13or the central connection,

10:14and then logically we have site two being one spoke

10:17and site three being another spoke.

10:19Another question is,

10:20do we wanna have site two and site three?

10:22Do we wanna have them logically

10:23be able to connect to each other

10:25or do we wanna have to go through the headquarter site?

10:27Because we may do this as well,

10:28we may say, you know what?

10:29We wanna go ahead and logically connect

10:31site two to site three.

10:32And if we did that,

10:33that's where every site has a connection to every other site

10:36that's referred to as a full mesh.

10:40Now, full meshes, we know with logical tunnels

10:43and connections between all sites is wonderful,

10:45but if you have like 30, or 40, or 50 sites,

10:49it may not be practical,

10:51especially if you're doing it manually

10:52to implement the logical topology,

10:54which is effectively an overlay

10:56on top of this cloud network,

10:58an overlay of the tunnels or the paths that you want

11:01between the sites.

11:02So a full mesh is when every site has a direct connection,

11:06whether it's physical or logical to every other site.

11:10And as an example of how that can get pretty dicey,

11:12if we had, let's just say five sites,

11:14it'd be the connection would be like this,

11:18and then like this,

11:20and then like this,

11:21and like that

11:22and I think we have all the connections in place.

11:24And so you can see the more devices or more nodes you have,

11:27the more complex a full mesh is going to be.

11:30So we could also do a hybrid approach

11:33where perhaps we have some of these connections

11:35but not others.

11:36Lemme show you what I mean.

11:38So let's say we have,

11:40I'll take out the links there

11:41and let's say this is HQ1 and HQ2 for fault tolerance.

11:46And then we have connections that go up to HQ1 and HQ2

11:51from everybody for fault tolerance,

11:52and then we have some connections here between.

11:54So this would be an example of a partial mesh or a hybrid

11:57where we don't have full connections

11:59from every device to every other device,

12:01but we do have connections from some of the spokes

12:02up to the hubs.

12:03In fact, for fault tolerance,

12:04we may wanna do this and that as well.

12:07So device one and two are connected to everybody,

12:10and site three, four, and five

12:12are only connected up to the hubs.

12:13And one of the big reasons that I talked about

12:15the underlay and overlay in a previous video

12:17is the fact that when we're designing wide area networks

12:20and we're logically putting these tunnels in,

12:22we can logically configure anything we want

12:25with the overlay.

12:26As long as we have connectivity

12:27from the headquarters site,

12:29to site two, to site three,

12:30we can then logically with an overlay

12:33apply the tunnels to provide the logical connectivity

12:36over the service provider,

12:38or the wide area network,

12:39or the internet,

12:41whatever we're using for the actual transmission

12:42of those packets.

12:43So as you can see, there's a lot of flexibility that we have

12:46on both the underlay network

12:47as far as the topology and the logical overlay

12:50that we place on top of that network.

12:52Now, not every single network

12:54has the same purpose or same function,

12:56and so in the next video I'd like to chat with you

12:58about various network types,

12:59their names and their purposes.

13:01I'll see you there in just a moment.

0:00<v ->Not every network has the same purpose</v>

0:02or same function or the same scope.

0:03And so in this video I'd like to chat with you

0:05about various network types

0:06we're very likely to come across.

0:09So let's begin by drawing a network,

0:10and I'm gonna do it in a cloud fashion

0:12that just represents a collection of network connectivity

0:15without having to describe the itty-bitty details

0:17inside that cloud.

0:19So let's imagine we have two nodes

0:20and we have device A that's connected

0:23and device B that's connected,

0:24and they wanna communicate with each other.

0:27However, perhaps they're not dedicated servers,

0:29maybe they're just workstations,

0:30they're running software

0:31and they wanna be able to share files

0:33or something else back and forth,

0:35and just do it between those two devices.

0:37That would be an example of a peer-to-peer network,

0:42where neither device has a dedicated role as a client

0:45or a dedicated role as a server,

0:46but they can act as either one,

0:48as they communicate and share data back and forth.

0:51A long, long time ago when the internet was still young,

0:54we had some things like Napster for music sharing,

0:57and that's an example of a peer-to-peer networking system,

1:00that was intended to share music.

1:03Now, a more typical example of a network type

1:05that we're gonna come across,

1:06is when we have dedicated servers that are out there.

1:09So let's call this server one,

1:11and maybe server one is serving up HTTPS,

1:15that's secure web services.

1:16Maybe it's serving up DNS services,

1:18maybe it's serving up some streaming services, et cetera.

1:22And so its role is primarily going to be a server.

1:25And then we have other devices like clients,

1:27that their typical role isn't to provide services,

1:30but rather to consume services,

1:31like the computer you're probably sitting at

1:33or the mobile device you're on.

1:35In fact, if you're watching this content

1:36or listening to it right now,

1:38you are on some type of a client device

1:40that's consuming the content.

1:42So with a client, it would reach out, contact a server,

1:45and then consume those resources.

1:47And that would be an example of a network type

1:48that's client-to-server.

1:50So peer-to-peer is two devices

1:52that aren't dedicated in any specific role,

1:55for example, server or client, they can do both.

1:57And client-to-server is clearly defined roles.

2:00Now just be aware though,

2:01that if a client is connected to the server,

2:02it's also very likely

2:04that that server is acting as a client,

2:06to some other backend system,

2:07maybe a database or some other resource.

2:10However, in the singular instance,

2:12when the client is connected to the server,

2:14that's an example of a client-server relationship.

2:16Now another example of a network is a personal area network,

2:21and that refers to something that's like really close,

2:23within a few meters.

2:24An example would be Bluetooth.

2:26You've got your mobile device,

2:27you have a friend who has a mobile device

2:29and you wanna share something really close

2:31or really local between those two devices.

2:33That's an example of a personal area network.

2:37So some of the technologies there could be Bluetooth

2:39or infrared or NFC, near-field communications.

2:43And for many corporate networks,

2:44they're gonna follow a three-tier architecture

2:47for access and working with their networks.

2:50And it involves, the first tier is the access layer.

2:52Think of the access layer where devices plug in.

2:54For example, we have device A and B and C,

2:58and over here we have device D.

2:59So they're plugging into layer two switches

3:03at the access layer.

3:04And then those access layer devices

3:06have connectivity down

3:07to what's called a distribution layer device.

3:10And so normally we have fault tolerance,

3:11so this could be access one and access two

3:14as far as switches,

3:16and this could be distribution one and distribution two,

3:19and in this case there are multi-layer switches.

3:21We'll have a separate set of videos

3:22on switching and multi-layer switching

3:24and what that means.

3:25But from a network type and topology,

3:27this is a very common architecture.

3:29And then those will be connected down to some core devices.

3:32We'll call this core one and core two,

3:34and then we're gonna have

3:35some interconnectivity there as well.

3:37So for fault tolerance,

3:38we're gonna have connections that go like that

3:39and that and that and that.

3:40And for fault tolerance from this access layer switch

3:44to the distribution layer,

3:45we're gonna have connections like that and that,

3:47and this is an example of the three tier hierarchical model

3:50in enterprise networks.

3:53That's quite often followed by the way,

3:54regardless of the vendor that you're using.

3:57So in the data center it's slightly different,

3:59but this three tier architecture

4:00is used pretty consistently in corporate networks for access

4:04to end users on local area networks,

4:06if they're using wired connections.

4:07If they're using wireless,

4:09they'll have some connectivity either off

4:10of the distribution layer or the access layer

4:12with something called an access point.

4:15So an access point is a little wireless transmitter

4:17that allows your wireless clients to connect to the network.

4:20So you have your wireless clients and they connect in,

4:23they associate with the access point

4:25and they can join the network using good old Wi-Fi.

4:28Now, in a large environment, one access point,

4:30which acts the radio for Wi-Fi, it's not enough.

4:33So in a large environment,

4:35it's very likely to have multiple access points

4:38across the enterprise

4:39that present what it looks like to the user

4:41one giant wireless local area network.

4:45And for that wireless local area network,

4:47there'd be something called an SSID,

4:49think of that like the name for that network.

4:51And in enterprise networks,

4:52as I mentioned, they're expandable.

4:54So you could have like 20 or 30 or 40 access points

4:56so that if a user shows up anywhere in the building,

4:59they could associate with the closest access point,

5:01get access to the wireless local area network,

5:04and then authenticate and then start using the network.

5:06So this type of topology network type is fantastic

5:09in like a campus network

5:10or at the same geographic location.

5:13And an acronym for campus area network would be CAN.

5:17And that'd be an example of a network

5:19that sprawls across a campus.

5:21And that campus could be multiple buildings.

5:23And normally a campus area network

5:25is gonna have relatively high speed connectivity

5:27across the campus.

5:29And something that's slightly bigger

5:30than a campus area network

5:31would be an MAN, a metropolitan area network

5:37that covers like the geographic location for a city

5:40or a portion of that city or some other metropolitan area.

5:43So a campus area network or a metropolitan area network

5:46are all gonna be fairly high speed

5:48between all the nodes in either of those network types.

5:52And as we start going to wide area networks

5:54where we have more geography between our sites

5:57and we're using service providers,

5:58one of the reasons that speed often decreases

6:01for wide area networks is because it costs more.

6:03You have to pay a service provider

6:05for the bandwidth you're gonna use between, for example,

6:08site one and site two.

6:09So let's do this.

6:10Let me go ahead and clear off some of this.

6:11And I wanna share with you one of the really amazing topic

6:14that's starting to be used more and more and more

6:16regarding setting up

6:18and managing the overlays for wide area networks.

6:21So let me go ahead and redraw a headquarters up here

6:24and I'll put site two over here and site three over here.

6:29And we've got some beautiful connectivity here between those

6:31that could be just the raw internet

6:33or it could be bandwidth provided by lease lines,

6:35by service providers.

6:37But effectively it means we have some WAN connectivity

6:39that's available to us to glue together

6:42or connect together these sites.

6:44So let me go ahead and put the routers for site one,

6:47the router for site two,

6:49and the router for site three there.

6:51So let's imagine for our topology,

6:53our overlay topology

6:54that we wanna have a tunnel

6:56from headquarters down to site two,

6:58and we wanna have a tunnel from headquarters to site three.

7:02And we also wanna have a tunnel

7:04between site two and site three.

7:05So this would be the green represents like an IPSec

7:08or a GRE tunnel that we could also protect with IPSec

7:11as the overlay for logical connectivity between those sites.

7:16Now just setting up three tunnels like that, not a big deal,

7:19but if we start adding additional sites,

7:21so let me pop some more sites in here.

7:23We have site four and site five and site six and site seven.

7:29Then as we decide on what we want our topology to look like,

7:32our logical overlay topology,

7:34if we want full mesh, we've got connections,

7:36basically there's too many,

7:37there's tons and tons of connections

7:39from everybody to everybody.

7:40And if we're doing that manually,

7:42it becomes very hard to manage.

7:43So what we may wanna do is we may wanna do a partial mesh

7:46or hub and spoke to simplify it,

7:48Sophie did hub and spoke would be from the headquarter site

7:51here, here, here, here, here, and here.

7:54So that would logically be a star.

7:56And maybe we want some connectivity

7:58between site six and site three

8:00and maybe some connectivity between site two and site four

8:03so we could draw the tunnels there.

8:04But instead of having to do that overlay all by ourself,

8:07a super popular option is called SD

8:11as in software defined wide area network.

8:14And here's what it means.

8:15We have our infrastructure in place, our underlay,

8:17and by using an SD-WAN solution,

8:19you and I can sit at a computer,

8:21we can logically draw what we want for our topology,

8:24who should talk to who, what tunnel should go where,

8:26and then we use a controller that then implements

8:30that full configuration to all of the devices.

8:33So you and I talk to the interface for the SD-WAN solution,

8:37and then the controller talks with all the devices

8:39and reigns down that config and makes it happen.

8:42So that's what software defined refers to,

8:45whether it's software defined wide area networking

8:47like this example,

8:48or software defined networking in general

8:50where we're defining our networking

8:52for a campus infrastructure

8:53and specifying what we want that overlay network to be.

8:56And one of the big drawbacks of SD-wan

8:59is that it's not free.

9:00You have to pay a vendor,

9:01and there's lots of vendors that do it,

9:03including Cisco and Juniper and VMware and others

9:06that have SD-WAN solutions.

9:09So the actual cost for an SD-WAN solution

9:11may be kind of pricey,

9:12but once it's in place,

9:13the overhead as far as managing and working

9:15with that SD-WAN solution is gonna be more effective

9:18because it can roll out the configs

9:19and keep track of all the configs

9:21and implement changes with just a few clicks

9:24as opposed to having to go and manually do it all.

9:26However, I do wanna share with you

9:27one other option that is pretty darn cool

9:30for wide area networks

9:31that doesn't require having an SD-WAN solution,

9:34but it's still pretty slick,

9:35and that is using MGRE.

9:38Let me clear up my screen

9:39and let's talk about multi-point GRE.

9:41With multi-point GRE,

9:43we could configure our headquarters site

9:44with a logical multi-point GRE tunnel interface.

9:49And when you hear multi-point GRE,

9:51I'd like you to think of the idea of dynamic,

9:53meaning that multi-point GRE interface,

9:55the tunnel interface,

9:56can dynamically accept incoming requests,

9:59verify who they are,

10:01and then allow that tunnel to come up.

10:02And so if we had a multi-point GRE tunnel solution,

10:06we could then go ahead and set up the headquarter site

10:08and then as we bring up multiple sites,

10:10we simply point them at the headquarters,

10:12they connect and it dynamically builds the tunnel

10:15so we don't have to manually configure the headquarter site

10:18over and over again.

10:19So site two could connect

10:20and use the MGRE tunnel at the headquarter site.

10:23Site three could use the MGRE tunnel.

10:25And also with a solution like this,

10:27these two sites could dynamically discover each other

10:30and dynamically build a multi-point GRE tunnel

10:32to each other.

10:33And that way if we're onboarding additional sites

10:36and we're using the MGRE solution,

10:38as we bring up those multiple sites,

10:41we simply bring 'em up, put on a base config,

10:42they connect home,

10:44build the multi-point GRE tunnel,

10:47and we don't have to manually configure those.

10:49So one example of using multi-point GRE tunnels

10:52is a solution called dynamic multi-point VPNs,

10:56and that's a solution from Cisco Systems,

10:59but there are other vendors as well.

11:01So as far as network types go,

11:02that is a dynamic and fairly inexpensive way

11:06without buying a full SD-WAN solution

11:08for bringing up multiple sites

11:10with very little effort once the infrastructure is in place.

11:14Alright, and let's talk about one last network type.

11:17And that is something we're likely

11:18to find at a service provider.

11:20So let's imagine that this is our service provider network,

11:24I'll call it SP for short.

11:26And the service provider has lots of customers.

11:29They've got customer one over here,

11:31they've got customer two over here.

11:34And let's say this is customer one, site one,

11:37and customer one, site two,

11:41and we're on the left of customer two, site one,

11:43and customer two, site two.

11:47And that service provider could be providing connectivity

11:50between the customer sites or internet connectivity.

11:53But the actual network type I'd like to focus on

11:55for the service provider

11:56is what's going on inside of this service provider.

11:59So let's imagine the service provider has lots of routers.

12:02They've got a router there, there, there,

12:04they've got some here,

12:05and they've got connectivity to their customers

12:09and to each other,

12:10and they've got some fault tolerance

12:11in case the link goes down and it's just a party.

12:14All right, let's see if I've got it all.

12:16There it is.

12:17So let's say this is router one, router two, router three,

12:19router four, router five, router six.

12:21Now in a typical router, a layer three router,

12:25it's forwarding packets

12:26and it's looking at the layer three information

12:28from our discussion on the OSI reference model

12:31and the TC/IP protocol stack

12:32that's based on the actual IP address.

12:34So the router gets a packet,

12:36the router looks

12:37at the destination IP address in that packet

12:39and then makes a routing decision

12:40to forward onto the next router in the path.

12:42So one of the techniques the service providers

12:44can use in their network

12:46is they're going to enable a feature called MPLS,

12:49and that stands for Multi-Protocol Label Switching.

12:53Whew, it's a mouthful.

12:54It basically boils down to this.

12:55What they'll do is instead of forwarding packets

12:59inside their network based on IP address,

13:01they're gonna forward it based on labels.

13:03So each of the packets, in addition

13:06to its normal layer three IP header,

13:09it's also gonna get a little 2.5,

13:11I call it layer 2.5 label.

13:14That's part of the MPLS header.

13:15And as these routers in their enterprise

13:17receive those packets,

13:19they're gonna make forwarding decisions

13:20based on those labels as opposed to based on the IP address.

13:24And it makes sense because this customer two,

13:26maybe they have 100 networks over here on the left,

13:29and they've got,

13:30let's say 200 networks over here on the right.

13:32And effectively as the service provider

13:34forwards traffic from customer two, site one

13:37over to customer two, site two,

13:38it's gonna use some path.

13:39And so if they identify this is the path we wanna use,

13:43that traffic as it goes through

13:44the service provider network,

13:45they can forward that based on the labels,

13:47which gives the service provider

13:49a whole bunch of opportunities for some very creative tech.

13:52So one of the features that we can use

13:54if they're using MPLS is something called layer three VPNs.

13:58Think of a layer three VPN as a group of routes,

14:02for example, for customer two.

14:03And then using labels as the service provider network

14:06forwards that traffic.

14:08And when working with labels,

14:09when a router receives an MPLS labeled packet,

14:12it's gonna go ahead and swap out that label

14:14for the next router on the path.

14:16So the process is called label switching

14:18because we're forwarding the traffic based on labels,

14:20but every router on the path is also doing label swapping

14:23by taking off the label that came in on it, removing it,

14:26and then placing the new label

14:28before forwarding it to the next device in the path.

14:30Another way of saying that is swapping the tag.

14:33So the concept of a label

14:35is also sometimes referred to as a tag in MPLS.

14:38And again, a typical corporate network

14:40won't be dealing with MPLS and labels.

14:42However, if we're working for a large service provider,

14:45it's very likely that they are enabling MPLS

14:47on their core network

14:48and then doing that label switching to forward traffic

14:51inside of their network.

14:52So if traffic is being sent

14:54from a computer over here at customer two, site one,

14:56and it's being sent over

14:57inside the service provider network, we have those tags,

15:00those labels that are being used for the forwarding,

15:03but before the actual traffic is sent out to the far side,

15:05any tags or labels that the service provider

15:08was using are stripped off.

15:10And so the endpoints at customer two, site one

15:13and customer two, site two,

15:14they have no clue

15:15that there was some label swapping

15:17happening inside of the provider network.

15:19All they know is that they got the packet,

15:21it looks like a normal IP packet, they processed it.

15:24And if that called for some response back to the other side,

15:26the response made it back through the other direction.

15:29Now, one of the cool things about logical networks

15:31and overlays is that we can be very creative

15:33about how we design our networks.

15:35We have the underlay, the real network, I call it,

15:37and then the overlay, which is the virtual network.

15:40But in addition to the overlay,

15:42we also have options in the virtual world

15:44for doing virtualized networking altogether.

15:47So in the next video,

15:48I'd like to chat with you

15:49about hypervisor's virtualized networking functionality

15:52and give you some examples of doing exactly that.

0:00<v ->In this video, I'd like to chat</v>

0:01with you about virtualization in the world of networking.

0:04So we're gonna take a quick peek at hypervisors

0:07and virtual switches, and then we'll extend that to

0:09network function virtualization.

0:12So let's lay some groundwork for virtualization

0:14and then we'll add on top of it with networking.

0:16If we're looking at a virtualized environment behind it all,

0:20we have some type of a hypervisor.

0:22And a hypervisor is an environment that allows

0:24for the creation of a virtual machine.

0:27And let's take a look at a specific example.

0:29This is a host, a computer,

0:32and this computer has things like CPU, it's got memory,

0:36it's got some storage.

0:38Now that storage may be local on that,

0:41or it may be storage that's reachable over a network.

0:43And on this hardware, if we run a type of software

0:47called a hypervisor, that hypervisor then allows

0:50for the creation of individual VMs.

0:54Now, when we talk about VMs, we have to whisper

0:56'cause they think they're running just

0:57on dedicated hardware.

0:58They don't know that they're running as a virtual machine.

1:01For example, this could be Windows, this could be Linux,

1:04this could be some other virtual machine

1:05running a guest operating system.

1:06And all those virtual machines are then running courtesy

1:10of this hypervisor, which is providing

1:12a logical home for them.

1:14Now, when we have a hypervisor that runs directly

1:17on hardware, that's referred to as a Type 1 hypervisor.

1:21If we had a situation where we had a computer,

1:23so here's our hardware,

1:25and it's running a operating system,

1:27let's say it's running Windows,

1:29and then we're running an application

1:31like VMware workstation, and that's a hypervisor also,

1:34but because it's running as an application

1:36and then we're running VMs inside of that,

1:38that would be an example of a Type 2 hypervisor.

1:42So a type one or a bare metal hypervisor runs directly on

1:45the hardware and provides an environment for VMs

1:48to go ahead and run in.

1:49And let me show you an example of exactly that.

1:52Lemme make the font a little bit bigger here.

1:55So this is an ESXi host.

1:57It's a hypervisor from VMware that's running on a server.

2:01So this server is, if you look at the hardware here,

2:04it's a Dell PowerEdge R630.

2:06It's got a whole bunch of CPUs and a whole bunch of memory.

2:08But this hypervisor software that we're looking

2:11at the interface for,

2:12that's providing an environment

2:14where we can then create multiple virtual machines.

2:17So I've got a virtual machine here running

2:19my network emulation.

2:20I've got a virtual machine right here,

2:22which is a Windows 2019 server.

2:24We right click on it and it's Launch remote console,

2:27that gives me a very quick and easy way to access

2:30that virtual machine.

2:32Lemme bring that into screen here.

2:33So here is that virtual machine.

2:35And if we wanna go ahead and log in,

2:37I'll put in my credentials,

2:38and boom, this is the the graphical user interface

2:42for that virtual machine, this Windows 2019 server.

2:45So literally we were just looking at ESXi-3

2:48and then we had our Windows 2019 server that was running

2:51as one of the VMs on that hypervisor.

2:54And in a VMware environment,

2:56the hypervisor name is called ESXi.

2:58But the same concepts regarding hypervisors apply,

3:01whether using HyperV or Google services in the cloud

3:05or Amazon Web Services or Microsoft Azure,

3:07those are all examples of platforms

3:10that provide hypervisor services so that we can spin up

3:13and run and use virtual machines.

3:15So let's pretend this is our Windows 2019 machine,

3:18if it wants to work with the hard disk, meaning

3:20write some data or read some data, that request goes

3:22to the hypervisor and then the hypervisor actually

3:25goes to the actual storage.

3:26So the hypervisor is the middleman between the resources

3:29that the virtual machine wants and needs

3:32and the actual resources themselves.

3:34That would include CPU access, memory access,

3:37and disc access.

3:38Also it applied to network access.

3:41All that access is being provided by the hypervisor

3:43for the benefit of the VM.

3:45So one of the questions that might come up is, "Okay,

3:47what if I need these VMs to communicate with each other?"

3:50Maybe I have the Windows 2019 server here

3:52and that's VM number 1,

3:55and I've got VM number 2 over here

3:57and I want them to talk to each other.

3:59How do we do that?

4:00And the answer is we do virtual networking.

4:02So we can create a virtual switch.

4:04We'll call that V switch that lives in the brain

4:07of this hypervisor running on this hardware,

4:10and then we can logically plug in those devices

4:12to that V switch.

4:13So we have a virtual network interface card on this

4:16virtual machine, and we have a virtual network interface

4:19card, a VNIC, on this virtual machine,

4:21and we plug them into the same switch

4:23and then they can communicate with each other.

4:25So if these two VMs on the same hypervisor

4:27and they're connected to the same switch,

4:28need to communicate, they can.

4:30Now if we want like this VM here to be able to communicate

4:33with this VM over here, then we start

4:35to network all the hypervisors together.

4:37So we have a physical network out here,

4:39and then logically we can forward the traffic

4:42through the physical interfaces to go from ESXi-3,

4:45for example, over to ESXi-2,

4:47which could then forward it to a virtual machine over here.

4:50But the concept from the VMs perspective is

4:52that all they know is that they're connected to a switch.

4:54In this case, it would be a virtualized switch,

4:56and then it's the hypervisor's responsibility

4:58to make sure the forwarding of the traffic happens

5:01appropriately so they can get to their destination.

5:03In fact, lemme show you an example

5:05of a virtual switch on ESXi-3

5:07that allows the virtual machines to talk to each other.

5:10So this represents my 2019 Windows server

5:12and this represents another virtual machine.

5:14And can they communicate together?

5:16It depends on the networking.

5:17So let's go take a look at the virtual switch.

5:20So over here on the left on the hypervisor,

5:22I'm gonna go down to networking

5:24and I've got this virtual switch right here

5:26I'm gonna click on it and this is showing,

5:28lemme go ahead and minimize that.

5:29This is showing a visual representation

5:31of the virtual switch.

5:32So here regarding virtual switch number zero,

5:35which represents this virtual switch right here,

5:38I've got my Eve-NG-Pro virtual machine

5:41connected to that switch.

5:43Again, this is V switch zero.

5:45I've also got the Windows 2019 server

5:47connected to the same V switch.

5:49So logically they should be able to communicate

5:51with each other logically via this V switch.

5:54And they're both up and running so let's test it

5:56with a quick ping.

5:57So the IP address on this virtual machine

5:59for the Eve-NG-Pro is

6:00192.168.1.115

6:03And on our server, let's go back

6:05to our server and take a look.

6:06So this is the graphical user interface for our server.

6:08Let's go to a command line and do an IP config.

6:12So we should be able to ping via the virtual switch from

6:15this VM over to the other VM if we ping out to its address,

6:18the Eve-NG machine is at 192.168.1.

6:23and it is 115, press enter.

6:25And sure enough, we're getting those replies

6:28from that other virtual machine.

6:29And all that traffic is logically going through

6:31that virtual switch zero on the ESXi hypervisor.

6:35Now, in addition to basic switching that we can have

6:38between VMs on the same hypervisor, for example,

6:41between this Windows 2019 server

6:43and this VM right there,

6:45we could also virtualize other network devices,

6:48not just switches.

6:49So if we go back to this topology,

6:51could we virtualize this firewall

6:53and have it be a virtual appliance, a VM

6:55that's doing the same exact functions

6:56and not have a physical device?

6:57The answer is yes.

6:59Could we use the same technique for a router

7:01and have a virtual router as a VM instead

7:04of having dedicated hardware?

7:05The answer is yes.

7:06Now, as far as physical devices,

7:08like physical computers that are plugging into a switch,

7:11they really do need to plug into something physical.

7:13So it has to be physical at the edges,

7:15but all the infrastructure,

7:16it could be virtualized if we wanted it to be.

7:18Now, one of the negatives of doing a virtualized device

7:22is this: Power.

7:25So if we purchase, for example,

7:26a dedicated firewall from a vendor that we trust,

7:29they've got hardware

7:31and application specific integrated circuits

7:34and a lot of CPU dedicated for that function.

7:37And so it might be able to forward hundreds

7:39of megabits per second

7:40and do all the analysis and all the hard work.

7:42But if we virtualize this,

7:44so let's say we have a hypervisor here,

7:46and that hypervisor has, maybe it has a ton of CPU

7:49and a ton of RAM and everything else that it needs,

7:52but once we start loading up, maybe that's our firewall

7:54and maybe we're running a proxy server

7:56as another virtual machine

7:58and we're running another virtual machine.

8:00All those VMs have to compete

8:02for the resources on this hypervisor.

8:04So the one negative of virtualizing a lot

8:06of our infrastructure is the raw CPU

8:09and dedicated hardware that otherwise might be a bottleneck

8:14if we're using virtual machines

8:15that are providing those functions.

8:17So going back to the hypervisor here,

8:18if we take a look at our virtual machines, I only have two

8:21that are running at the moment, but I'd like you

8:23to take a look at some of these virtual machines.

8:25Here is this guy right here,

8:26PA-FW1 is a virtualized Palo Alto Next Generation firewall.

8:31When I was working with the Cisco Firepower firewall,

8:35I also had it virtualized.

8:37And so if the virtualized flavor meets the needs as far

8:40as CPU and throughput, there's not a lot of negatives

8:43in virtualizing that gear as opposed to having

8:45dedicated physical appliances for those functions.

8:48And as far as what could we virtualize?

8:50We could virtualize firewalls,

8:52we could virtualize layer 2 switches,

8:54we could virtualize layer 3 routing and routers.

8:57So effectively, anything that we can run on

8:59a dedicated appliance is very likely we have the ability

9:02to virtualize it using some flavor

9:05of hypervisor to provide that environment.

9:07Another trend that has been growing

9:09and will continue to grow is instead of having systems

9:13with dedicated attached storage, like a dedicated hard drive

9:16that's physically on that device, instead,

9:19why don't we use network attached storage

9:21or network storage that's reachable over a network?

9:24So in the next video, we're gonna take a closer look at some

9:26of our options and some of the technologies involved

9:28with dedicated networks for the purpose of storage.

9:32I'll see you in that next video in just a bit.

0:00<v ->In the old days, we used to have storage</v>

0:02that was only local on that local computer.

0:04So you had 10 computers,

0:05they each had their own little hard drive

0:07and you could use those hard drives.

0:08And if they were full, you have to either delete stuff

0:11or figure another solution out.

0:12In the world we live in today,

0:14a lot of our storage options are available over a network.

0:18And let's use this as the backdrop for this discussion.

0:21This represents three hosts that are running a hypervisor,

0:26and then on top of that we have virtualized machines.

0:28And now those virtual machines need disc access,

0:31so they make the request to the hypervisor,

0:33and then the hypervisor actually facilitates

0:35those read and write requests to the storage.

0:38Now, there's several options in general

0:40that we could use for storage regarding these hypervisors.

0:42One would be local.

0:44That means we have a local disc drive

0:46that's attached to that physical hardware,

0:48and that's the storage it uses.

0:50So that's certainly one option.

0:51And the local is represented down here in this diagram

0:55right here.

0:55So it's directly attached storage,

0:57and the host can go ahead and use it.

1:00Now that doesn't scale very well

1:02because we're limited to just the physical resources

1:04in the directly attached storage there.

1:06For the benefit of the VMs, what makes better sense

1:09is use some type of network access to storage.

1:11So one of those options

1:12is use a dedicated proprietary network

1:15with a technology called Fibre Channel,

1:17spelled F-I-B-R-E for Fibre Channel.

1:19And that's a dedicated network that these hosts connect to

1:23that they can go ahead and use

1:24to reach out and access storage.

1:26However, dedicated Fibre Channel networks are expensive.

1:28So another option is to use

1:30a more traditional ethernet network,

1:32a high-speed dedicated ethernet network

1:34with a technology called Fibre Channel over Ethernet.

1:37So it's still using the technologies of Fibre Channel,

1:41but it can run over a more traditional ethernet network,

1:43which makes it more affordable.

1:45So then our storage would be attached

1:47to that Fibre Channel over Ethernet network

1:49or reachable over that network,

1:51and then we could have multiple hosts

1:53who are sharing that storage.

1:54Now, typically when we have multiple devices

1:57that are working with a SAN,

1:59which is an acronym for storage area network,

2:02that's gonna be block-based,

2:04meaning individual blocks or spaces of data

2:06are being read from and written to.

2:08So if you have systems that can communicate

2:10with a storage area network with block-based storage,

2:13they can share those resources very efficiently.

2:15And most of the time

2:16when we're working with storage area networking,

2:18we're using this technology right here, iSCSI.

2:21And one of the cool things about iSCSI,

2:22it uses the same commands

2:24as if we were working with local storage,

2:25but it can use iSCSI over an IP-based network.

2:29And with iSCSI, it doesn't have to be

2:31a Fibre Channel over Ethernet.

2:32We can have an iSCSI adapter on this host

2:35that is able to communicate over ethernet

2:38to a storage device that also speaks iSCSI.

2:41And that way it can send

2:42the read and write request via iSCSI commands

2:45over a traditional ethernet network.

2:47And normally we're gonna have

2:48dedicated high speed ethernet networks

2:51just for that storage.

2:52And let me show an example of that.

2:53This is the ESXi host.

2:55If we go over here to storage, it has some local stuff,

2:58but it also this Synology storage area network appliance

3:02that it's connected to.

3:04So this ESXi host that we're currently on, which is ESXi-3,

3:07it can write to and read from this storage appliance

3:11using iSCSI.

3:13And as far as how it pulls it off,

3:14it's got a virtualized iSCSI adapter built into it.

3:18So here on this host,

3:18if you click on storage and go to adapters,

3:21here we have this virtualized iSCSI software adapter,

3:24which at the end of the day

3:26is communicating via an ethernet network

3:29to reach the SAN appliance.

3:30And that SAN appliance is a little Synology device

3:34that has some fault tolerance built into it as well.

3:36And here's the graphical user interface

3:38for that little SAN device.

3:39So when we're using a SAN or see the term SAN,

3:43it's very common that we're using iSCSI,

3:45that it's block-based storage.

3:47And on that appliance that's providing that storage,

3:49it's very likely, in production environments,

3:51that it just has tons and tons of storage available

3:54that can be shared across this storage area network.

3:57Now, not all network based storage is gonna be block-based

4:00where individual computers and systems can share it

4:03and write across it in individual blocks.

4:06Some network attached storage uses file-based,

4:09and they have a different term for that

4:10and it's called NAS for network attached storage.

4:14I know the acronym is like flipped, (laughs)

4:16but network attached storage is traditionally file-based.

4:21And here's what that means.

4:22Instead of having individual blocks

4:23that are written to and read from by multiple devices,

4:26with file-based, the devices who are using it,

4:28let's say this is our computer right here,

4:30and we are attached to a network

4:32and that network has a NAS here with storage.

4:37As we interact with that network attached storage device,

4:39we are reading and writing individual files.

4:42So if we're working on a Word document here and we save it,

4:44it's gonna be saved as a Word document.

4:46If we pull up a JPEG, it'll pull up the JPEG.

4:48So it's file by file by file with network attached storage

4:53versus a storage area network which is using iSCSI

4:56and is traditionally block-based.

4:58So my intention for us in this video

5:00is to reinforce the idea

5:02that not all storage has to be local.

5:03We may have dedicated networks

5:05and types of networks just for storage,

5:07which include iSCSI-based networks

5:09for storage area networking, and also file-based storages

5:12with more typical network attached storage devices.

5:15So thanks for joining me in this video,

5:16and I'll see you in the next set.