Tip:
Highlight text to annotate it
X
00:00:00 - In the previous, Nugget, we discussed the block protocol
00:00:04 - of SCSI and how great it's going to be for the commands
00:00:09 - associated with reading and writing to a storage facility.
00:00:13 - Now, we know that one of the options for carrying these
00:00:17 - SCSI commands was Fibre Channel. 00:00:19 - And I promised you that we would
go into much greater 00:00:22 - depth on this important storage
area networking 00:00:25 - technology.
00:00:26 - So guess what? 00:00:27 - That time has arrived.
00:00:27 - Let's do it now. 00:00:29 - In this Nugget entitled Fibre Channel
00:00:31 - and More Fiber Channel. 00:00:33 - Now, as you might guess, this Nugget
is pretty darn 00:00:36 - important when it comes to SANs.
00:00:38 - And we've got lots to do as a result. 00:00:40 - I'll take you on a little history
lesson, and we'll look 00:00:42 - at the topologies that have evolved
through the years when 00:00:45 - it comes to Fibre Channel.
00:00:47 - Then Fibre Channel is unique compared to local area
00:00:50 - networking technologies. 00:00:51 - We'll talk about all the different
port types we have, 00:00:54 - many of which were invented by
Cisco. 00:00:57 - And we'll talk about the wonderful
world of addressing 00:01:00 - that we have in a Fibre Channel
environment. 00:01:02 - Finally, I'll take you behind the
scenes, and we'll see 00:01:05 - exactly how a Fibre Channel storage
area network is going 00:01:09 - to function.
00:01:10 - Do you remember the language of SCSI?
00:01:13 - We have that concept of an initiator and that
00:01:16 - concept of a target. 00:01:17 - Well, of course, we still have
that in the Fibre Channel 00:01:20 - environment, because remember,
we are sending SCSI commands 00:01:23 - here in our storage area network.
00:01:25 - Now, the very first topology that was invented for Fibre
00:01:30 - Channel storage environments was the
00:01:32 - point to point topology. 00:01:35 - The initiator has a direct connection
to a particular 00:01:40 - target device in that infrastructure.
00:01:43 - The initiator can obviously send commands directly to that
00:01:48 - particular target that's connected in that point to
00:01:52 - point connection. 00:01:54 - Now, obviously the storage here
functions 00:01:57 - as a dedicated server.
00:01:59 - And the initiator functions as a client.
00:02:02 - This does great for performance. 00:02:04 - It does great for security, but
obviously one of the 00:02:08 - issues with this point to point
topology for Fibre 00:02:11 - Channel networks is, it doesn't
give us scalability. 00:02:16 - This direct connectivity is a lot
like that concept of 00:02:20 - directly-attached storage that
we've actually wanted to move 00:02:25 - away from for scalability concerns.
00:02:29 - So we bring this up as a bit of historical reference, this
00:02:34 - point to point topology that is capable in a Fibre Channel
00:02:39 - environment. 00:02:41 - So in an attempt to address the
scalability concerns 00:02:44 - presented by point to point topologies,
the inventors of 00:02:48 - the Fibre Channel networks decided
they would do a Fibre 00:02:52 - Channel arbitrated loop, yeah,
the FC-AL. 00:02:57 - And this Fibre Channel arbitrated
loop would be 00:02:59 - obviously much more scalable than
that 00:03:01 - point to point topology.
00:03:03 - But guess what? 00:03:06 - In actual practice, even though
theoretically 120 00:03:11 - devices can be connected in the
arbitrated Fibre Channel 00:03:16 - loop, in actual practice, you can
only get about 12, due to 00:03:21 - performance concerns in this type
of infrastructure. 00:03:25 - You see, because there's only one
data path in this 00:03:29 - looped-type topology, only one
pair of devices can 00:03:32 - communicate at any one given time.
00:03:36 - So we have a loop-type situation here, where
00:03:38 - everybody's working with shared available bandwidth.
00:03:42 - There's latency concerns in this topology, because these
00:03:47 - devices have to negotiate for control of the loop before
00:03:50 - they can send traffic. 00:03:52 - Is this sounding familiar to you?
00:03:55 - Does this bring back the limitations and problems that
00:03:58 - we had in early token ring environments? 00:04:02 - It certainly should, because it
is no coincidence. 00:04:07 - So another issue with the loop,
just like we would find 00:04:11 - in a token ring-type environment,
would be how 00:04:15 - susceptible this environment was
to a device failure. 00:04:19 - If we had one of our hosts, one
of our initiators in the 00:04:23 - loop fail, we could have many,
many different devices 00:04:27 - impacted by that failure.
00:04:29 - So kind of a second-generation topology, if you will, the
00:04:33 - Fibre Channel arbitrated loop. 00:04:36 - Now, I bet you know where we're
going into the third 00:04:39 - generation of topology that I'm
going to 00:04:42 - show you right now.
00:04:44 - Yes, thank goodness, the concept of the Fibre Channel
00:04:47 - switch was invented. 00:04:50 - In fact, when they did this, they
really started talking 00:04:53 - about the fabric.
00:04:55 - Fabric equates to the concept of a storage area network when
00:04:59 - we're talking about a Fibre Channel environment.
00:05:01 - With the Fibre Channel switch, we can have the more efficient
00:05:05 - connectivity of Fibre Channel devices that
00:05:08 - we're looking for. 00:05:09 - And think about it.
00:05:10 - As you need to scale, you can go ahead and plug your Fibre
00:05:14 - Channel switch into another Fibre Channel switch, just as
00:05:18 - we would do in a local area network type of environment.
00:05:23 - So this third generation switched fabric topology is
00:05:28 - what we're so excited about today, and what Cisco's
00:05:31 - certainly built upon with great inventions for Fibre
00:05:35 - Channel with their MDS line of storage
00:05:39 - area network equipment. 00:05:42 - Now, one of the things that's really
interesting about Fibre 00:05:44 - Channel is, our ports have different
roles. 00:05:48 - They have different functionality.
00:05:50 - And they get these port names that we need to get
00:05:55 - comfortable with. 00:05:56 - Let's start with the standard ports
that we have in a Fibre 00:06:01 - Channel environment before any
fancy inventions from an 00:06:04 - organization like Cisco Systems.
00:06:08 - The first port that I want to talk to you
00:06:09 - about is the N-Port. 00:06:12 - And here we see an N-Port on a
host that connects to the 00:06:15 - Fibre Channel environment.
00:06:17 - An N-Port is a node port. 00:06:21 - This would be a port connecting
to a particular 00:06:24 - host or a disk, a node in the infrastructure.
00:06:27 - This is pretty easy-- 00:06:30 - no problem at all.
00:06:32 - Now, what does that connect to? 00:06:35 - Well, here you can see we have
this Fibre Channel device. 00:06:38 - And it can connect to an F-Port.
00:06:41 - We refer to F-Ports as our fabric port.
00:06:45 - This is the port that connects to our peripheral device, like
00:06:49 - a host or a disk that's operating as that N-Port.
00:06:52 - Notice that an F-Port can only be attached
00:06:56 - to exactly one N-Port. 00:07:00 - Now, notice we have a Fibre Channel
switch here connecting 00:07:03 - to another Fibre Channel switch.
00:07:05 - These connections are made with E-Ports.
00:07:09 - What's an E-Port? 00:07:10 - It's an expansion port.
00:07:12 - An E-Port allows us to create an interswitch, an interswitch
00:07:20 - link, or an ISL, between two of our switches.
00:07:23 - These E-Ports are going to carry frames between these
00:07:27 - switches for configuration and fabric management.
00:07:31 - And they're also serving as a conduit between switches for
00:07:35 - frames that are going to be destined for remote N-Ports.
00:07:42 - Now, look over to the right in this topology.
00:07:45 - We have the old-school Fibre Channel 00:07:48 - arbitrated loop situation.
00:07:51 - Well, in this particular case, connecting to this old
00:07:56 - antiquated mess, we have an FL-Port. 00:08:00 - So this port right here is an FL-Port.
00:08:03 - And what the job of that port is, obviously, is to connect
00:08:09 - to that fabric loop in the topology. 00:08:13 - Sure enough, the ports that connect
to that infrastructure 00:08:18 - get a special designation.
00:08:20 - And they are the NL-Ports. 00:08:23 - So here we have NL-Ports.
00:08:26 - These are node ports, but there are node ports in that
00:08:30 - special arbitrated loop type of topology that we can have
00:08:35 - with Fibre Channel. 00:08:37 - So take a good look here.
00:08:38 - These are the standard ports that we have in a Fibre
00:08:43 - Channel architecture-- 00:08:44 - our N-Ports, our fabric ports,
our expansion ports, our 00:08:49 - FL-Ports for connecting to an arbitrated
loop type of 00:08:53 - environment, and then the node
ports in there that are called 00:08:57 - NL-Ports in that special topology.
00:09:02 - But the engineers at Cisco Systems weren't just going to
00:09:05 - sit around and not make glorious inventions for the
00:09:09 - Fibre Channel infrastructure. 00:09:10 - So here we have a couple of Cisco
MDS devices, and they 00:09:14 - can be connected by what are called
TE ports, the trucking 00:09:23 - expansion port.
00:09:24 - This Cisco invention allows MDS switches to expand on the
00:09:30 - functionality of those old E-Ports or expansion ports
00:09:34 - that we had in the classic Fibre Channel infrastructure.
00:09:38 - Thanks to the TE, or trucking expansion ports, we can take
00:09:42 - advantage of virtual storage area networking.
00:09:46 - Can you believe Cisco did this? 00:09:48 - Capitalizing on the popularity
of VLANs, Cisco invented the 00:09:53 - concept of a VSAN.
00:09:55 - These TE ports allow us to do quality of service over this
00:09:59 - particular trunk and special Fibre 00:10:02 - Channel trace route features.
00:10:04 - So when we have a TE port in our infrastructure, we're
00:10:09 - going to transmit frames in what is called an EISL format.
00:10:16 - This is going to contain the virtual storage area network
00:10:20 - information. 00:10:22 - Interconnected switches use this
VSAN identifier in order 00:10:27 - to multiplex traffic from one or
more virtual storage area 00:10:32 - networks across this same physical
link. 00:10:36 - Again, this has parallels to our
local area network 00:10:40 - environment.
00:10:41 - That's why it makes it pretty darn easy for us to master.
00:10:44 - Now, another neat Cisco invention is an NP port or a
00:10:49 - node proxy port. 00:10:51 - Remember we said we had our F-Ports
on these devices, and 00:10:54 - they would connect down a node
port for a particular host? 00:10:58 - Well, in Cisco environments, we
can have an NP-Port, a node 00:11:03 - proxy port.
00:11:05 - And what this device is doing is, it's allowing other end
00:11:10 - ports to connect through it, for multiple physical end port
00:11:13 - connections through one F-Port. 00:11:17 - So this is a great enhancement
for scalability in our 00:11:23 - Cisco-based fabric infrastructure.
00:11:27 - Now, what if we have a host device that wants to send in
00:11:29 - these tagged virtual storage area network frames?
00:11:32 - Well, some special port terminology for this
00:11:35 - environment-- 00:11:36 - we would have a trunking fabric
port in this case, a TF 00:11:41 - port, and then a trunking node
port. 00:11:45 - So a TN port would be on the device
that has the ability to 00:11:51 - send those tagged frames for virtual
storage area 00:11:55 - networking.
00:11:57 - Now, in our LAN environments, we often want to connect a
00:12:01 - protocol analyzer to the environment. 00:12:04 - And we use the switched port analyzer
or 00:12:07 - span feature for this.
00:12:09 - Well, this has a corollary in our Fibre Channel
00:12:12 - environments, thanks to Cisco. 00:12:14 - And we can get into what's called
an SD port. 00:12:18 - The SD port is a span destination
port in our Fibre 00:12:23 - Channel switched infrastructure.
00:12:27 - Another cool invention is an FX port.
00:12:31 - Now, when you think FX, don't think special effects, by the
00:12:34 - way, but think of the X as a variable. 00:12:37 - Sure enough, an FX port can operate
either as an F-Port or 00:12:43 - an FL port, depending on what connects
to 00:12:48 - that particular port.
00:12:49 - So if you're going to connect via the switched fabric, it'll
00:12:52 - be an F-Port. 00:12:53 - If you're going to connect an arbitrated
loop type 00:12:56 - situation, it will automatically
convert itself 00:12:59 - to the FL port.
00:13:02 - Now, if you're connecting another 00:13:04 - geographically-dispersed device
from another storage 00:13:09 - area network that's pretty far
away, some vendors will 00:13:13 - implement a B Port for this particular
functionality. 00:13:18 - This is called a bridge port.
00:13:20 - And it's described in the Fibre Channel standards, and
00:13:24 - Cisco will support these particular B Ports.
00:13:28 - Finally, you can have ports that are in what are
00:13:30 - called an auto mode. 00:13:33 - So auto mode ports can operate
as an F-Port. 00:13:37 - They can operate as an FL port.
00:13:39 - They can operate as an E-Port. 00:13:42 - They can operate as a TE port.
00:13:44 - They can operate as a TF port. 00:13:47 - You get the idea.
00:13:48 - So these ports are extremely flexible and can automatically
00:13:52 - be initialized for a particular role when the
00:13:56 - interface comes up and determines 00:13:59 - what exactly is attached.
00:14:02 - By the way, I know we have covered a lot of terminology
00:14:07 - and concepts already, and we're very early this
00:14:10 - particular Nugget. 00:14:11 - This is definitely an area where
we are learning a new 00:14:15 - vocabulary, aren't we?
00:14:17 - Just like you would do when learning a new vocabulary for
00:14:20 - a new language, you'll probably want to develop some
00:14:25 - aids to assist you. 00:14:26 - This would be a great area where
flash cards, either 00:14:30 - traditional index cards or electronic
flash cards, would 00:14:34 - come in very, very handy.
00:14:37 - In our local area network environment, you're very
00:14:39 - familiar, I'm sure with, the concept of a MAC address.
00:14:43 - In fact, we looked at these in depth in the first of our two
00:14:47 - courses here for CCNA data center at cbtnuggets.com.
00:14:51 - Now, that's great. 00:14:53 - The MAC address is that unique
identifier for Layer 2 in our 00:14:59 - local area network environment.
00:15:00 - And sure enough, we have a similar methodology over here
00:15:05 - in the Fiber Channel world. 00:15:06 - They're called World Wide Names.
00:15:10 - They are 64-bit, or more commonly, 128 bits in length.
00:15:17 - And they are indeed presented in hexadecimal.
00:15:21 - Now, there's two types of World Wide Names that we need
00:15:25 - to be familiar with. 00:15:27 - There's the Node World Wide Name.
00:15:30 - This is one and only one address that uniquely
00:15:34 - identifies that particular node in the Fibre Channel
00:15:37 - infrastructure. 00:15:39 - Now, we know a lot of devices will
have 00:15:41 - multiple ports, of course.
00:15:44 - And sure enough, each and every port will get its own
00:15:48 - unique Port World Wide Name. 00:15:52 - The only exception to this rule
would be a particular 00:15:56 - device that only has one port.
00:15:58 - If you have a device that only has one port, you'll often see
00:16:01 - the Node World Wide Name and the Port World
00:16:04 - Wide Name as identical. 00:16:07 - But that's uncommon, obviously,
in our environment. 00:16:10 - We have Fibre Channel switches,
which will possess a 00:16:13 - unique node worldwide name that
uniquely identifies that 00:16:17 - device in the infrastructure, and
then multiple Port World 00:16:21 - Wide Names that uniquely identify
each port in that 00:16:26 - particular infrastructure.
00:16:28 - Now, just like MAC addresses, these World Wide Names are
00:16:32 - indeed hard-coded into our Fibre Channel devices.
00:16:35 - But there's another type of Fibre Channel addressing I
00:16:38 - want you to be aware of, and that's called the Fibre
00:16:41 - Channel IDs. 00:16:43 - You'll often see this as FCIDs.
00:16:46 - These are dynamically-acquired addresses that are routable in
00:16:51 - our switch fabric. 00:16:53 - Remember when we were talking about
the Fibre Channel 00:16:56 - arbitrated loop and I told you
that had a 00:16:59 - limitation of 127 devices?
00:17:02 - That's because of the addressing scheme that is used
00:17:05 - for the dynamically-chosen Fibre Channel IDs in that type
00:17:10 - of environment. 00:17:11 - So let's focus on our Fibre Channel
IDs in the topology 00:17:16 - that we're really passionate about,
the switched fabric 00:17:19 - infrastructure.
00:17:20 - We have a 24-bit Fibre Channel ID address that is utilized in
00:17:26 - this environment. 00:17:27 - And it consists of three 8-bit
elements. 00:17:31 - The first bit is our domain ID,
absolutely 00:17:36 - the first 8-bit element.
00:17:38 - And this is used to define the switch.
00:17:42 - Each switch receives a unique domain ID.
00:17:46 - Then we have our area ID. 00:17:50 - And this area ID, sure enough,
is used to identify a group of 00:17:54 - ports within a domain.
00:17:57 - Areas can be used to group ports within a switch.
00:18:00 - Areas can also be used to uniquely identify
00:18:03 - fabric-attached arbitrated loops. 00:18:07 - Each fabric-attached loop receives
a unique area ID. 00:18:11 - And by the way, what appears to
SRP there is my penmanship 00:18:15 - work for GRP, for group.
00:18:18 - So the area ID helps us identify a
00:18:21 - group group of systems. 00:18:23 - And then finally, our third part
is the port portion of 00:18:27 - this Fibre Channel ID.
00:18:30 - Now, remember we said that each switch needs a unique
00:18:34 - domain ID and that it's an 8-bit field.
00:18:37 - But in all actuality, there's only 239 domain IDs available
00:18:44 - to the fabric. 00:18:45 - This is because domain IDs 01 through
EF 00:18:52 - are actually available.
00:18:54 - Domains 00 and F0 through FF are reserved for use by
00:19:03 - internal switch services. 00:19:06 - Now, keep in mind, even though,
through the 00:19:09 - mathematics of the situation here,
there could be 239 00:19:14 - switches in the fabric, based on
the Fibre 00:19:16 - Channel IDing system.
00:19:18 - In practice, the number of switches that you would get in
00:19:21 - your fabric typically is a lot smaller because of particular
00:19:25 - storage vendor built-in restrictions. 00:19:29 - Now, let me share a little bit
about the Fibre Channel 00:19:31 - standards with you in case you're
reading some of the 00:19:34 - literature and you come across
these terms utilized. 00:19:37 - When they designed Fibre Channel,
they did it in a 00:19:40 - five-layered architecture.
00:19:42 - There's FC-0, which defines the physical interface stuff,
00:19:47 - like cabling, connectors, et cetera. 00:19:49 - There's FC-1, which defines the
line coating of signals. 00:19:54 - There's FC-2, which indicates framing
and 00:19:58 - flow control matters.
00:20:00 - There's FC-3, for generic services that are required for
00:20:05 - fabric management. 00:20:07 - More on that in a moment.
00:20:08 - And then FC-4, which is where we do our upper
00:20:12 - layer protocol mapping. 00:20:14 - Now, a big area here is the generic
services that are 00:20:18 - going to need to be provided that
would be really required 00:20:22 - for fabric management.
00:20:24 - The specifications outlined a name service.
00:20:28 - They outlined a login service, an address manager, an alias
00:20:34 - server, a fabric controller, as it is known, a management
00:20:39 - server function, a key distribution server, and also
00:20:44 - a time server. 00:20:46 - Now, while a lot of Fibre Channel
devices will provide 00:20:50 - many of these, like an MDS switch,
for instance, from 00:20:53 - Cisco, it's important for you to
realize that some of those 00:20:57 - functions can easily be provided
00:21:01 - by an external server. 00:21:03 - Remember at the beginning of this
Nugget, I promised you 00:21:06 - that we would peel back the curtains
and we would take a 00:21:09 - look at Fibre Channel behind the
scenes. 00:21:12 - Let's get into that now.
00:21:13 - One of the things a moment ago we learned that Fibre Channel
00:21:16 - needs is a Fibre Channel name server. 00:21:18 - And do you know why?
00:21:20 - We have all those crazy names. 00:21:21 - We have the Fibre Channel IDs,
we have the Port World Wide 00:21:25 - Names, we have the Node World Wide
Names, certainly a more 00:21:29 - robust, a more complex naming architecture
that we have in 00:21:32 - our LANs that just have the simple
MAC address. 00:21:35 - It is the job of the Fibre Channel
name server to go 00:21:39 - ahead and track all of this information
for us, including 00:21:42 - some operational parameters that
are used, like the upper 00:21:46 - layer protocol mappings and the
classes of service that 00:21:49 - can be utilized in our Fibre Channel
infrastructure. 00:21:52 - The Fibre Channel name server is
a distributed database that 00:21:57 - our MDS switches implement in the
Cisco infrastructure. 00:22:02 - You know something else about this
Fibre Channel name server 00:22:06 - process that's pretty interesting,
is it does what 00:22:08 - we call soft zoning.
00:22:10 - Now, what soft zoning means is that the worldwide node
00:22:15 - lookups can be utilized to verify zone membership.
00:22:19 - So we have a very passive kind of zoning that's enforced here
00:22:24 - by providing information only about those nodes that are
00:22:28 - truly in the zone of that device that is requesting zone
00:22:33 - information. 00:22:36 - Now, if this is a distributed database
across these 00:22:39 - different devices, we need some
type of intelligence to 00:22:44 - handle that synchronization.
00:22:46 - That intelligence is another acronym I want you to know.
00:22:50 - It's the registered state change 00:22:53 - notification, or RCN process.
00:22:57 - Now, something else that we are going to get into here is
00:23:01 - the fabric login process. 00:23:04 - This is probably one of the most
hysterical acronyms in 00:23:08 - all of computer networking.
00:23:10 - I love to say it. 00:23:12 - Say it with me.
00:23:12 - You ready? 00:23:13 - FLOGI--
00:23:14 - isn't that fun to say? 00:23:15 - FLOGI.
00:23:16 - If you love that, pretty soon, we're going
00:23:18 - to talk about PLOGI. 00:23:20 - That's right.
00:23:21 - So let's get into the details here of this
00:23:24 - fabric login process. 00:23:27 - Now, a key element to this fabric
login process is the 00:23:31 - fact that the highest 16 addresses
in our 24-bit fabric 00:23:37 - address space are reserved.
00:23:40 - And this is going to be a comfortable concept for you if
00:23:44 - you are coming from the world of local
00:23:47 - area networking expertise. 00:23:49 - For instance, the broadcast alias
is all F's. 00:23:54 - If we have all F's and then the
address ending in an E, we 00:23:59 - have the reserved, well-known address
for the 00:24:02 - fabric login server.
00:24:04 - So there are plenty of reserved, well-known addresses
00:24:09 - in the Fibre Channel system that are going to facilitate
00:24:12 - processes like login and accessing key services in the
00:24:18 - storage area network infrastructure. 00:24:21 - So here you can see a conceptualization
of a couple 00:24:24 - of N-Ports and their respective
F-Ports. 00:24:28 - Now, before these N-Ports can begin
exchanging data in our 00:24:32 - Fibre Channel infrastructure, three
things need to happen. 00:24:37 - You ready?
00:24:38 - First, the N-Ports must log into their attached F-Ports.
00:24:43 - This process is indeed known as our fabric
00:24:47 - login, or FLOGI process. 00:24:50 - Next, the N-Port must log in to
its target N-Port. 00:24:55 - This process is known as the port
login or PLOGI process. 00:25:03 - Then what happens is the N-Ports
must exchange 00:25:06 - information about the upper layer
protocols. 00:25:10 - They need to ensure that the initiator
and the target 00:25:13 - process can communicate properly.
00:25:15 - And this process is known as the process login, or PRLI.
00:25:23 - So pretty easy, right? 00:25:25 - We have the fabric login process,
the port login 00:25:29 - process, and then the process login
process. 00:25:33 - So these storage area network devices
can communicate in 00:25:38 - that infrastructure.
00:25:39 - Now, realize when these processes succeed, the node
00:25:43 - ends up getting its unique port number.
00:25:46 - And it's going to be communicating with the name
00:25:49 - server this important information for the Fibre
00:25:52 - Channel infrastructure. 00:25:54 - Realize that after all this process
is complete, there are 00:25:58 - additional port and address discovery
commands that can be 00:26:02 - utilized to query and verify particular
fabric and port 00:26:08 - parameters without having to do
this whole port login 00:26:11 - process all over again.
00:26:13 - Let me give you some of these particular commands that can
00:26:17 - be utilized. 00:26:19 - A discover address can be used
to confirm the address of 00:26:23 - another port or to discover whether
the other port has a 00:26:27 - hard-coded address.
00:26:29 - Please note that this is abbreviated ADISC.
00:26:35 - A discover fabric service parameter can be used to
00:26:38 - verify the Cisco fabric service parameters that are in
00:26:41 - use in your fabric. 00:26:43 - And notice the abbreviation here
is FDISC, or F-D-I-S-C. 00:26:49 - And finally, we have the discovery
port service 00:26:52 - parameter that can be used to verify
the service parameters 00:26:55 - that exist for another N-Port.
00:26:57 - And this is abbreviated PDISC. 00:27:01 - Now, let's transition to a topic
that is so important and 00:27:04 - we've been at this for such a while
now. 00:27:07 - I need you to do something for
me. 00:27:09 - Pause this Nugget.
00:27:10 - Please trust me. 00:27:11 - Find the Pause button, pause this
Nugget. 00:27:14 - Stand up wherever you are right
now, stretch. 00:27:18 - Get re-energized.
00:27:19 - Check back into this Nugget, because we're about to cover
00:27:23 - something that is super, super critical and is going to
00:27:26 - impact your ability to understand a future Nugget
00:27:29 - called Fibre Channel over Ethernet Explored.
00:27:33 - So did you do it? 00:27:34 - Did you pause me?
00:27:36 - Welcome back. 00:27:37 - What we're going to do right now
is discuss Fibre Channel 00:27:41 - flow control.
00:27:42 - When they designed Fibre Channel, they said, hey, we're
00:27:45 - going to get a leg up on those silly guys over in the local
00:27:49 - area network world. 00:27:50 - Over there, packets can become
dropped. 00:27:53 - That would be a disaster for those
SCSI commands. 00:27:57 - Remember the SCSI commands think
they're over that 00:28:00 - highly-reliable short parallel
cable inside a machine. 00:28:05 - They are not going to tolerate
any kind of packet loss. 00:28:09 - So what Fibre Channel engineers
did was they 00:28:12 - implemented what is called a credit-based
strategy for flow 00:28:18 - control within the particular technology?
00:28:22 - Specifically, here's how it works. 00:28:25 - The transmitter won't send a frame
until the receiver tells 00:28:32 - the transmitter it can accept another
one. 00:28:36 - What does this mean?
00:28:37 - It means that the receiver is always in control of this
00:28:42 - particular flow of information from the
00:28:44 - transmitter to the receiver. 00:28:47 - This eliminates the possibility
of frames being 00:28:50 - lost due to things like buffer
overruns, and it maximizes 00:28:56 - performance under very high load
conditions in your 00:29:00 - storage area network.
00:29:02 - But wait a minute here. 00:29:04 - We know that the introduction of
the switch fabric gives us 00:29:09 - a very scalable solution for Fibre
Channel. 00:29:13 - But what about multiple paths?
00:29:16 - What about the potential for loops when we get a redundant
00:29:20 - environement built here with our MDS
00:29:23 - switches from Cisco Systems? 00:29:26 - Well, they borrow from other great
concepts 00:29:30 - that we're used to.
00:29:31 - They created the fabric shortest path
00:29:35 - first routing protocol. 00:29:37 - Let me say it again, because you
probably 00:29:38 - don't believe me--
00:29:39 - the fabric shortest path first routing protocol.
00:29:43 - This will avoid loops. 00:29:46 - It will ensure that frames in a
communication 00:29:49 - follow the same path.
00:29:51 - It will ensure in-order delivery in our
00:29:54 - storage area network. 00:29:55 - And just like OSPF does, it will
use a cost-based 00:30:02 - algorithm in order to make these
routing decisions. 00:30:05 - The cost of a given path ends up
being determined by two 00:30:10 - factors in the Fibre Channel world--
00:30:12 - the speed of each interswitch link along the path and the
00:30:17 - number of hops that are in the path.
00:30:21 - Now, in an earlier Nugget, I did allude to the various
00:30:25 - methodologies of implementing security that we can have in
00:30:29 - the storage area network environment. 00:30:31 - Let me just refresh you on that
here, as we wrap up this 00:30:35 - Nugget on Fibre Channel.
00:30:36 - One of the things you know that we can do is zoning.
00:30:40 - And I'm not talking about that soft zoning that we looked at
00:30:44 - with our name service. 00:30:45 - I'm talking about really hard zoning
your environment so 00:30:50 - that devices can't access other
devices 00:30:53 - outside of their zone.
00:30:55 - But it turns out that zoning has tremendous scalability and
00:31:01 - manageability problems. 00:31:03 - So what did Cisco invent as an
alternative? 00:31:06 - We talked about it in this Nugget,
didn't we, the virtual 00:31:10 - storage area network.
00:31:12 - VSANs really bring tremendous benefits to an environment
00:31:16 - that had previously just been zoned. 00:31:20 - Something else to consider is how
we protect our LUNs. 00:31:25 - When it comes to the LUN protection,
there are two 00:31:28 - techniques, remember.
00:31:29 - There is a LUN masking, and there is LUN mapping.
00:31:35 - You could do both of these simultaneously in your
00:31:38 - infrastructure. 00:31:40 - But typically we find that LUN
mapping is utilized only in 00:31:45 - environments where LUN masking
is not available. 00:31:50 - So you might be experiencing a
bit of deja vu right now. 00:31:54 - If you are new to storage area
networking and Fibre Channel, 00:31:58 - you might be experiencing that
same feeling that you got when 00:32:03 - you first started to study switches
in local area network 00:32:08 - environments.
00:32:09 - Just like back then, don't be overwhelmed. 00:32:12 - This might be a Nugget you have
to watch two, three, or 00:32:15 - four times.
00:32:17 - You may have to make flash cards to learn this miraculous
00:32:21 - new vocabulary that we have talked about as we've gone
00:32:25 - throughout this Nugget. 00:32:26 - Have no fear.
00:32:27 - You will master this material. 00:32:29 - You will understand where it fits
in your modern data 00:32:33 - center architecture.
00:32:34 - And we will even build upon this material as we go
00:32:37 - throughout this course and many other courses together
00:32:41 - here at cbtnuggets.com. 00:32:43 - My name is Anthony Sequeira.
00:32:45 - I really sincerely hope that this has been
00:32:48 - informative for you. 00:32:50 - And I'd like to thank you for viewing.