Fibre Channel SAN Storage Overview Tutorial Video

in this lesson we're going to start covering our sound protocols and the best one to start with is the original sound protocol which was fiber channel before we start getting more in-depth on fiber channel fool let me give you some general sign terminology first we have a 1-1 is a logical unit number that represents a disk that will be presented to a host so the client connects to it someone it sees the one as if it was a local hard drive ones are specific to our Sun not our knives protocols other terminology the client is known as the initiator and the storage system is known as the target fiber channel is the original sign protocol and it's still very popular today it uses dedicated adapters cables and switches and it's different than Ethernet at all layers of the OSI stack including the physical level so you can't use an Ethernet network card or an Ethernet switch for fibre channel it uses dedicated fibre channel infrastructure this is different than our manaos protocols which run over Ethernet the FCP fibre channel protocol is used to send scuzzy commands over the fibre channel network so if your client had a local hard drive it would send scuzzy commands to that local hard drive with Sun it's sending scuzzy commands same commands but over a network now favorite channel is a very stable and reliable protocol so it's very popular with old-school storage engineers fibre channel is lossless unlike TCP and UDP with TCP the sender sends traffic to the receiver the receiver will periodically send acknowledgments back if the sender doesn't get an ignore which meant and back then it knows that the traffic was on lost in transit and it will resend the traffic with UDP its best effort we aligned apparently our protocols to resend traffic if anything gets lost so that's how traffic works over Ethernet networks it's different in favor Channel it's built into the protocol that we never lose any traffic that we're going to have to recover from fibre channel currently supports find widths of 2 4 6 8 and 16 gigabits per second depending on the hardware that you've got to plate fibre channel uses a dedicated network for the storage so you see here we've got our standard local area network running over Ethernet which gives clients access to their servers in the example here the server is going to be using fibre channel storage for that we have separate dedicated network infrastructure so putting them both together let's see the server that you see in the middle here is a web server and the client up at the top is going to be accessing a web page on that web server the client will access the server over the normal Ethernet local area network then to fetch the actual web page from its storage the server will connect to its storage over the fibre channel network so the network you see down here on the server it has got HBAs which are host bus adapters that's the equivalent of a network card in Ethernet and that will connect to dedicated Fibre Channel switches and then our storage system also has HBAs to connect to our fiber channel network fibre channel uses worldwide names WWN s for its addressing the w w ends our 8 byte addresses that are made up of 16 hexadecimal characters and you can see an example of their format it's a big long hexadecimal address the worldwide node name the WWN n is a same to annoyed in the storage network and the same WWN n can identify multiple network interfaces of a single network node so we've got a host it's got may be multiple HBAs or multiple ports in the HP is the WWN and signifies that one host w WN n stands for worldwide node name you may sometimes see this also being referenced as NW w and the nord worldwide name that is exactly the same thing so w WN n or n www n they're the same thing just two ways of seeing it our hosts also have worldwide port names w WP ends a different w WP n is assigned to every individual part on a node so if we had a multi-part HBA in the same host each part on that h ba would have a different w w pn w WP ends are the equivalent of MAC addresses in Ethernet the w w pn is burned in by the manufacturer of that HB a and it's guaranteed to be globally unique just like a MAC address in Ethernet w WP ends are sent to HBAs on both the clients and the storage system as well so the parts on the client and the parts on the storage system they'll all have unique w WP ends we are primarily concerned with w WP ends not w WN ends when we're configuring fiber channel networks and just like w WN ends can also be known as n w w and w WP ends are also sometimes known as pw w and again it means the same thing aliases can be configured to make configuration and troubleshooting easier on your storage for example we could create an alias named exchange server for the exchange server which has got that big long wpn that you see there by configuring an alias it means that when we are configuring settings for that server we can specify the alias rather than the wwp n this is more convenient and it also makes it less likely that we're going to put in any typos for a fiber channel you're going to need to configure the Fibre Channel switches and the storage system aliases can be used on both of them prosecutor zoning is configured on the switches to control which hosts are allowed to communicate with each other the servers which are clients in the storage system will be able to communicate with the storage system but servers will not be allowed to communicate with each other over fibre channel this increases security and it also makes the fibre channel network more reliable and stable I've got an example of how to configure zoning here the command syntax in this example is for a school fibre channel switch brocade are another popular manufacturer of Fibre Channel switches so in the example here I've got a couple of servers down at the bottom which are clients of the storage system each of the servers have got an HB a with a w WP N and I've configured an alias with the server's name for that particular w WP n the storage system up at the top also has its HBA and it's w WP n and we've configured an alias for it as well with the zoning we've got zone name server one so we configure is zone which is dedicated for server one and we see member FC alias server one and member FC alias NetApp controller one this allows server one to talk to its storage we also configure a zone for server two so I've got zone name server to member FC alias server two and member FC alias net up controller one that zone allows server two to talk to its storage so I configure is all for each of my different servers which are clients office storage system then I group all of those zones together into a zone set so I've got z1 set name my zone set and actually a member server one and member server to the zoning configuration is done on the Fibre Channel switches and with this configuration server one can talk to its storage and server two can talk to its storage that the two servers can't talk to each other over the fibre channel network as well as configuring zoning on our switches we also need to configure on one masking on the storage system it's critical that the right one is presented to the right host if the wrong host was able to connect to a one then it would be liable to corrupt it so if you look back if sample here I don't want server two to be able to connect to server ones and one the zoning on the switches make sure that the server's can't talk to each other that they can both talk to the storage so how do I make sure that they can't connect to each other's and lungs that's where one masking comes in so zoning prevents unauthorized hosts from reaching the storage system and it prevents hosts from talking to each other over the favorite channel network but it doesn't prevent a host from accessing the wrong one once it gets to the storage system on one masking is configured on the storage system to walk on one down to the host or hosts who are authorized to access it to secure your storage you need to configure zoning on your switches and one masking on your storage system here's an example of how we would configure one masking so I've got my storage system up at the top and I've configured a boot one for server one and I've also configured a boot one for server two for my server one boot one I see that the only initiator that can connect to that is server ones wwp N and I see that the server to boot one the only initiator that can connect to that is server two's w WP n IQ dogs who have used aliases here rather than typing in the w WP n in my configuration next thing to talk about is switch the main IDs each switch in the favorite channel network will be assigned a unique domain ID I think the name of e is quite confusing because I think of a demain ID I think that would mean an ID for the entire domain of switches but it doesn't mean that at all the domain ID means a unique ID for each switch that is in that favorite channel network one switch in the network will be automatically assigned as the principal switch it is in charge of assigning domain IDs to the other switches each switch and learns about the other switches in the network and how to wrote to them based on their domain ID when our servers are started systems HBA horizon-- it will send a log e which is a fabric login request - it's a nucleotide favorite channel switch the favorite channel switch that's plugged into the switch will then assign it a 24 bit FC ID which is the favorite channel ID address so what's happening is the server is powered on the HB a comes online it sends a flog e to the favor channel switch that is plugged into seeing here this is my W wpn please assign me an FC ID so that I can communicate on the favorite channel network the FC ID our seem to horst's is made up of the switches the main ad that is plugged into and the switch part the FC ID is similar to an IP address in Ethernet it's used by Fibre Channel switches to route traffic between servers and their storage switches maintain a table of FC ID 2w w pn address mappings and what part the cost is located on so every switch in your favorite channel network it learns about the domain IDs of all of the other switches in the network it oxen learns about the wwp ends of all the hosts that are attached to that network and based on the FC ID it knows the domain ID of a switch that that host is plugged into because we have all this information they're able to route the traffic between the hosts here's a picture of the fabric login process working so I've got server one down at the bottom which is a client of my storage system up at the top when server one power is on its HP a will send a fabric logon a foggy to the switch that it is attached to and the switch will assign it an FC ID if I know did a short log E database on that switch I would see the interface that the server is plugged into and its FC ID - w WP n the same thing happens with our storage system up at the top when it powers on its HP a will send a flog e to the switch that it's plugged into the switch will assign that part an FC ID and if we do a short log E database on that switch will see the interface that the storage is plugged into it's FC ID and it's wwp n the Fibre Channel switches share the flog e database information with each other using the fcns the fibre channel name service each switch in the network learns where each w WP n is and what its FC ideas and how to route traffic there here's an example of visual fcns database so the short log E database command that will only work on the switch where the clients are directly plugged into the switches share that information with each other through using FCS so if we do a show F G and s on either switch we will see the FC ID and the w WP n of all of the hosts out in our network because the FC ID is derived from the domain ID which is how we identify our switches the switches now know how to route traffic to any horse than the network after the foggy fabric Morgan process is complete the initiator will send the palagi the part long again based on the zoning configuration on the switch the host will learn it's available target wwp NS on its storage so you can see the part login process here server one down at the bottom sent in a flog e a fabric login and it was assigned its FC ID from there when that process is complete it will send a plug-in a port login to its locally attached switch the switch will check its zone in configuration and as you can see from the example here it will allow you the server to talk to its storage finally we have the PLR I the process login the initiator host will send APL RI process login request to its target storage the storage system will grind access to the host to its lungs based on its configured one masking so back to the same example again after the server has done its flog e and its pluggy it will send the processor login directly to the storage system and this will allow it to connect to its ones servers access server storage will invade ibly be mission-critical for the enterprise so we're not going to want to have any single points of failure redundant fibre channel networks should be pute in place muin is fabric a and fabric b or san a and san B each server and storage system host will be connected to both fabrics with redundant HP a parts Fibre Channel switches distributes shared information to each other such as domain IVs the fcns database and Zoning when we configure zoning and a fabric we only need to do it in one switch and it will then be distributed to the other switches from there this makes things more convenient for us but resolves or a potential downside here as well where if we make a miss configuring it's going to be replicated between all the switches and the fabric if an adder and fabric I was able to propagate to fabric B this would bring down both fabrics and it would drop the server's connection to their storage this would be disastrous for this reason switches and different sides of the fabric are not cross connected to each other both sides of the fabric are kept physically separate this is different than how we often do things in Ethernet line networks where switches will be cross connected to each other in fibre channel networks we have two fabrics fabric a and fabric B and hosts are connected to both fabrics but the switches are not switches are dedicated to either fabric a or fabric B let's have a look at what that looks like so you see here I've got server one down at the bottom and it's got two HBA parts for redundancy the first part is connected to fabric II the second part is connected to fabric B I also do that on my storage my storage has also got redundant HBA parts one is connected to fabric a the other is connected to fabric B the two fabrics are kept strictly physically separate from each other you can see I've signified that by drawing a big red line up the middle here so you can see my hosts are connected to both fabrics but the fabrics are kept physically separate from each other this means that if I have a Miss configuration in fabric a fabric a could be taken down but that Miss configuration cannot be propagated to fabric B my server would lose connectivity to it started over fabric a but it can still get there over fabric B so I don't have an outage okay but wait we are going to have at least two controllers for redundancy of our storage system so our network topology is actually going to look more like things we've still got the fabric a and the fabric b fiber channel networks where our switches are kept physically separate from each other in the example i've got a couple of clients server 1 and server to which are connected to both fabrics and now up at the top I've got two separate storage system controllers for redundancy the controllers just like the server's again actors and hosts so my storage controllers are connected to both fibre channel fabrics the way that I'm going to configure my zoning on my switches now on one of the switches which is in fabric a I configure is own first server one so actually user name server one member FC alias is server one a that is the HP a part on server one which is connected to the fabric a network and then I also see a member FC alias controller one a and member FC alias controller to a both controller one and controller two are connected to my fabric a network and my server can reach its storage through either controller also on that same fabric a switch I'll configure a zone for server two so actually so name server to member FC alias s to a that is the W wpn of the HBA which is on server two and connected to my fabric a network and then member FC alias controller one in and member FC alias controller to a the HP is on controller one and two that are connected to the fabric a network and I tie it all together into a zone set IC 0 and set names when set a member server 1 and member server 2 I can figure that on one of the to fabricate switches and it will propagate it to the other switch so it saves me having a do a duplicate configuration on both I also need to configure my fabric B switches so on there I also configure is Owen for server one I say zone name server one member FC alias s1 be the HB a on server one which is connected to my fabric B switch and member FC alias controller one B and member F cos controller 2 be for the HBAs on controller 1 & 2 which are connected again to the fabric B Network I also need a zone for server 2 so I do a similar configuration there zone name server 2 member FC Elias s to be the HPA on server 2 and member FC Elias controller 1b and member FC alias controller to be the w WP ends for the HB a s on my controllers that are connected to fabric B and tie it all together in my zone set I've got zone set name zone set B member server 1 and member server 2 I can figure that on one of the two fabric B switches and it will propagate it to the other fabric B switch so that takes care of my zoning on my switches I also need to configure on one masking on the storage system on my storage I've got a boot one for server one and I've also got a boot one for server two I say that the wwp ends for server 1a and for server one B are both able to connect to that one so server one will be able to connect to it over either fabric a or fabric B and for the server to boot one I do the same thing I say that the members of its one masking our server to a and server to be next topic to tell you about is TP G's target portal groups all of the parts on the storage system which initiators can access our storage through are added to a target part or group in the example that you see on the slide here parts controller one in controller one B controller to a and controller to B are added to a target portal group each of those parts will have its own unique w WP n I add all four of those w WP ends to a target portal group and the hosts are going to learn that they can connect to their storage through any of the w WP ends in that t PG n Lua is asymmetric illogical unit assignment it's used beber storage system to tell the claim to which I would preferred perhaps for it to use direct pipes to nodes I know it is another name for a controller in the storage system owning the one are marked as optimized paths other pipes are marked as non-optimized paths let's look at how this is going to work so we've got the same example we were looking at earlier where I've got a storage system which is made up of two nodes which are controller one and controller to controller one into the disks where the one for server one is currently located server one can get to its one through either controller one or controller - but it would be better for it to go to controller one because that is a direct path the storage system can give the server all of this information lets it know all of the paths that it can take to get there and which are the preferred paths and it uses an WA to do that so the server learns about optimized path one which is going through fabric a and terminates on HB a controller one a it also learns of a optimised path - which is going through fabric B and which terminates on HB a controller 1b path one and path - are optimized paths because we go to controller one which is where the line is the server will also learn about non-optimized path 3 which goes through fabric a and terminates on the HB a controller to a and non-optimized path for which goes through fabric B and which terminates on controller to B so the server has four different paths so that it can take to get to its storage and two of them are better optimized paths during the login process initiators will detect ports available to connect to the storage on in the target portal group we will notify whichever preferred paths multipathing software on the initiator will choose which path or perhaps to take to the storage all popular operating systems to all flavors of Windows Unix Linux VM ware etc have multipathing software which supports active active or active standby paths the client will automatically film over to an alternate path if the one is using fields so considering our example where we had the two optimized paths and the two non-optimized perhaps on our multipathing software we could choose to do active active load balancing over both optimized pipes or we could do active standby where we send the traffic over one of the optimized paths and if it feels we feel over to the other optimized path for example here's an example of what multipathing software looks like from a client a couple of popular manufacturers for each pas are mu X and Q logic here we're looking at a screen shot of Q logics and server and this is software which is installed on the server it's going to be a client of the storage we get a nice graphical interface and you can see the server is over here on the left this is the server that the software is running on it's plugged into fibre channel switch one and it's detected four possible paths that it can choose to get to its storage we can go on in the server run the software and then we can configure it with the width we want to get to the storage so as you've seen client connectivity to sand storage is fundamentally different to how Ethernet networking works I already had a lot of experience in Ethernet networking when I came to storage and I found this pretty amazing in Ethernet if you want to connect a client to a server you have to point the client at the server's IP address but with fibre channel because of the login process the client will automatically detect it you can go on to the favorite channel switch that the client is connected to and you could do a short foggy database from there to learn what it's w w pn is you could then configure the zoning on your switches you can also again using that w w pn create the one on your storage and configure the one masking on your storage you can then power it on that client you could boot it up from the operating installation CD and it will automatically detect it's in one without you having to configure anything on the client and you could install the operating system on its remote one in Ethernet networking all the routing and switching decisions are handled by network infrastructure devices all of them but in science storage multipathing intelligence is enabled on the client and host

Channel: Flackbox

Views: 95,063

Rating: 4.9376855 out of 5

Keywords: FC, Flackbox, Storage, Storage Area Network, Fibre Channel, EMC, HDS, SAN, NetApp

Id: zb2kEtjMmPg

Length: 29min 53sec (1793 seconds)

Published: Sun Jul 03 2016