Help with CX4 cables and Virtual Connects

[looking_for_help]

Hi,

We have 8 virtual connects on our c7000 enclosure. In Bays 1,2,5,6,7, and 8, we have VC-Enet. In Bays 3 and 4, we have VC-FC.

We are using BL465c blades with 6 NICs and 2 HBA's in them.

Do we need to use CX4 cables in this scenario for redundancy between the virtual connects?
If so, can CX4 cables be used between VC-Enet and VC-FC? Or at the minimum VC-FC to VC-FC (between bays 3 and 4).

On HP's website, I found documentation where the CX4 cables were linked between VC-Enet modules only. If this were to be used for redundancy, how do we take into consideration the VC-FC's?

Sorry if none of this makes sense. New to blades and virtual connect equipment.

[The Brit]

The CX4 cables are only used between VC-Enet modules. They are used as "stacking links" to join modules together. This provides a way for network traffic to move between ENet modules if necessary, to get out to the network.

you will find a diagram showing how to connect the stacking links for your configuration in the VCM User Guide.

http://h20000.www2.hp.com/bc/docs/support/SupportManual/c00865618/c00865618.pdf

it doesnt matter if your FC modules are geographically situated between your ENet modules, just run the cx4 cables across them.

[looking_for_help]

Thanks for the reply "The Brit".

If I'm understanding your reply correctly, are you saying that CX4 cables are used for a "shortcut" path for network traffic between VC-Enet in Bay 1 to say Bay 8?

What are other reasons for stacking / joining the modules together? Would redundancy be one of them?

I found this link from HP's website:

http://h20000.www2.hp.com/bizsupport/TechSupport/Document.jsp?lang=en&cc=us&taskId=115&prodSeriesId=3885293&prodTypeId=329290&objectID=c01724621

It seems like it's possible to stack VC-FC's. If it's true, can I stack my VC-FC for redundancy purposes between bays 3 and 4?

Is it possible to stack VC-Enet and VC-FC modules together?

Thanks for any help provided!

[Adrian Clint · Accepted solution]

First...
You cannot stack VC-FC modules. There is no connector to do this, no such concept. No communication needs to happen between VC-FCs or to VC-Eth. Redundancy is achieved with MPIO/Securepath over the two fibre connections and having more than one cable into each VC-FC and setting up dynamic balancing on the VC-FC.

What one of the questions in the link you found is talking about is that if you have two enclosures in a multi-enclosure domain and they are stacked then the VC-FC modules need to be in the same slots in both enclosures.

Second. Unless you have the VC-Eth 1/10Gb-F Ethernet Fibre modules in interconnects 1 & 2 then you cant do the multi-enclosure domain upgrade anyway. But you could use Virtual Connect Enterprise Manager to move profiles between enclosures.

Third. Stacking is linking all the VC-Eth modules together in a ring or a mesh. To..
(a) reduce number of uplinks from a blade enclosure to core switches.
(b) keep as much network traffic as possible within the VC modules.
(c) redundancy of connections out of the blade enclosure.

The VCs are internally stacked between 1-2, 5-6 & 7-8. On our config I put stacking links between 1-7, 2-6 & 5-8 so I have a VC ring of 1-2-6-8-7-1. So now blade nic can uplink thru any VC port if it is configured to do so....and it has two paths to get there.

[looking_for_help]

Thanks "Adrian Clint". The info you provided answered my questions.

Could you elaborate more on the VC-FC redunancy using MPIO/Securepath. I think MPIO/Securepath is software from HP. Where would have this software installed to implement the redundancy for VC-FC?

[Adrian Clint]

Yes MPIO and securepath are both pieces of software from HP.
http://h20000.www2.hp.com/bizsupport/TechSupport/ProductList.jsp?lang=en&cc=us&prodTypeId=18964&prodSeriesId=421492&taskId=135

http://h18000.www1.hp.com/products/sanworks/secure-path/spwin.html

MPIO is free.

What they both do is remove visible redundant paths to the same disk within the operating system. So if you have many paths to the same disk (which you will have at least 2) you only see one disk wether 1 or more of the paths are ok.
So if one VC-FC module goes down or all the uplink cables to the SAN switch(s) from one VC are removed - everything stays up.

[looking_for_help]

Hi "Adrian Clint". The redundancy you're talking about seems to be from a client's point of view where you can install the MPIO on an operating system. But what if you have fibre connectivity from VC-FC in bay 3 and 4 going to the same switch or to two different switches? Will be be able to achieve redundancy by itself?

[Adrian Clint]

This all depends on what is "redundancy" to you?
What do you think is going to fail more often/most?

Cables from two VCs to one switch obviously do not have redundant connections....but if the SAN connected unit you are connecting to has only one SAN port then you dont care.

If you go from a VC to a SAN switch (twice) then you do have more redundancy.