BladeSystem Virtual Connect
cancel
Showing results for 
Search instead for 
Did you mean: 

Virtual Connect Configuration

duhaas
Advisor

Virtual Connect Configuration

New to the blade game and we have two Cisco Virtual connect in ports 1 and 2. My question is how I should configure it network wise. Right now I am thinking of taking four ports and trunking/etherchanneling them on my 4507 Cisco Switch. Than taking another four and trunking them into my iscsi switch. Is this a decent practice, anyone know of any gotchas???
7 REPLIES
Jeroen_Kleen
HPE Pro

Virtual Connect Configuration

The best starter would be to take a look at the VC ethernet cookbook. There are a lot of good best practices listed. See: https://h30340.leveragesoftware.com/group_file.aspx?FileID=797fc8d6d91b4d78bab04a1043d227aa Enjoy, Jeroen
(I am an HPE employee and a HPE OneSphere evangelist)
Engage as well with HPE OneSphere experts at our new slack channels for quick and agile discussions around HPE OneSphere and HPE Dev: https://www.labs.hpe.com/slack
If your question is resolved then please acknowledge that and/or provide Kudo's.
tubbaguts
Occasional Visitor

Virtual Connect Configuration

D Billot hosted two very good labs at the HP Tech Forum last year. He has posted the lab guides on this site. Both are detailed, step by step guides to setting up VC. "HP Virtual Connect 101 Lab Guide" https://h30340.leveragesoftware.com/group_file.aspx?FileID=b642a9618e6d461f819163bd3248473c "HP Virtual Connect 201 Lab Guide" https://h30340.leveragesoftware.com/group_file.aspx?FileID=47ffbc8dc5a746d1a9d3a0edb9e6e268
duhaas
Advisor

Virtual Connect Configuration

Thanks for all your help guys
duhaas
Advisor

Virtual Connect Configuration

One more question, now that I made my first attempt @ configuring thinga, I created two sets of four port lacp vlan trunks (questioning if i could of just created an 8 port lacp trunk), four ports go to one VC, the other four go to VC2. After setting up all the ports to be in my uplink group, only the first port on each VC switch show linked active, not sure i understand why the others all show linked/standby, looks like this: Bay1 Port 1 Active/linked Bay1 Port 2 standby/linked Bay1 Port 3 standby/linked Bay1 Port 4 standby/linked Bay2 Port 1 Active/Linked Bay2 Port 2 standby/linked Bay2 Port 3 standby/linked Bay2 Port 4 standby/linked
David Billot
Frequent Advisor

Virtual Connect Configuration

Ok, for the first question, yes, you could have simply created an 8 port LAG. Whatever suits your needs best. As for your Active/Standby results, this can happen for a couple of reasons, but I'll give a few to start with: 1) Ensure that your vNet connection mode option is configured for Auto (default) and not for Failover. The Auto setting supports port trunks, while the Failover mode does not. 2) Verify your switch configuration. A common oversight is to either set an improper channel mode (channel mode needs to be either Active or Passive), or to not configure the channel for LACP (Cisco uses PAgP by default which is not compatible with VC port trunks). If none of the above suggestions resolves this, then you might want to post your switch configuration so that we can review your settings. Refer to the VC User Guide for examples of common switch configuration settings. The VC cookbook also provides an abundance of configuration examples. The VC cookbook can be found in the download area of this Interest Group. Thanks, Dave [Updated on 2/26/2008 10:04 PM]
duhaas
Advisor

Virtual Connect Configuration

Thanks for your feedback, you helped lead my network guys in the right direction. What are your thoughts on the two sets of 4 vs an aggregate of all 8??? Any benefit one way or the other.
David Billot
Frequent Advisor

Virtual Connect Configuration

I don't really have enough insight into your overall design to make an intuitave recommendation. However, from a generalized standpoint, having two sets of port trunks associated with single vNet will put one port trunk in standby mode -- loop prevention. The only way to have two port trunks enabled in active/active state, is to create separate vNets for each port trunk. So you would then have for example, vNet-1A, and vNet-1B, but they could both be carrying the same network/VLAN traffic, just in separate channels. You would then need to create NIC teams/bonds in a transmit load balance (TLB) mode in order to take full advantage of both vNets/port trunks simultaneously (see the VC cookbook for such a scenario.) The advantage of this design is primarily one of load balancing at both the NIC and the upstream switch level. On the otherhand, if your intent is to provide 8Gb of aggregate bandwidth for your blades using a single port channel, then you would also likely want to have a redundant port channel on another module (eg., module 1 and module 2), and you would then associate both port trunks in a single vNet. As stated above, this configuration will force one of the port trunks into a standby state. So that would mean two port trunks each with 8 ports each. BTW, redundant port trunks don't have to match bandwidth. You could, if your design allows, have the primary port trunk configured with 8Gb, and the secondary as something less than 8Gb (e.g., 4Gb.) This would reduce your upstream port cost, but of course you would be limited to that 4Gb bandwidth during a failover period. Now, if big pipes is what you really need, then you might consider just using the 10Gb ports instead, and forgo the port trunking altogether (if you have enough 10Gb ports on the upstream switch of course) if you want to further reduce your cable footprint. We also have a VC fiber based Ethernet module now that provides fiber SXP 10Gb links as well as fiber SFP 1Gb links and copper links as well. That may be something to consider. And, BTW, you can even trunk the 10Gb ports as well if you are really needing a big pipe -- that would give you 20Gb per port trunk. Hope that helps. Thanks, Dave