Location: Ask the Storage Experts

Discussion: Setting up NIC Teaming with Link Aggregation for Failover/RedundancyReported This is a featured thread

Category: Discussion Forum / General Discussion
Showing 5 posts

bmccleary
Setting up NIC Teaming with Link Aggregation for Failover/Redundancy
Sep 9 2009, 5:58 PM EDT | Post edited: Sep 9 2009, 5:58 PM EDT
We have a R805 server that has the embedded Broadcom 5708 NIC cards that are capable of NIC Teaming. We also have two PowerConnect 6224 switches that have the capability to do Link Aggregation. Now I believe that these features on the NIC and Switch allow for redundancy/failover. I would like to find a step-by-step manual that outlines how to connect and configure both of these units (in Windows 2008) so that there is full redundancy. Meaning that if a NIC port fails, a cable fails, a switch port fails or the entire switch fails, that the other network path automatically takes over all requests without any intervention. All of the tech specs say that this is able to be done, but I have found nothing that shows how to do it. Even the manuals describe about each setting (in a highly technical manner), but not what to set them to in order to achieve our needs. I would like to find a cable diagram and instructions, using both switches, on how to properly setup the Link Aggregation. Any help is appreciated. Do you find this valuable?    
andreaserson
andreaserson
1. RE: Setting up NIC Teaming with Link Aggregation for Failover/Redundancy
Sep 10 2009, 5:07 AM EDT | Post edited: Sep 10 2009, 5:07 AM EDT
There are four types of teaming with your Broadcom LOMs.

This documentation should contain everything you need to know:
http://support.dell.com/support/edocs/network/BroadCom/R125805/en/index.htm

Check in particular out these chapters:
Introduction - Teaming
Configuring Teaming
Using Broadcom Advanced Control Suite 3 Do you find this valuable?    

bmccleary
2. RE: Setting up NIC Teaming with Link Aggregation for Failover/Redundancy
Sep 18 2009, 3:40 PM EDT | Post edited: Sep 18 2009, 3:40 PM EDT
Thank you for your post. I have been reading that document and am now setting up our lab environment to test. I have a few questions for you regarding, in your opinion, best practice and recommended settings, especially when it comes to the switch (since the document link you provided only really addresses the NIC cards).

<Questions follow in next reply...> Do you find this valuable?    

bmccleary
3. RE: Setting up NIC Teaming with Link Aggregation for Failover/Redundancy
Sep 18 2009, 3:40 PM EDT | Post edited: Sep 18 2009, 3:40 PM EDT
1) Of the four methods available to setup teaming, it seems like the "Smart Load Balancing and Failover" and "Link Aggregation" are the best choices. Do know which method proves the easiest to maintain while still establishing full failover that should a switch fail, the other connection automatically takes over with no latency or down-time?

2) I am new to these terms, but do we need to enable/configure spanning tree protocol or link aggregation on the switch? If so, do you know of a white sheet/manual (similar to the one you provided for the NIC's) that is specifically related to configuring this for the switch? I have the manual for the switch, and it talks about what settings are available for these options, but doesn't give a plain text reason of what settings to use in this scenario. Again, the goal is to prevent any network outage in the event a total switch failure. Performance is secondary to this need.

3) We purchased the stacking module for our 6224 switches thinking that this was needed to allow for automatic failover, but I have read in some other Dell forums that if the master switch fails, that the entire stack can go down for 10-12 seconds while another member switch reconfigures itself as master. This downtime would be detrimental to us because our Virtual Machines are accessed on the SAN via iSCSI. If the stack goes down for 10-12 seconds, that basically means that the VM's will crash. If this is the case, is it better to configure each switch as a standalone switch with no stacking module? I know it will be more maintenance, but if it prevents the down-time for a member switch to become the master, then that's completely worth it.

Your input is greatly appreciated!
 Do you find this valuable?    

bmccleary
4. RE: Setting up NIC Teaming with Link Aggregation for Failover/Redundancy
Sep 23 2009, 6:38 PM EDT | Post edited: Sep 23 2009, 6:38 PM EDT
Any thoughts here...? Do you find this valuable?