LACP accross multiple switches

I would like to know if I can nic team a server accross a pair of cisco switches (2960G)
I am unable to stack the switches at the moment but I have them trunked via an etherchannel.
Is is safe to run a single server with multiple nics in a LACP configuration to both switches?
What precautions do I have to take ex. spanning tree.  Or this not best practice?

thanks

FREDARCEAsked:
Who is Participating?
I wear a lot of hats...

"The solutions and answers provided on Experts Exchange have been extremely helpful to me over the last few years. I wear a lot of hats - Developer, Database Administrator, Help Desk, etc., so I know a lot of things but not a lot about one thing. Experts Exchange gives me answers from people who do know a lot about one thing, in a easy to use platform." -Todd S.

fgasimzadeCommented:
I dont think it is possible to configure LACP across 2 different switches
0
fgasimzadeCommented:
Small addition - Multi-chassis etherchannel is supported begging from Cisco 3750


https://supportforums.cisco.com/thread/2054801
0
Don JohnstonInstructorCommented:
If they are not stacked, then no.

If they are stacked, then yes.
0
Keep up with what's happening at Experts Exchange!

Sign up to receive Decoded, a new monthly digest with product updates, feature release info, continuing education opportunities, and more.

harbor235Commented:

Right, but that is for switch to switch etherchannel not server to switch, so the answer is no.

Curious though, if you had a VMware hypervisor with a nexus 1000V could you do multi chassis etherchannel? This is possible, check it out:

http://www.cisco.com/en/US/prod/collateral/switches/ps9441/ps9902/guide_c07-556626.html#wp9000289


harbor235 ;}

0
Don JohnstonInstructorCommented:
Actually, host to switch multi-chassis etherchannel is supported.

High-availability features such as EtherChannel and FlexLinks will work across stack members, increasing uptime and network connectivity.

http://www.cisco.com/en/US/prod/collateral/switches/ps5718/ps6406/white_paper_c11-578928.html
0
harbor235Commented:

Question said they were not stacked Don but I agree

harbor235 ;}
0
giltjrCommented:
It is safe to NIC team them, but you can't use LACP.

We do NIC teaming  with the NIC's going to different non-stacked switches all of the time.

If you NIC team them in active-active mode this gives you 2 times the outbound bandwidth and availability if one path fails.  It does NOT give you 2 times the inbound bandwidth that LACP would give you.

If you NIC team active-standby then you still have availability, but you only get the bandwidth of a single NIC in both directions.
0
FREDARCEAuthor Commented:
So far in my testing I setup the switchports to be lacp mode active on both switches.
It works,  with the server connected to both switches but failover takes a bit longer as opposed to it being connected to the same switch.  I am assuming spanning-tree is the cause of the extra time to recover the connection to the server.
0
fgasimzadeCommented:
Can you show us your config?
0
Don JohnstonInstructorCommented:
You cannot implement etherchannel to two separate chassis (in a Cisco environment) unless you use stacking, VSS or vPC.  

The only possible exception is if the use of the term "NIC teaming" is used to describe something completely different than etherchannel (or port aggregation).
0
fgasimzadeCommented:
I agree, that is why I asked to show us config
0
FREDARCEAuthor Commented:
There is no stacking.  just 2 switches that are trunked via an etherchannel.

Yes,  I am using "NIC teaming" on an HP server.  So,  I am wondering how this is working between 2 switches that are not stacked?  can someone explain?

thanks
0
fgasimzadeCommented:
LACP is not in place, this is 100%. It is just connected to 2 different switches and that's is
0
Don JohnstonInstructorCommented:
>So,  I am wondering how this is working between 2 switches that are not stacked?

It's not working. More than likely, one of the links is blocked by spanning-tree. But you do NOT have an etherchannel group in place.
0
fgasimzadeCommented:
Spanning-tree is a good idea, it explains why it "takes a bit longer" to failover, as you said.
0
giltjrCommented:
NIC teaming is not the same exact thing as LACP/Etherchannel.  You can do NIC teaming without using LACP/Etherchannel.  

NIC teaming without LACP/Etherchannel gives some of the same thing you get with LACP/Etherchannel, but not everything.

In a active-standby setup only one NIC is used.  If the path from that NIC goes down (unplug the cable, shut the port on the switch, turn the switch off) then the NIC teaming software will switch to the standby NIC.

In a active-active mode, the NIC teaming software will send packets out on both NIC's.  Data to the server only comes in on one NIC because when the server responds to an ARP it uses the MAC address of the "primary" NIC.  

The "primary" NIC is the 1st NIC in the list of NIC's.

Spanning tree should not come into play here.  The NIC teaming software will respond to ARP's with the primary NIC's MAC address.  There should not be any traffic coming into the server via the secondary NIC.  Outbound traffic can flow from ether NIC.
0

Experts Exchange Solution brought to you by

Your issues matter to us.

Facing a tech roadblock? Get the help and guidance you need from experienced professionals who care. Ask your question anytime, anywhere, with no hassle.

Start your 7-day free trial
FREDARCEAuthor Commented:
ok.  so I guess the real question is since I can't LACP between both switches.  What is the best setup?  If I am 'nic teaming' on the server and plug both nics to the same switch, then this means I will be able to run LACP on the switch and increase performance.  However, if the switch fails I lose both links to the server.  If I connect the server to both switches then I can't utilize LACP however I increase redundancy to the server in case of a single switch failure.  Am I correct in thinking this way?  Or should I not even consider one setup over the other?

thanks
0
giltjrCommented:
What is your ultimate goal? More bandwidth?  Or availability?  Pick one.

I am assuming you have gigabit NIC's.  Are you actually ever exceeding 1 Gbps for a length of time that somebody actually notices?
0
FREDARCEAuthor Commented:
Availability? but I also want to follow good practice.  

Are both methods acceptable?
0
giltjrCommented:
Good practice is pick what is most important to you and go with that.

Do you activity monitor your network bandwidth usage today?  If so do you exceed 1 Gbps for any length of time?  Do you even come close to that?

If the answer is you don't know, then start monitoring.  A couple of simple and easy free programs to do this are PRTG or MRTG.  I perfer MRTG, but although it runs under Windows, it is much more Linux friendly.  PRTG is the same basic thing as MRTG, but it is designed to run under Windows.

You will need to enable SNMP on your server.

If you don't come close to 1 Gbps, then go with availability.  When you start having problems because a single 1Gbps NIC can't provide the bandwidth needed, then you start looking at other options. Upgrading to 10 Gbps.   Getting stacked switches and LACP.

Get a second server, setup a cluster, do NIC teaming with LACP on both servers and connect each server to different switches.

0
FREDARCEAuthor Commented:
thanks all
0
It's more than this solution.Get answers and train to solve all your tech problems - anytime, anywhere.Try it for free Edge Out The Competitionfor your dream job with proven skills and certifications.Get started today Stand Outas the employee with proven skills.Start learning today for free Move Your Career Forwardwith certification training in the latest technologies.Start your trial today
Network Architecture

From novice to tech pro — start learning today.