Login | Register For Free | Help
Search for: (Advanced)

Mailing List Archive: nsp: foundry

BigIron RX LAG load sharing behaviour

 

 

nsp foundry RSS feed   Index | Next | Previous | View Threaded


dana at zeroloops

Oct 29, 2010, 1:40 AM

Post #1 of 2 (892 views)
Permalink
BigIron RX LAG load sharing behaviour

Hi all,

I have noticed a little problem with our LAG configurations in our
Bigiron RXes. The problem description is as follows:

We have configured a LAG of two or more ports. From time to time we are
getting a high utilization port message from our monitoring system on
one of the two/more ports in that LAG. Thus, it seems that the software
does not share the traffic evenly amongst the available ports.

After reading the manual, I understand that there is no way we can
manipulate the load-sharing method of our LAGs. BigIron RXes only
load-share using HASH-BASED method. Although Foundry/Brocade claims that
this method should do the job properly, it is not the case in our
network. Let me give you some lines to show the problem:

lag "trunk-1" static
ports e 2/7 to 2/8
primary-port 2/7
deploy

#sh int e 2/7 | i util
300 second input rate: 155247818 bits/sec, 39455 packets/sec, 15.90%
utilization
300 second output rate: 25029300 bits/sec, 11300 packets/sec, 2.61%
utilization

sh int e 2/8 | i util
300 second input rate: 189413832 bits/sec, 44085 packets/sec, 19.36%
utilization
300 second output rate: 482937687 bits/sec, 79502 packets/sec, 49.05%
utilization

As you can see above, the output rate (from BIRX to an FWSX) has a ratio
of 49% vs 2%. However, if you look at the input rate (from FWSX to
BIRX), it has more or less the same amount of traffic 19% vs 15%.

My question is now, is there anyway we can do in the BIRX to get the two
ports more balanced? I hope some of you has experienced the same
problem and could share the solution/trick with me.

Thanks in advance.


--
Dana


_______________________________________________
foundry-nsp mailing list
foundry-nsp [at] puck
http://puck.nether.net/mailman/listinfo/foundry-nsp


bdflemin at gmail

Oct 29, 2010, 6:55 AM

Post #2 of 2 (879 views)
Permalink
Re: BigIron RX LAG load sharing behaviour [In reply to]

I don't believe there's a way to reconfigure the BigIron; however,
you'll likely get even load balancing if you can create two separate
sessions / flows. Is that possible in your situation?


On Oct 29, 2010, at 3:40 AM, Dana wrote:

> Hi all,
>
> I have noticed a little problem with our LAG configurations in our
> Bigiron RXes. The problem description is as follows:
>
> We have configured a LAG of two or more ports. From time to time we
> are getting a high utilization port message from our monitoring
> system on one of the two/more ports in that LAG. Thus, it seems that
> the software does not share the traffic evenly amongst the available
> ports.
>
> After reading the manual, I understand that there is no way we can
> manipulate the load-sharing method of our LAGs. BigIron RXes only
> load-share using HASH-BASED method. Although Foundry/Brocade claims
> that this method should do the job properly, it is not the case in
> our network. Let me give you some lines to show the problem:
>
> lag "trunk-1" static
> ports e 2/7 to 2/8
> primary-port 2/7
> deploy
>
> #sh int e 2/7 | i util
> 300 second input rate: 155247818 bits/sec, 39455 packets/sec,
> 15.90% utilization
> 300 second output rate: 25029300 bits/sec, 11300 packets/sec, 2.61%
> utilization
>
> sh int e 2/8 | i util
> 300 second input rate: 189413832 bits/sec, 44085 packets/sec,
> 19.36% utilization
> 300 second output rate: 482937687 bits/sec, 79502 packets/sec,
> 49.05% utilization
>
> As you can see above, the output rate (from BIRX to an FWSX) has a
> ratio of 49% vs 2%. However, if you look at the input rate (from
> FWSX to BIRX), it has more or less the same amount of traffic 19% vs
> 15%.
>
> My question is now, is there anyway we can do in the BIRX to get the
> two ports more balanced? I hope some of you has experienced the
> same problem and could share the solution/trick with me.
>
> Thanks in advance.
>
>
> --
> Dana
>
>
> _______________________________________________
> foundry-nsp mailing list
> foundry-nsp [at] puck
> http://puck.nether.net/mailman/listinfo/foundry-nsp

_______________________________________________
foundry-nsp mailing list
foundry-nsp [at] puck
http://puck.nether.net/mailman/listinfo/foundry-nsp

nsp foundry RSS feed   Index | Next | Previous | View Threaded
 
 


Interested in having your list archived? Contact Gossamer Threads
 
  Web Applications & Managed Hosting Powered by Gossamer Threads Inc.