1Gb host becoming saturated writing to 10Gb host.

bfrericks · ‎05-25-2012

Hello,

I have a situation that I am not 100% sure how to resolve. I have a Windows host, Host A, that has a 1 gig connection to the core and is part of a Windows cluster. I have another host, Host B, that has a 10gig connection to the core. Host B is essentially a NAS and a target for disk-2-disk backup. Host A is a SQL Active/Passive cluster. The problem I have is when Host A runs its backup dumping to Host B it saturates Host A's link and causes the cluster to failover becuase SQL/Windows Clustering thinks it has lost network connectivity. I have enabled flow control on both ports and ensured it is enabled on both interfaces or both hosts. The only solution I can come up with is to setup some type of rate limiting. Does anyone have any other thoughts?

Thanks in advance!

smehrnia · ‎05-25-2012

Hi,

can you tell how did you do the rate limiting and where u enforced it?

thx.

Soroush

Hope it Helps!

Soroush.

bfrericks · ‎05-25-2012

I have not rate limited yet, I am wondering if that is my only option. I have done nothing other than enable flow control.

smehrnia · ‎05-25-2012

I think its best to define your Back Up traffic flow from Host A (source destination port #) and use a policy map to police that particular flow from host A, while letting other flows freely use the remaining bandwidth.

i donno if you've ever done that b4, but it sounds reasonable. lemme know if you need further help wth it.

plz Rate if it helped.

Soroush.

Hope it Helps!

Soroush.

hobbe · ‎05-25-2012

Hi

Yes you could do rate limiting on the link exiting the switch/es that Host A cluster is on.

Or you could open up a interface more and use that for backups only.

I remember there is a possibility in some backup software to choose how much bandwith to use both for backing up and restoring.

you do not happen to have one of them ?

Good luck

HTH

Joseph W. Doherty · ‎05-25-2012

Disclaimer

The Author of this posting offers the information contained within this posting without consideration and with the reader's understanding that there's no implied or expressed suitability or fitness for any purpose. Information provided is for informational purposes only and should not be construed as rendering professional advice of any kind. Usage of this posting's information is solely at reader's own risk.

Liability Disclaimer

In no event shall Author be liable for any damages whatsoever (including, without limitation, damages for loss of use, data or profit) arising out of the use or inability to use the posting's information even if Author has been advised of the possibility of such damage.

Posting

Sounds like the cluster keep alive is being lost. If correct, unclear from your description where it's being lost. For an outbound dump, 1st point of possible congestion would be the host's NIC. One would think Windows would be smart enough to not delay a special keep alive, but perhaps Windows needs to use its QoS scheduler.

If the congestion is on a network switch, which doesn't seem to match your description, normally you would use QoS to guarantee the keep alive isn't impacted by a bulk high-rate flow.