What is acceptable latency over two active BGP WAN links?

[ah0672863]

Hello World -

We have multiple remote offices with redundant routers, circuits, and switches.  Circuit-A is connected to Router-A which is connected to Switch-AB.  Circuit-B is connected to Router-B which is connected to Switch-AB.  WAN circuits are with a single ISP and Active/Active.  eBGP is WAN protocol with multiple path routing enabled.  iBGP is running between the two routers over a dedicated link on the routers.  HSRP is defined on the LAN.

Some users were complaining about performance...We took a look and noticed that the latency over one path is about 36-45 msec higher over the other. Say there are two people sitting in the same office on opposite sides of the office.  Johnny-A traffic flows up and out over Router-A and Jane-B traffic flows up and over Router-B both destined to Location C.  Whenever traffic went over Path-B we knew the users were going to have a poor experience.  

Is there a standard that speaks to WHAT IS an acceptable level of variance between the two circuits of similar media and performance?  We are trying to stay away from it depends on the application.  I think 36-45 msec within North America is too high.

Thoughts?

[Joseph W. Doherty]

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 wha2tsoever (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

What might be a rough rule of thumb, for WAN provider latency could be up to about twice the direct path fiber latency for the distance between sites when you're not jumping across oceans.

An additional 36-45 ms latency seems high, if it's not due to your traffic or different CIR commitments.  You see this delta even if both paths are unloaded by your traffic?