Does anyone have experience with Nimble replication and WAN optimizers such as Riverbed or Silver Peak?

[ninmikes98]

We are seeing very little traffic reduction with our Silver Peak WAN optimizer.  Silver Peak believes it is because the data is already encrypted and compressed.

Does anyone know if this is the case, or if there are any best practices relating to using a WAN optimizer?

Thanks!

Michael

[edfops66]

Correct from my experience as well, unless you can turn off encryption and compression and let SilverPeak handle those aspects (which would be a bit wasteful or hard to do for an existing Nimble array) there will be close to no reduction on the SilverPeak.

This was true for me as well with other products that also do their own encryption and compression (Veeam is a great example).

[dbeaulieu132 · Accepted solution]

My Experience with Riverbed was no good. The Optimization result was around 45 to 50% but the throughput was not as expected. Turnout after talking with Nimble support, Riverbed was blocking the capacity to auto adjust the TCP Windows. I have add a pass-through rule in the Riverbed and now I have a 0% optimization but Everything is synchro in 35 minutes instead of 1h15.

So My recommandation is dont use Riverbed or any optimizer.

[paul_napoli]

In theory, a WAN optimizer WAN should find very little "room for improvement" with Nimble SmartReplicate.    In general, WAN optimizers tend to work their magic using 3 "moving parts" to minimize the impact of network latency.... 

1.  Data Reduction (aka Dedupe), which Nimble SmartSnap and SmartReplicate already do,  (this is the biggest value-add of a WAN optimizer)
2.  Transport Streamlining which uses the optimal TCP packet size for the WAN link without having to constantly determine it dynamically
3.  Application Streamlining which may not be applicable since Nimble > Nimble replication doesn't require support for a myriad of "chatty" protocols

The previous posts prove the theory correct.