Replication over slow WAN link

[cackerson]

I have replication setup to run over our WAN. I use seeding and am able to get the replicas to work successfully sometimes.
But out Exchange server almost always fails with a "Error: Client error: A connection attempt failed because the connected party did not properly respond after a period of time, or established connection failed because connected host has failed to respond Unable to retrieve next block transmission command. Number of already processed blocks: [9565"
We have a 1.5 Mb link and it is throttled to 600 kbps. Even when it isnt throttled, if the amount of data to replicate is large it will often fail.
Is this too small of a pipe to make replication work?

[Gostev]

Bandwidth is never an issue (as long as it is enough to replicate the changes), it looks like the connection itself is not stable in this case.
But yes, of course 1.5 Mbps is quite slow connection overall for 2012... even at home, I have 30Mbps up/down for like 20 bucks a month

[dellock6]

Anton, ask Gianluca about Veeam customers in Italy, there are even worst scenarios than 1.5 mbits. We talked about it in our last webinar

About the error, I think too 600 kbps is really really too slow, probably some activity times out during the process. Craig, is this line dedicated to replica activities at least during replica taking place? Any other traffic going through this connection?
Also, when it succedes, how long does it takes? If the hours it takes goes well beyond your desired RTO, your best chance is to upgrade that line...

Luca.

[cackerson]

Luca,

The line is not dedication to just the replication. There is often high utilization on this line.

One job we have replicates succesfully most of the time is 200 - 300 MB and takes about an hour.

One job runs successfully %50 of the time is about 1 GB and takes an hour and a half.

Our Exchange server rarely replicates succeffully. The last time it did it was 10 GB and 33 hours. Usually it fails even after small amounts of data have replicated.

[tsightler]

I can almost guarantee that 1.5Mb is not enough to replicate an Exchange server no matter how small, unless you have WAN acceleration involved (real WAN acceleration with caching engines, not just compression/optimization).

For reliable replication you will need to prioritize replication traffic to no be drowned out by other traffic, basically to guarantee at least some minimal bandwidth to replication. Without this contention with other traffic are likely to make the replication fail.

[dellock6]

Agree, that's why I asked for other traffic going through the line.
Exchange server is known to be heavy to replicate since its disk activity creates much CBT changing, and this must be replicated.
As Tom suggested, check if you can:
- use some wan accelerator to increase the effectivness of that line
- use some QOS on the line to prioritize replication.

[cackerson]

Replication Success!!!

We had a slow link to our DR site. It was only 1.5 mbps. I setup replication, seeded the VMs are and followed all the recommendations about proxies etc. I throttled the link to 666 kbps at first and had very little success replicating. I was replicating machines with large amount of changes like Exchange Server (3-5 GB). It would never work. I increased the throttling to 1000 kbps and was able to get my machines with smaller amount of changes like my Domain Controllers to replicate (200 - 400 MB).

This month we upgraded the link to 3 mbps. I seeded the VMs with large changes like Exchange and set the throttle to 3000 kbps. I was able to replicate successfully over the weekend. 4 servers with (5 - 7 GB) of changes all replicated successfully.

I know Veeam says that there is no minimum requirement for bandwidth, but for us we needed more bandwidth to replicate our servers that have a large amount of changes.

Veeam rocks!