Take the 2-minute tour ×
Server Fault is a question and answer site for professional system and network administrators. It's 100% free, no registration required.

What factors into choosing a distribution server for transactional replication? In our topology, we've always had the distributor reside on the publishing server. We rarely generate snapshots and performance is good enough, so this is okay for us today. One of our instances is moving to a windows cluster with availability groups, so we need to move the distributor off for resilience/symmetry.

Right now our two choices are to use a server physically close to the publishers, or our single subscription server. Our publisher is in our main office, and our subscriber is in a colocation facility off-site which our ISP runs. We have a pretty good line to it. The reason we're even considering the latter is to save work and licensing costs.

share|improve this question
add comment

1 Answer

up vote 1 down vote accepted

I'd keep the distributor as close to the publisher as possible. My reason for saying this is that one of the conditions for a log reuse is whether or not all of the records in the VLF have been processed by the log reader. If, for some reason, your line goes down between your publisher and the distributor, your log is at risk of filling up.

That said, I have to question your statement of moving the distributor away from publisher. I've run clustered instances that house both publisher and distributor without any problem.

share|improve this answer
    
I should have been more precise with my words. This isn't a failover cluster instance, it's an AlwaysOn availability group. If the node hosting distribution goes down, replication stops. Wouldn't replication stop if the line between pub/dist goes down as well? That would be more of a concern to us than the log filling up, I think. –  Stuart Branham Sep 14 '12 at 21:50
    
It's all what's important to you, but in every environment that I've ever worked in, the pub was the business database and replication was set up in support of reporting, ETL, etc. So, the pub is the most important db in the topology. But to answer your question, repl'n goes down if either of the pub-dist or dist-sub links goes down. I haven't read up on AGs and replication, but if you can put the pub and dist in the same group, I would do that. –  Ben Thul Sep 14 '12 at 21:58
    
Thanks! Very helpful. –  Stuart Branham Sep 14 '12 at 21:59
    
By the way, this page (technet.microsoft.com/en-us/library/hh403414.aspx) seems to suggest that the distribution db can't be part of an AG. That's... unfortunate. –  Ben Thul Sep 15 '12 at 4:26
    
Unfortunate, but makes sense at least for simple cases. You don't want the server that you're allowing to go down be a distributor. –  Stuart Branham Sep 17 '12 at 18:39
show 1 more comment

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.