SharePoint 2013 Request managment Verus having a load balancer

Question

the higher management ask me to provide a high level description about what are need in case we want to define two SharePoint servers inside the same Farm . so if one SharePoint server is down the other server starts. I was reading about this and seem in general I can define a hardware load balancer that can check if one server is down to route the requests to the other server. But on SharePoint 2013 there is something called Request Management , where I can define routing rules and I can define the routing rules to be based on the server health.

So in my case can I use one of them (load balancer or Request Management) or I have to use both of them to implement the requirements of having 2 SharePoint servers inside the same farm, where in case on server went down the other server will handle the users requests.

Answer 1

You'll need both to give the highest level of uptime.

SharePoint Request Management will ensure the request lands up at the best server within the farm to handle the request. If a server is down then Request Management will not send the request to that server and so on. All this only kicks in once the request lands up in the farm via your front end servers.

Your users need to connect to the front end web servers somehow and if one is down and you don't have something intelligent enough in front of the farm to know that one front end server is down some (or all) of your users will experience downtime. Your choices to balance inbound traffic across 2 FE servers are DNS round robin, Windows NLB and a hardware load balancer.

The best one to use is a hardware load balancer because they work on a lower level to SharePoint RM and typically will only balance traffic to your available front end servers. We configure ours to check if the server is up (ping) and if IIS is up and only then send them FE's traffic.

We have used Windows NLB successfully to balance SharePoint as well but removed it in favour of a pair of F5 LTM devices so we can do extra checks and we use the F5 devices for multiple applications. Windows NLB isn't smart enough to know IIS or SharePoint is down. It only does server level checks.

http://technet.microsoft.com/en-us/library/jj712708.aspx has lots of good information about how request management works. In all the scenarios they reference a hardware load balancer.

Hope this helps.