Some of you may have just noticed, my blog is now missing the Amazon associates crap. Finally got around to removing that after they stopped paying Colorado residents. Not that I was making much anyways, kinda sad they couldn't come to an agreement though.
Anyways I've been now in Newark setting up a big SharePoint server farm now, and have had to answer the tough question "Just how big should the farm be?" from here we start talking about availability, performance, etc...let me make this short and sweet, start with the basics:
- 1 SQL Server
- 1 Index Server
- 2 Web Front Ends load balanced with Search Services behind them
- 1 "Everything Else" Server
This alone can handle a few hundred users simultaneously (assuming modern hardware), next steps moving up as resources permit...
- Cluster the SQL Server, remember this is your data, high availability very good
- Split out apps even further, SharePoint is good at self-balancing all but the index server, if you notice a particular app "pegging" in performance monitor time to buy a new server to split out the loads.
- SQL Server is typically very heaily hit, some of the first things you'll do is break out more storage engines on separate hard drives, offloading the bigger sites to these hard dries to further distribute the load, ultimately you may be doing so with a new SQL Server
Just break out your performance monitor tools when the users start to complain and be a good sleuth...analyze what is "pegged" (Disk? Memory? Network?) and start working accordingly. Infrastructure guys in particular are good at this, use them well in your organization to get the vital clues you need... |