Most of the customers need self manageable SharePoint farms. Something they “can put hands on”. Not large “sacred” farms in distant data centres. Operated by high priests of the SharePoint church.
How to serve this organizations? Is it at all feasible to do this? I started thinking along these lines, way back in Nov. 2010. I knew it is feasible to go against the dictate of the “common wisdom” and not use Virtual Servers. I have searched for a while and found an excellent hardware platform in the HP C3000 blade enclosure.
The result is surprisingly capable SharePoint farm which is also highly manageable. Almost like your own “data center under the desk”. It certainly looks and feels like that.
This Platform Architecture is not redundant as is uses DAS in the same enclosure. But of course it can be configured like so.
One can configure SQL SVR redundant installation by adding two or more of these “pod’s” and Shared Storage “in between”. I do recommend this “repeatable pod” concept of scale-ability and resilience over horizontal expansion (adding more servers) which very quickly leads to maintenance issues.
It is much better to have 1..N small(er) farms than a single large one.
Just very recently (Sep 2014) I have found out, good DELL engineers had a similar idea. And not very different implementation based on blades. intentionally or not they also adopted the key principles:
- Small(er) Hardware (when compared to DC Racks)
- No Virtual Servers
They had of course much bigger resources and developed the idea further. Although it seems I was first 🙂
Here is a bit more detailed view of what, I have delivered on an HP C3000 unit.