Servers
Fujitsu makes Windows Server 2012 see double
Small and medium-sized businesses want high availability (HA) capability for their core applications, just like the largest enterprises do. But server clustering can be intimidating and expensive, so Fujitsu has rejiggered some of its Primergy hyperscale dense servers to support the clustering of Windows-based servers inside of …
wtf
So they are putting a HA cluster in a SPOF??? What are they smoking??? Making a very complicated setup (means much more change to f*k-up) in a SPOF rig? What's the point? So you can check "we have an HA cluster"? Or is w2k12 so unreliable you always need two? or....
Re: wtf
If i'm reading it right, logically it isn't too bad.
The servers are effectively independent of each other, and most people would plug the power into separate feeds. True the shared disk is a single point of failure, but it is in any standard cluster (unless you start going down the resilient SAN route).
Not saying I'd go for one, but it's also not a terrible idea for a SMB that wants a bit of resilience.
Re: wtf
Not really a spof, unless you also consider a mainframe a spof. Multiple psus, separated and duplicated internal components etc.
Re: wtf
Actually, Windows clustering has come a very long way since the days of NT4 when the cluster service caused more failures than anything else. Windows is rock solid stability-wize these days (assuming you know how to configure it and have a good hardware/firmware/driver supplier) particularly if you're using the DC edition.
Re: wtf
Apparently Cillit Bang is the SPOF of a one box cluster....
Erm, SATA is fine ... but not dual-ported
Actually SATA is supported with Storage Spaces, but it cannot be shared / multiple-connected like SAS can. Basically the drives need to be "dual-ported" (connected to both controllers). It's a small thing, but we like accuracy, don't we?
Storage Spaces then manages having the disks online on one node or t'other.
