[Linux-cluster] Low cost storage for clusters

[Brendan Heading]

Riaan van Niekerk wrote:

> 1 The reasoning behind it being similar to why software RAID 1 (with a 
> big, idle Xeon behind it beats the pants of dedicated hardware RAID 1 
> with its relatively underpowered ASIC.

Surely though that doesn't apply in cases where the server is performing 
CPU-bound work.

The servers in my case are used mostly for running large parallel builds 
of projects comprising multiple millions of lines of code. As well as 
the disk activity for shifting the source off the disk and writing back 
the compiled objects, your CPU is spending all it's time doing the 
compilation work. If it has to stop compiling in order to do TCP/IP as 
well, well that's bad.

Basically in that environment you can never have enough CPU.

> 2 Also, the TCP offload engine (TOE) in a hardware initiator should give 
> it an edge, but most server NIC vendors are starting to bundle TOE in 
> their regular NICs, negating this performance benefit of dedicated iSCSI 
> HW.
> 
> If possible, get a demo QLA iSCSI HBA and test the two.

Good advice..