A tale of an RFP gone wrong

… sadly this appears to be true. Specifications were given, and we met the requirements, which entailed a demonstration of a particular level of performance over NFS. In case you aren’t sure, we demonstrated a sustained 1GB/s over NFS between 2 boxes over 10GbE last year. There aren’t too many companies that can do this. Our results were with RAID6 storage target, and an NFS client with small RAM size. Total read and write size each was much larger than either system memory. This wasn’t a cache test, data was going to and coming from disk. This stage set, a very high data rate over NFS was required for the RFP, which we met, quite easily. Our competitors … not so much. They never performed the measurement we did.

Apparently, our competitors went to their RAID vendor, and got some nice RAID0 numbers which they appear to have substituted for the NFS performance over the 10GbE. I really feel bad for the customer here. His purchasing person has bought a machine he may not be able to use, because it won’t give anything close in performance to the required specs. We know. Good performance is hard. Excellent performance is very hard. Anyone can throw together enough marketing numbers to fool a purchasing agent not sufficiently knowledgeable. This is sad.