The other situation of a LOB wanting to get local cheap disk has been around for a decade, ever since SANs and pooled storage started popping up. Once you tell them that the local DAS has to be backed-up, DR protected, encrypted, end-of-life migration, optimized, secured, and protected for long-term discovery, they soon get the message and let you grow the pool again. Some don’t learn, then they find out how hard some basic functions can be with localized disk.

As you mention, the $/usable SAN port can get skewed very quickly when you include all the cabling, patch panels, ISL ports etc. Lets don’t let that metric turn people off from SANs all together.

What about intelligent cards in a director for some fabric service (encryption, compression, CDP, CRR, etc)? These intelligent cards are quite expensive and have no usable ports for all effective purposes. Yet when dealing with a customer’s procurement analyst in an RFx scenario, they don’t want to hear the techno jargon. They want price / # ports.

I won’t even get into patch panels and trunks, but I’ll say this. I’ve customers buy $2M worth of directors and spent $3.4M cabling them up.

This is exactly why economists like to use the phrase “ceteris paribus”…..I wish it were more widely accepted in IT disciplines.

I understand the large BI, analytic solutions that are sold as an all inclusive “An appliance” like Teradata, Nettezza, Datalegro,etc but Microsoft is now pushing DAS for not only Exchange but SQL in some cases too.

In Microsoft’s case I think it’s a tactical move such that capabilities that are typically provided by enterprise storage (local and remote replication for instance) are now being provided by their application. And or it’s another way to shield a thick resource hungry application (dedicated spindles = no neighbors). And finally they are scared to death that Exchange running as a VM is gaining in popularity. This way they finally box out VM.

IGG-09172003-01 C. Stanley 17 Sept 2003