The most resilient storage in the business doesn’t just sit there and take reads and writes in a corner. A SAN or NAS array is made up of disks that fail, and requires network to be administered and storage to be administered. So you need to calculate personnel costs as well. I’d be very surprised if you could find someone reliable who could manage your storage in the 20TB range for $10k per year blended salary rate leftover for ‘admin’ in your model. Unless you trust your colo folks to manage your storage as well, which I would find *very* dangerous – that’s not what they do, and they are not generally subject-matter experts in this area.

Next, the best guarantee in the business for utilization (shameless plug here for my company – Pillar Data Systems) for SAN or NAS-attached storage is 80%. Most of our competitors average 40% usable space. So your 20TB array has only 16TB max (if you buy from http://www.pillardata.com) usable space – you’ll have to buy 25TB from us, or 50TB from our competitors, to give you 20TB usable space.

You’re also forgetting growth – data doesn’t stay at a static level over 4 years, or the business is dead, and it doesn’t matter what business that is – transaction records pile up, emails pile up, stored customer data piles up, and most data retention requirements for SOX / HIPAA / Data Protection Act are 7 years now.

Add up footprint, power, network, air con, management, support agreements, management overhead (i.e. configuring and managing things touching the service), data growth, etc etc etc and it is a far more expensive play to host your own storage than to go for the Cloud until you get to fairly high data requirements – say over 50TB usable.

But that’s not the reason pay for local storage – customers want (or need) the capability to have their data local to their environment for a variety of reasons. Some of our customers can’t even let a failed Hard Drive leave their data centres, because of data security requirements. Some have audit requirements (especially financial services) which mean that the data must physically be present in a given location and can be audited from there. Some want to be able to dynamically grow their storage quickly, and thus want to have physical management of it locally. Some just don’t want to pay for big pipes for all their users, and instead want to have big pipes only within their infrastructure and within their buildings. And some don’t trust the security, or don’t want to trust someone else’s security, for their business critical data.

YMMV, but that’s my experience.

This is the same question I hear all the time. Isn’t this just SSP 2.0? Isn’t this just the same old stuff wrapped up in a new word (cloud)? End users and analysts are right to be skeptical, and can be forgiven for dusting off the old “dumb disk” strawman as a way to start the conversation.

The truth is that major companies are using public cloud computing and storage services today. They’re not moving everything to it, and they’re giving it a thorough examination before taking the plunge, but they’re doing it. And they’re not just doing it because it’s cheaper (which it is) but because it offers them compelling benefits:
- They can get their data off-site immediately for compliance reasons and can distribute it geographically from there for DR
- They can offload the effort and focus of managing the part of their storage capacity that isn’t their core concern so they can focus on their real business
- They can move certain content outside the firewall to a programmable storage platform, enabling new services like collaboration and content distribution

No one has a crystal ball and can see which service providers will succeed and fail in this market. The telcos did OK after the collapse of the hosting business and SSPs, but they didn’t walk away holding all the marbles. In fact, that market is shared by a large number of mid- and large-sized companies. The winners were determined by their ability to offer a differentiated product or an integrated stack or both. I think the same will happen here – “cloud” services like compute and storage will evolve, cast off that silly name, and thrive (as you say) based not on their price point but on their unique value-add.

One great value is integration of the whole stack, from client to hardware and software to operation. This enables all sorts of features that a Lego-brick approach can’t deliver and (yes) allows a lower price point. Looks like a winner!

And no, I don’t think you’re evil. You just like stirring things up, and we all love you for it!