Thursday, May 28, 2009

Data Center Space in the Northeast

Yesterday I wrote my personal blog for the week, so today's blog is a return the typical issues of a CIO - building and renting data centers in the Northeast.

I was recently asked about a report being published this week on health care data center costs in Boston compared with other U.S. metropolitan areas. The report claims that, although huge opportunities exist for data center providers to house growing amounts of health care data, the high costs of running a data center in Boston and other Northeast cities will drive providers to house their data in low-cost areas in the Midwest.

Here's my view of the needs of healthcare CIOs for data center space in the Northeast.

My sense is that most IT organizations are embracing virtualization which reduces server space needs. However storage needs are increasing 25% per year, consuming more space.

Thus, the demand for data center real estate, on average, will experience modest growth. Healthcare data center space is not likely to cost much more than typical data center space.

The choice of build/owned verses co-located data center space is mostly a function of network connectivity and capital availability.

The breakout of BIDMC data center expenses, not including the operating and capital needed to suppport our applications is roughly as follows:

Space and Utilities 15%
Salaries & Benefits 29%
Elect Wiring/KVM/LAN Cabling/Racks/Etc. 5%
Storage 18%
Tape/Backup 4%
Servers 14%
Monitoring Software 3%
Network 1%
UPS/PDU/CRAC 11%

Moving the data center to the Midwest MAY save on space cost and would likely save on energy. Most of the other costs would be the same whether we were located in Boston or Lincoln, Nebraska. Some costs may actually be higher in the Midwest as we are within driving distance of engineering support centers for some of our OEM's such as EMC. We currently have a favorable space rental rate compared to the square foot cost of hosting facilities in Boston. Consequently, savings for us is not going to be as great as it may be for other companies who use hosted space. Moving the data center MAY also reduce computer operations salaries, but that's questionable and assumes the labor pool is readily available in the Midwest.

I estimate the annual savings for space, utilities, and salaries, in our case, would be $150k to $300k per year. Offsetting this would be increases in wide area networking expenses and, given the distance, there would need to be dual paths for redundancy which would increase the expense. There would be a significant one time expense for the relocation cost. The combination of these could easily overwhelm any projected savings.

Relocating the data center would also require a major project that would absorb much of IT's time thereby causing us to halt or slow down other, more pressing requirements of the Medical Center. There would also be a risk element in that we would be placing our IT assets in the hands of a party we do not directly control except through contract terms and conditions.

Unlike some companies that have national or international presence, we are limited to metro-Boston. The incentives to relocate the data center to the Midwest are not compelling for us.

I predict Boston and Northeast data center space will continue to be well utilized.

4 comments:

Bernz said...

I just moved two high-security, high-availability datacenters from Boston to the midwest. About 1000 physical servers.

On the books, this looks to save us about $500k over a few years. The process took about 1 year and had 4 full-time IT employees attached as well as many hours from operations and other groups as it affected their workflow. Still, bottom line savings were there.

I do think the labor pool is better in Boston for systems engineering/security than it is in the midwest, but I'm incredibly biased.

As far as the Wide Area issues now that the machines are two thousand miles away, I'm proud of my solution. Internet Line with VPN makes a cheap line and only about 10ms slower latency than a traditional WAN line. Using various kinds of compression, optimization and virtualization, data moves quickly over this line. Resources halfway across the country are readily available. If the Internet line goes down we do have a slower, cheaper backup, so we wouldn't be down for the count.

The "midwest consolidation" can work and save money. But in my experience, you're 100% right: it takes time and effort that might be better spent doing other things. And I like our Boston Datacenters and their personnel. That's worth something to me, as intangible as it may be.

I think we'll see less "data centers" and more "cloud and utility computing". That's not to say that data centers are going away, but just that the line between "where my data lives at rest" and "where my data is now" will become increasingly blurry.

Storage does increase, but the space that storage uses decreases. An 18GB SCSI drive and a 300GB SCSI drive take up the same room and use the same power. There are really interesting storage systems these days that are built to continually migrate their shelves, so you can physically have the data going into its larger, space-saving brethren on a rolling basis. In the same space where I had 15 TB 4 years ago, I now have 80 TB.

james.cooley said...

Anything likely to come of the idea of floating data centers, cooled by sea water? The North Atlantic is pretty darn cold and right next door.

krubin9 said...

I run a digital storage business for Iron Mountain, focused on healthcare.

We have data centers 220 feet underground where we enjoy natural benefits of cooling and low utility costs plus favorable real estate and scalability leverage.

Our data centers are used for secondary storage, back-up and archving of medical images and other data. We routinley save hospitals 25-50% of their storage cost by operating storage as a service.

Hospitals can then re allocate IT resources and capital to more strategic purposes than secondary storage

Ethan Bechtel said...

Thanks for the breakdown of costs. I'm interested in seeing how much of a cost difference there is within the Northeast (urban vs. rural). Salaries/benefits in Boston are, of course, much different than in Burlington, Vermont. Space and Utilities may vary as well.