How many datacenters is too many?

How many datacenters is too many?

Summary: Gartner recommends that two datacenters per continent is the optimal configuration

SHARE:
TOPICS: Data Centers
3

Though they do provide themselves an out by stating that that their recommendation of no more than two datacenters per continent is an ideal model and that there will often be some variation, the analysis firm Gartner came out and strongly recommended that the continental  twin sites approach should be the standard goal for IT operations. This means two sites each for North America, South America, Europe, Africa and the Asia/Pacific region.

Their statement seems to be based on the issue of dealing with the haphazard growth of datacenter operations due to business acquisitions and changing business needs. It reflects on how IT often laments the fact that maintaining multiple sites and systems hinders them when they need to rapidly respond to changing business needs. The goal of the twin datacenter topology is to reduce operational and deployment expenses, while providing a more efficient datacenter architecture. But there does seem to be a huge number of situations where focusing on this type of architecture seems very impractical.

The report acknowledges that there are many common situations where multiple datacenters are a business requirement, such as, designing for disaster recovery and business continuity, where the epigram “one is none, two is one” is often repeated. And of course the bottom line issue where the cost of moving to this architectural model is not economically justified.

While the concept is an interesting one, it does come across as more of an intellectual exercise than a practical approach to datacenter topology and design. The concept has a limited subset of datacenters to which it is completely applicable; those where the business runs all of their IT datacenter operations internally or uses collocation datacenters where the facility is operated by a third party and the IT equipment is owned and operated by the business.  And while it does suggest dual sites per continent, to a certain extent a large business would be putting all their eggs in one basket, if you consider that the dual pairing represents a single redundant infrastructure.

On the surface, it is difficult to envision a broad range of IT situations where this topology is a practical goal. If a business is completely re-architecting their infrastructure for efficiency of service delivery and optimization of expenses, there are a broad range of choices, ranging from modular datacenter deployment to hybrid-cloud services, to completely outsourced datacenters that present themselves as more practical options that this “two per” design.

If the goal is consolidation of multiple resources, this does seem a laudable target. But in areas such as Europe and Asia, the technical issues can often be trumped by the geopolitical nature of how datacenters and their contained data need to be treated, as well as the technical issues of bandwidth and site suitability that need to be addressed. So while the concept is an interesting intellectual exercise presented by a respected group of analysts, its overall practicality seems suitable for only an extremely limited set of circumstances.

Topic: Data Centers

Kick off your day with ZDNet's daily email newsletter. It's the freshest tech news and opinion, served hot. Get it.

Talkback

3 comments
Log in or register to join the discussion
  • listen up all you One Global Datacenter nuts

    Your wrong, get a clue.
    greywolf7
    • Yeah, we should do what Gartner says

      Not sure why anyone would care what their so called experts recommend.
      otaddy
  • For a while maybe

    There is an apochryphal story about Thomas J. Watson of IBM announcing a reorganization of the sales force. Some years after centralizing sales and marketing in Armonk, IBM was decentralizing it, pushing responsibility and decision making authority out to the regions.

    A trade magazine reporter asked him if the move to centralization had been a mistake. Watson's answer was that neither centralized nor decentralized management were wrong. It was just that no matter which one you had, you had to smash it every so often and go the other way, because otherwise things got strange. Little fiefdoms sprung up, guys who weren't even on the org chart were de facto in charge of places, policies and procedures outlived their usefulness and were strangling the business.
    Robert Hahn