For the past umpteen years there has been a trend towards consolidating data centers – from many sites to as few as possible. Or if building your own data center wasn’t plausible, consolidating from owned sites to leased, hosted or collocated sites became the option.
Arguably there are many valid reasons to consolidate, and in most cases there can be significant savings to be had, which I won’t go into here. However, consolidating for the sake of “less is better” may not be the best decision, especially when we apply the right metric to the issue – value to the business.
Value from an IT perspective too often gets defined as cost savings or operational efficiencies (which is a politically correct way of saying “cost savings”). But from a business perspective “value to the business” equates to agility, performance and service continuity.
The problem here is that you could argue both sides of the equation and have valid points from both perspectives. Consolidation can make IT more costs effective, and it can make IT more agile while improving performance (often through tech refresh cycles). But oftentimes the achilles heal of these projects are site or geographic specific issues revolving around latency, redundancy and offline support, and those arguments always favor the business case of keeping dispersed compute resources where they are.
But what if you could get the benefits of consolidation, while still satisfying that value to the business metric?
An obvious example of this might be an organization with multiple small manufacturing plants, or a retailer with multiple stores, each supported by common services provided by a consolidated IT environment. While 90%+ of the workload could easily be supported using remote services, if the central site had a failure anywhere in the IT path (network, servers, storage, applications, etc), operations at those remote sites could be compromised. If multiple sites are impacted simultaneously the lost opportunity or transaction costs could be significant.
To solve this issue many of these sites leave behind “some” IT in the form of small server rooms – but these are often managed by partial FTE’s and are not maintained and updated using the same process or methodologies as central IT, which over time increases costs and increases complexity (platform again, OS releases, tech refresh cycles, etc).
There is an emerging market for “fit for purpose” or Micro Data Centers; IT environments designed to be easily installed, self contained, scaleable and remotely managed. Designed right these mDC’s (for lack of a better term) could be as small as a single rack, but would contain servers, storage (SSD or HD), switching, UPS (if needed), and support their own cooling. With current and next generation technologies some pretty significant capacities could be attained even within one or two racks, and they could be environmentally supported in an office area or a shop floor. With a standardized OS/VM and a reasonable network, almost all maintenance could be handled remotely as well. And as a side benefit – varying tier levels could easily be designed in, based on business needs and budget requirements.
Food for thought…. all comments are welcome of course….