Mirrored Data Centers - Three Approaches
I emphasize final form, as the draft version of the report released in September of 2002 caused an uproar in the financial industry. The reason? The original report mandated that backup sites be situated at a minimum distance from the primary location (if I remember correctly, in some cases this would be in excess of 100 miles). The response was predictable, as financial industry participants made the case that creating mirrored sites so far away would be financially unattainable, not only because of the cost of installing the necessary infrastructure, but because it would impact other areas including human resources. The committee relented and watered down many of its original proposals in the final form. Although the recommendations put forth in this paper were targeted toward the financial industry, other industries have undoubtedly been taking notes.
I thought it might be interesting to see how organizations have gone about addressing their own backup data center strategies, and use three quick case studies that demonstrate that there are still no hard and fast rules. In the first instance, Computer Associates of Islandia, NY made the decision to create its backup data center in the midwest. It seems that this was a decision that was very much based on geography, as their range of nearby possibilities was somewhat limited:
"We are headquartered on Long Island. If we'd gone 50 miles west (of Islandia), the mirrored site would have been located in New York City. So the company made a decision that if we had to go farther west anyway, we might as well put the site in the Midwest," says Walt Thomas, CA's CIOIn this example, the company has moved to a different geographic region, which requires a completely seperate workforce.
This article shows the Florida operation of the Royal Bank of Canada determined that its maximum threshold for operational downtime was two hours. In assessing the range of risks, RBC likely weighted its mitigation strategy not against terrorism, or blackouts, but weather - specifically hurricances. Instead of building a heavily reinforced fortress, it decided to build a data center located 60 miles from the original site. It is interesting to note that this decision was made so RBC could use the same workforce:
Execs decided that the second data center had to be within driving distance, but also had to be somewhere that didn't normally experience the same weather patterns as the primary data centerYet another approach is demonstrated by a non-financial player - Deloitte & Touche in the creation of the Deloitte & Touche Cyber Center in Amersterdam - a centralized mirrored site designed not only to combine its IT resources distrubuted over 109 locations in The Netherlands, but to do so in a well-designed aesthetically pleasing environment:
On a white board we outlined our vision of what the data center should look like: a corporate facility comprised of two equal columns connected by a central space. We envisioned the columns containing the employee workstations, with each column also featuring a mirrored data center on the ground floor. The central facility was to contain a grand cafe, a place where clients, colleagues and partners, could meet and relaxOne of the main reasons Deloitte & Touche chose this option was the inherent cost savings associated with reducing costs on human capital (the mirrored facility is administered by a single team of professionals), a greater degree of location-wide standardization and an overall reduction in the costs associated with systems architecture. It is also interesting to note the emphasis Deloitte & Touche put on design and building selection.
Three distinct businesses, three distinct strategies based on a different evaluative criteria.
Please join the discussion about mirrored data centers.