Ashley Davis is a managing director at JPMorgan Chase with responsibility for the financial services firm’s data center strategy worldwide and for driving its Technology Sustainability program.

Consolidation and alignment

How has the data center estate changed at JPMorgan Chase in recent years?
We’ve had a significant consolidation program over the past five years. From operating 100-plus data centers, we’ve adhered to a roadmap that will get JPMC to circa 30 core sites globally by the end of 2012.

There is always going to be an element of what might be called “infrastructure mismatch.” Buildings are built with a lifecycle for 15 to 20 years; on the other hand, the refresh cycle for applications, services and platforms is between one and three years and the network they depend upon has typically got a three- to five-year refresh.

So the broad challenge is maintaining the currency of the infrastructure in the data center, taking a building that was put up with a 15-year term in mind through multiple refreshes ― it’s like performing open-heart surgery.

How do you ensure alignment between what the business units demand and the data center facilities available?
In the latter half of the last decade there was a one-size-fits-all mentality, but that’s not the case any more. Data center capacity was a key challenge and hence application placement was typically landed in our most resilient facilities (Tier IV). This “goldrush” has had implications; now, we’re driving the appropriate behavior around right-sizing these centers and ensuring the appropriate behavior around application placement. And technology refresh and factors like high availability disaster recovery are forcing a rethink on application placement ― that is, which app is suited to which environment and which tier of data center.

Defining a taxonomy, a suitable class of service for a given application, presents an opportunity for organizations to reduce costs. If we can force the application owners to think about the required resiliency and so put applications in the most appropriate sites (aware that a Tier IV site costs 30% to 40% more than a Tier II site, for example) then we can save money. That has not been an easy journey with the business, but it’s been a journey driven by the new reality of the cost of data centers.

Cloud opportunities

What role will cloud computing play in this more targeted placement of applications?
Cloud ― both internal and external ― is going to increase the need for a defined taxonomy of applications so you can determine the application type best suited for the cloud. Our data center strategy has four themes: “sub-millisecond,” which delivers latency competitiveness, where we want to host near or with a local stock exchange engine; “millisecond,” where the data center is next door to a hot metropolis site in London, New York, Hong Kong and so on; what we call “sync/async,” big out-of-town sites where we do the majority of the back office processing; and “remote,” where we’re less agnostic on location but need proximity to a fast network connection for performance and available cheap energy. At this stage, I’d classify the biggest cloud opportunity as the “remote” level, where we’re not agnostic about location and have validated the application type for hosting there.

Is there pressure internally from the business to take advantage of the flexibility and responsiveness of cloud?
Cloud means many things to many people, but it doesn’t mean a guy can go and use his corporate credit card to buy the hosting of an application. Cloud is a purposeful, deliberate adjunct to our current strategy and therefore we’re not just buying a hosted rack at a time. It’s the application that is important, not the footprint of the application.

Sustainability

What is the JPMorgan Chase agenda for sustainability within its data centers?
This is a pervasive, important issue for us. In 2007, we set the goal that by 2012 we would consume 20% less power than 2005 levels. Much of our data center consolidation program has been around achieving those targets.

We use things like computational fluid dynamics to continuously monitor the performance of the inside of data centers, using that for both placement decisions and also performance of sites. When purchasing products, we are always cognizant of their energy ratings. And we’re now doing smart metering and tracking usage of power down to rack level and trending that usage.

Further reading

The changing shape of the corporate data center: an in-depth report.