Back when the US economy was in freefall, State Street could have put any pending technology projects on hold. After all, banks were failing and merging hastily left and right, layoffs were the rule for the financial services arena and no one knew when the bleeding would stop. Instead, the Boston-based financial behemoth commenced an ambitious IT project that was years in the making, groundbreaking in scope and required near constant promotion inside and outside the company.
At the Wall Street & Technology Capital Markets Cloud Symposium, held at the NY Hilton on May 17th, State Street's Kevin Sullivan revealed how the firm adopted a cloud-based architecture that is now being used in 20 cloud-based projects -- even when the technology was not entirely ready for primetime.
As Chief Infrastructure Architect, Sullivan and his team approached State Street management with plans to overhaul its internal systems and exploit cloud computing. It was a tough time for a large-scale project: The US and global markets were in a recession and skidding to avoid a depression.
"We started in 2009," says Sullivan. "Not a great time in this business and an even worse time to talk about large technology projects. We had to go slow and in a measured way -- it took time to get traction."
It turns out that the economic slowdown was a prime time to review the firm's infrastructure. Sullivan and his team's evaluation looked at what systems could be improved and consolidated. "Out of this review, we saw growing complexity and a need for virtualization and we still had a lot of complexity in the operations," says Sullivan.
For example, when Sullivan started at State Street in 1990, he recalls that the firm worked with three major technology partners. "By 2009, we had 150 tech partners. And the complexity of managing this was an issue. How do we do this going forward?"
During the review, his team found that while the hardware had become commodities, usage had blown through the roof. In 2009, on the infrastructure side, provisioning time and utilization was not what we wanted them to be. He says that it "took weeks to roll out a new serve," adding later that, depending on the project, provisioning sometimes took six to eight weeks.
Sullivan had to prove that his cloud migration project could save money. "We had to prove benefits were there in our Proof of Concepts (POC) exercise," Sullivan continues. "We had to measure what we had done compared to the baseline to see the progress.
In terms of buy vs. build, State Street did a combination of both since it could not rely on a single vendor for this large and complex cloud project. "There is no out-of-the-box solution from one vendor. We had to add some things," he says.
Sullivan told the myriad vendors -- he declined to name them -- that they needed simplification: More standards, consolidation and self-service. Later, Sullivan did give a shout-out to Platform Computing, which he said has been very helpful to State Street.
"Once the POC was done, you had to go to 'wide participation' such as Application Support and Administration," Sullivan adds. "There were 30 groups that had to be approved in this matter. We created 400 use cases and [whittled the number down] to 150. We had to see if these can be accomplished on the cloud," says Sullivan.
This POC process took six months.
As with any large-scale project, the firm and its IT leaders had to submit to an external review phase with industry experts. "We talked to regulators and auditors. We had to make sure that they were aware of what we were doing," he says.
State Street went with POD (processing on demand) on the cloud and declined to store its data on the cloud. Why? Volume. "We didn't put data on the cloud because we had a lot of data and moving it would take too much time," says Sullivan.
State Street's data that interacts with the cloud are stored on two data centers -- a primary center and a second backup center.
While this project is far from finished, Sullivan says from the start of this project to today took two and a half years. In fact, the proof of concept portion took one year by itself, he told the audience of IT professionals. The major cause for the lengthy project was culture.
"It wasn't because of the technology, it was the people you had to orient. It was like the movie Groundhog Day everyday. We had the same discussions all over again," he recalls. "We forced the managers and business heads to get together once a week for two hours to review everything that was going on so no one group got left behind."