This is how Goldman Sachs is cutting staff through cloud computing
Goldman Sachs has been quietly cutting costs in its tech team by using cloud technology. Over the past nine years it's managed to reduce the amount it spends on information technology staff compensation by 50%.
Financial services organisations have been slow to adopt cloud-computing technology, lagging behind most other industries, but Goldman Sachs is an exception.J Ram, managing director and co-head of cloud platforms at Goldman Sachs, spoke at the TABB Group's MarketTech 2016 conference in New York this week about how Goldman is utilising cloud technology.
“The cloud has changed our entire outlook on computing, how we approach the growth of the firm and how new applications are developed – it’s transformed how we operate in many ways,” Ram said.
One major way it's affected the way Goldman Sachs operates is by allowing it to cut staff costs. In 2007, Goldman had 35,000 servers and it spent 45% of the division's budget on staff and 55% on non-staff.
“That became the biggest inhibitor for growth,” Ram said. “If we wanted to spend $1m on a computing grid, we had to spend close to $1m on staff, and you can’t scale the environment [operating with that cost structure], so we had to rethink that.”
Goldman currently has more than 200,000 servers. It now allocates 29% of its budget to staff and the remaining 71% to non-staff. But it still employs more than 8,000 engineers who focus on software development across the front-, middle- and back-office. It has more than 5,000 applications currently up and running (in comparison, Morgan Stanley has approximately 2,000).
By moving to the cloud, Goldman has grown its plant size – a measure of productive capacity or output – by 580% while reducing the total costs by 20%, including a 50% reduction of staffing costs over that nine-year time-frame.
“Now we’re spending more dollars on the technology itself rather than managing the technology, so we’re better able to tap into the innovation of the latest technology,” said Ram.
Goldman has moved over to 100% virtual desktops.
“The cloud has helped us to manage the workplace environment from a security standpoint and a mobility standpoint,” Ram said, adding that 90% of Goldman’s servers are housed in the cloud.
“At a big financial institution, it’s never a cookie-cutter model, and there’s a high degree of complexity – this [cloud-based infrastructure] is what differentiates our business on a day-to-day basis,” he said. “We had the basic framework of how you define requirements and departments, and the language to understand a lot more about the applications, which is part of development cycle and one of the most significant investments we’ve done enabling our team to design on the fly and expedite their workflow.”
How does the cloud differentiate Goldman Sachs?
Ram said that cloud computing has helped Goldman to shift away from hand-holding. It can now consistently enforce policy through increased control points. This means a greater degree of oversight from risk and compliance; reduced deployment time so that in some cases it’s measured in minutes rather than months; and the use of open standards to enable developers to innovate. All this has helped cut costs.
“The cloud model and stack are helping us to manage our legacy risk, and we’re able to maintain our compliance much more effectively than before,” he said. “Transparency in audit is a core competency – if anybody did anything in the system, it would be tracked, attributed and reviewed, and that degree of control happens without adding any more bureaucratic processes, because it’s based on peer-review and electronic tracking of changes."
The next step: moving to the public cloud
What next? The public cloud.
“We’re just in the starting phase of the public cloud journey – we do have a few apps running, but we see a lot more promise coming out of it, because it offers great flexibility," said Ram.
For example, Goldman had a new business plan it wanted to try out, and previously it would have required the firm to build two data centers, which typically would take four-to-six months on build out and another 12 months for execution. Instead, they are able to make a plan and issue a proposal on the public cloud for planning and evaluation purposes.
“We can go about it relatively quickly using the public cloud, and in this instance we found out that the business plan would not move forward, so we were able to cut our losses and move on,” he said. “Usually when you’re trying out new technologies such as GPU-based computing you would have had to generate set of expertise in house to build that platform, but didn’t have to do that, because we took advantage of the public cloud.”
Photo credit: BlackJack3D/GettyImages