Hermes Ramps Up Cloud DevOps Strategy

Tweet
Facebook
LinkedIn
Save this article
Send to
Print this page

After years of patching holes in its hull, one of the UK’s largest fund managers has found its technology is finally ready to take off, thanks to its use of the cloud.

Hermes Investment Management is capitalizing on its cloud development framework, through which it can roll out new releases close to two-to-three times per day. Using Microsoft Azure, the London-based firm has built its developing and testing environments from scratch on the cloud infrastructure, accelerating its ability to release technologies into production.

During a presentation at The Summit for Asset Management, held on February 28 in London, Darren Hodges, head of development at Hermes Investment Management, explained that the two-to-three years ago its IT department, which comprises around 12 percent of the firm’s 470 staff, were struggling with operational overhead from monitoring work environments, managing as many as 220 servers, fixing bugs and patching up systems. At that time, the firm was outsourcing a development team from India, whereas their primary focus was to stabilize the core functions and keep the business ticking over.

“They were fighting fires, they weren’t really thinking about the strategic direction,” he outlined.

Between 2016 and 2017, Hermes shifted its approach, and set its sights on growing the business and revamping its approach to technology development. Over the preceding years, it dismissed the outsourced developers and built a London-based, in-house development team.

“We need them to have a range of skills and we need them to be willing to actually roll up their sleeves and get stuff done. This is also part of our interview process in trying to find those broad-minded people that can get stuck into anything that needs to happen,” Hodges explained.

Today, Hermes has reallocated 50 percent of its IT team to focus on strategic direction and roughly 25 percent to support the business. As part of the new approach, development teams are also set 10-week sprints as opposed to shorter two-week cycles. The objective is to provide the developers with an effective roadmap, to enable them to think ahead, focus on more important problems that require more time and present tangible results at the end.

Hermes has also reworked its development lifecycle by automating testing and implementing tests at earlier stages of the process. As each developer creates code, it triggers a build which is automatically deployed and tested in the cloud environment. This removes the need for dedicated team for testing, where every developer is made responsible for ensuring the code is viable and works. To automate the process further, developers are not required to use manual logins, but any activity where the technology is deployed in or out of the cloud environment is subject to approvals.

During the presentation, Sejai Modi, a DevOps automation test developer at Hermes Investment Management said the flexible cloud environment and this strategic approach to development have allowed the firm to have overnight releases. Using a step-by-step demonstration, she explained that throughout the day, developers write the code, it is tested, a build is done and at 4 AM UK time, a release is quoted. The code is then deployed into the cloud test environment, an end to end progression suite is run and results are published back.

“This means that when our development teams come in first thing in the morning, they are able to see the results of the [release and the] fix that was made, and they can then move on to the next task that they were working on,” she continued.

Herding Cattle

Contrary to many investment firms that have migrated their operations to the cloud, Hermes decided it would take a different approach and build native cloud environments for development and testing. The infrastructure has provided the firm with the necessary scalability to accomplish these tasks, and allowed it to offload the cost and pressure of maintaining its proprietary servers.

“You can scale up and scale down. Take your test environments, for example, they are not used 24 hours a day so why keep them on 24 hours a day. Why pay for an infrastructure when we are not actually using it? Similarly, today our [test] workstations are all switched off. The point is why pay for work stations when they are not in use, so they don’t exist anymore,” said Hodges.

Hodges explains that every night, when the code has been deployed and tested, a record of the release is made. After 4 AM, the development and testing environment is dismantled and shut down, including the servers, networks, firewalls, security layers, and storage. The next morning everything is rebuilt from the ground up in 30 minutes. All the applications and configuration components are resintalled, including Microsoft’s SQL server and all the data is normalized.

“We don’t need to worry if a server isn’t big enough or if we need more CPUs. We change a tag, we run the code, it destroys the machine, it starts it up again and bingo we have got the new machine,” he explained.

Towards the end of the presentation, Hodges described a unique analogy of how cloud infrastructure is similar to cattle, whereas proprietary servers are treated as pets. He explained how pets require dedicated care, resources and time.

“Whereas with cattle they are bred for use [much like cloud],” he added. “They are the same, they are not unique. You can cut up bits, from your network to your storage and everything is modularized.”

In the coming months, Hermes is expected to complete the move of production to Azure’s cloud infrastructure. Speaking on the sidelines of the presentation, Hodges explained that this is part of a wider strategy for the firm to shift its entire operations stack to the cloud.