Wealth management firm St James’s Place (SJP) has begun a project to move its entire IT infrastructure to the cloud, helping to meet its scalability demands and focus on its core financial business rather than data centre management.
SJP, which has around £52 billion of funds under management, operates from 20 offices in the UK, and has recently grown its operations to provide financial planning and investment advice in three new offices in Asia.
Despite operating in a highly regulated industry, SJP has decided to begin moving its infrastructure to Amazon Web Services to support its expansion plans and provide greater flexibility for the business.
“We are a rapidly growing company so speed and agility is really key to us, and we need to be able to scale our operation, especially at the tax year end,” said Andy Montgomery, St James’s Place head of division for IT operation, speaking at last week’s AWS Summit in London.
“As we approached the end of the year, there was a lot of business that needed to go through our systems, and we had to scale our environments so that we [cope with] application submissions, fund switches and various other things. Amazon allows us to scale our environment to be able to cope with that.”
Montgomery explained that management of SJP’s IT estate, which involves 241 servers and several hundreds petabytes of storage across two regions, had become a burden on the business. Now, rather than spending time operating data centre infrastructure to support its various enterprise applications, this has been addressed by moving its infrastructure to AWS, using a virtual private cloud (VPC) connection.
“We don’t want to focus on infrastructure and we don’t want to focus on managing data centres – our key business is wealth management, that is what we do, that is what we are good at," he said. "Amazon provides a wrapped set of services that we can leverage to build applications on.”
The move to the cloud began last year when the IT department was tasked by the business with managing a programme of change during a short timescale, Montgomery said. This required multiple environments to be built in parallel in order to for the work to be completed on time.
“When our senior management team gave us that challenge, I didn’t have the option to go back to them and say 'we don’t have enough infrastructure, storage and servers – this is going to take months to provision',” he said.
“After we went through the discussions of how to do all of that work in parallel, it was very easy to create the infrastructure. We went from three environments to 12 in a matter of days, which allowed us to crack on with those programmes and deliver to the timescales we were set.
"That was a very compelling reason for being in AWS, and it give us that broad flexibility that allows us to deliver based on the pace our business is moving.”
SJP now runs the majority - around 85 percent – of its systems in the cloud. AWS is used to run a variety of software, including its Microsoft Exchange environment, intranet, many enterprise applications and analytics tools.
“We also run our CRM, our data warehouse, and all of our business processing systems apart from one legacy one in AWS. We don’t run Redshift right now but it is something we are aspiring to do," Montgomery said.
However, SJP continues to rely on some on-premise systems, as part of what Montgomery describes as a hybrid cloud. The firm's legacy data centre – which constitutes around 15 percent of its overall estate – is linked to the AWS virtual private cloud through a “fast, low-latency, high bandwidth, direct connect pipe”.
For example, the firm runs its virtual desktop infrastructure (VDI) platform in its on-premise environment. “That works directly over the direct connection straight into AWS, so people get fast access to email from our virtual desktop infrastructure,” he said.
Meanwhile. some applications sit across the two environments. This includes some legacy databases and software that run on "outdated operating systems which remains on-premise", he said.
"But a lot of the stuff we have built around that – our APIs and so on – that sits in AWS and it talks over the fast direct connect,” he said.
“So you can look at your legacy applications you can choose which bits to port over time - your application can sit across both, not just one or the other. By taking that approach we will move more and more to the cloud, until that remaining 15 percent is running in AWS.”
Next section: Migration challenge
The move to the cloud has not been entirely straightforward, however. Towards the end of 2014, the firm began to experience performance issues with its cloud environments.
“We saw screens not loading, and some screens taking in excess of 60 seconds to load,” he said. “We worked with all of our software vendors that were involved in the project to try and troubleshoot the issue. And of course each of the software vendors said ‘my code is good, it must be somebody else’s code’.”
An investigation into the infrastructure revealed no faults, and a decision was made to swap its magnetic storage for solid state drives in AWS after a brief shutdown of its systems. However, this saw only limited benefit.
Montgomery said:“Some of [the issues] were resolved with the solid state drives, but then we looked at the instance type: did we size properly? Did we choose the best storage from the outset? Did we choose the best instance type from the outset? Potentially no.”
This time, a decision was made to swap AWS server instances, switching from CPU-optimised to memory-optimised overnight. This involved doubling both the amount of CPUs and memory.
“Since then we have had absolutely no issues with that system. So it just shows that having some flexibility in your underlying infrastructure platform – which Amazon provides – allowed us to solve that issue over a couple of nights,” he said.
“If we were to do that in a traditional way, swapping out storage and swapping out servers is days and even weeks of work, assuming we have done all of the procurement and handled all of the traditional data centre change aspects.”
Going forward – cloud first approach
While SJP has been focused on moving from its physical data centres to an infrastructure as a service model, the next step is to move up the stack and rely on platform as a service where possible for new developments, as well as deploying software as a service where relevant.
“I want to be as high up in this stack as possible. What we are interested in at St James’ Place is adding value to our business, adding value to our customers and providing business functionality to get work done. The rest of it is largely overhead. So we have become largely a cloud first company,” he said,
For example, this will mean carrying out mobile development using Paas tools.
“So when we sit within our architecture teams, and we say ‘OK, we want to build a new iOS app’ the conversations start with 'how many web servers, database servers do we need' and all the rest of it, which is not the right way to go. You have to change your thinking around cloud,” he said.
“Once you have done Iaas, and once you have ported your workload to the cloud, anything new you are developing, you should be considering about getting above the operating systems and the Iaas and tapping into APIs.”
He added: “So we are looking at DynamoDB, we are looking Cognito, we are looking at various SDKs and other APIs that Amazon makes available to you. Can we develop an app that just uses services where we don’t have to manage the operating system at all? [We want to] let Amazon scale and manage our solutions for us. It is really important for where we are setting our direction.”