With digitisation in banking gathering pace, the benefits of using cloud-native software are starting to hit home. Advantages include the ability to quickly scale up and down, easily reconfigure processes, rapidly get new products and services to market, access data in real time and adapt systems for regional differences.
So when Google Cloud approached Avaloq, the leading specialist in wealth management core banking software, to become a strategic partner, Avaloq knew the timing was right.
Making the move to the cloud
More than 150 private wealth managers in over 30 countries rely on Avaloq’s platform to help them manage some CHF4.5 trillion in assets on behalf of their clients. With the fast-growing popularity of cloud technologies and interest among major financial institutions, Avaloq’s move to Google Cloud needed to be faultless and fast.
While Google and Accenture already have a strong partnership in moving independent software vendors (ISVs) to the cloud, in this case Orbium, part of Accenture Wealth Management, was brought in for its in-depth Avaloq expertise. As its leading implementation partner, Avaloq trusted Orbium to plan, test and implement its transition to a cloud environment in a pilot project.
Getting it right from the start
To manage the move, Accenture and Orbium put together a team of five in-house experts, supported by a cloud engineer and a performance specialist from Avaloq, as well as a Google architect. Everyone worked remotely due to the COVID-19 pandemic.
As with any project, the planning stage was critical. Get that right and any installation is always much smoother. However, the team was also aware that much of the work they would be doing was going to be a first.
At the end of June 2020, planning started. This involved assessing and reviewing the technology before writing the high-level architecture plan that would form the basis of the installation. Meanwhile, Google provisioned the bare metal server that would host the Avaloq software and its Oracle stack.
Everything had to be analysed and designed from scratch – from how different networks would communicate with each other in the cloud to how the firewalls would work. And the team had to find the optimal configuration to make sure the underlying software could run at the speed required by banks that need to be able to process millions of transactions every hour.
Installation and testing
With the plan complete, the installation kicked off in August with the supporting infrastructure – in this case that meant preparing two Windows virtual machines and a Google machine. Both connectivity and storage were carefully managed: the former to optimise performance; the latter to comply with data-storage regulations for the different jurisdictions in which Avaloq banks operate. Within two weeks we had a working Avaloq core banking system on Google Cloud and began testing its performance.
Critical to the success of the project was how many transactions per second the software could execute. Initial tests revealed that the system was too slow relative to Avaloq’s internal servers, a key benchmark. The network latency between the smart client – the front end where users click to access the banking system – and the servers was good. However, special attention was needed to reconfigure the Google server and turn on simultaneous multi-threading (SMT). SMT, also known as hyperthreading, is where each central processing unit (CPU) splits its physical cores, known as threads, in two, one being virtual. This allows each core to do two things simultaneously, increasing the processor’s efficiency without any loss of speed. Subsequent tests for all transactions, such as payments and forex orders, showed that Avaloq on Google Cloud has a great potential.
Success and learnings
By early September, testing was complete and Orbium stepped back. In a timely manner, we had prepared and installed a proof of concept that could give wealth managers and private banks using Avaloq the increased agility of running software in the cloud, as well as improved performance. With an optimal mixture of cloud expertise from Accenture, a leader in Google Cloud Platform Services, Avaloq skills from Orbium and great support from both Avaloq and Google, we were able to stick to the original plan and architecture design when executing the deployment and testing. We did find out, however, that connectivity between different zones and servers in the cloud might be non-trivial.
Ultimately, however, while the script execution itself is no different from when it is running on an internal server, there are some improvements in key benefits with Avaloq on Google Cloud. For example, built-in tools and dashboards monitoring cloud performance made it easier to find bottlenecks. There are also tools to automate the building and networking of infrastructure, making it simpler to replicate or rebuild environments as needed. Both are typical benefits of using cloud technologies. Another big advantage is easy connection to Google Cloud managed services – making it possible to leverage the value of innovations like AI/ML, for example, by connecting Avaloq data with services like BigQuery.
Installing Avaloq on Google Cloud took one month of planning and five weeks of preparing and testing. The results of this effort are now available to Avaloq clients and prospective clients, helping to make the services they offer safer, faster and more agile for years to come.