The Singapore Exchange (SGX) has launched an application migration initiative that involves shifting workloads to Kubernetes to reduce its dependencies on virtual machines from both VMware and Red Hat.
The global provider of a platform for trading stocks and other financial instruments is wrestling with two core challenges pertaining to virtual machines. In addition to increased VMware costs stemming from licensing changes implemented by Broadcom, Red Hat has also announced end-of-life support for its virtualization platform based on kernel-based virtual machines (KVMs) as part of an effort to encourage its customers to migrate to the Red Hat OpenShift platform.
SGX, however, has decided to opt to rely on open source kubevirt software to convert its applications into a set of containers that comply with the Open Container Initiative (OCI) format, enabling its legacy monolithic applications to run on Talos Linux, an instance of a lightweight distribution of Linux developed by Sidero Labs that has been more tightly integrated with Kubernetes to create an immutable IT platform.
That approach makes it possible to deploy applications on bare-metal servers in a way that not only aligns with existing GitOps workflows and infrastructure-as-code (IaC) tooling but is also simpler to manage and secure, said Rushan Ratha, head of platform engineering for the SGX FX Group.
It remains to be seen how legacy applications will perform at scale using kubevirt, but SGX is also rebuilding the underlying IT infrastructure to provide additional compute resources. Most legacy applications should be able to make the transition but even if a handful of applications remain on virtual machines the level of dependency on Broadcom and Red Hat will be greatly reduced, said Ratha.
In fact, the overall goal is to make it simpler to rebuild Kubernetes clusters by invoking a Talos Linux application programming interface (API) provided by Sidero Labs, said Ratha. “The nodes should be disposable,” he said.
Initially, SGX is counting on a half dozen members of a platform engineering team to manage its migration project spanning six global data centers, with three of them already running legacy applications on Talos Linux.
Other elements of the SGX stack include the Cilium service mesh from Isovalent, an arm of Cisco, monitoring tools from VictoriaMetrics, open source Flux tools to deploy software and NFS-based storage systems from Pure Storage. However, longer term SGX is looking to migrate to an object-based storage platform, noted Ratha.
It’s not clear how many organizations are actively migrating workloads off of legacy virtual machines, but clearly a lot of these initiatives are now being led by platform engineering teams. The challenge, as always, is ensuring that platforms selected are capable of running legacy applications at scale. That’s especially critical for organizations such as SGX where the speed at which transactions are processed is core to how it differentiates itself from rivals. After all, while reducing costs is always welcome, achieving that goal at the expense of application performance isn’t always going to be met with the same level of enthusiasm.