The theory of thinking differently: banks moving to cloud without multicast support

Einstein once said that the “measure of intelligence is the ability to change.” In the complex world of financial markets, change is very much afoot, as the major institutions seek to overhaul their legacy trading platforms from an on-premise, to a public cloud environment. But exactly how easy the migration will prove to be is very much up for debate.

Whilst this migration will be hard for most system types, in these institutions, it can offer large savings and is achievable. From a front office perspective, there is no doubt that with a huge increase in bandwidth, the public cloud providers are putting a huge amount of effort into improving their network stack. The likes of Microsoft Azure, Amazon Web Services and GPC have all been rolling out larger piped enhanced network adapters which significantly reduce CPU overhead and jitter, providing more consistent latency. From a performance perspective, this means that the public cloud is ready to onboard front office real-time trading systems - from Single Dealer Platforms to algorithmic trading systems.

However, for these real-time trading systems there is a major blocker, UDP multicast. The trouble is that so many trading platforms are still distributed on systems relying on UDP multicast messaging - which makes any transition to the cloud problematic to put it mildly. Because as things stand, UDP multicast is not supported by cloud providers. This makes the migration path anything but straightforward.

As a result, financial institutions face a whole plethora of challenges when trying to move to the public cloud. To start with, a bank’s entire messaging layer needs to be re-designed. If this wasn’t enough, front office trading systems then need to face up to the impossibility of using UDP multicast. That’s a real challenge when a firms entire existing platform is based on such messaging architecture. Certain products support UDP unicast, which is available on public cloud, but it significantly limits the functionality of those system as they’ve been designed with multicast in mind.

Another aspect which firms need to consider when moving to the cloud, is if there will be an increase in the number of instances they are running. For instance, if they want to establish more points of presence for their clients, deploy their system in (many) more regions, or move to a serverless architecture. Since most messaging systems are licensed on a per core or per node basis, the bill can rise significantly. With these issues in mind, what are the alternative solutions currently available to financial institutions?

Unfortunately, many of the well-known messaging systems currently employed, are simply not fit for this purpose. This is why, as we speak, banks are asking us for a specific solution that enables them to migrate their existing platforms using multicast, to the public cloud.