About a-team Marketing Services
The knowledge platform for the financial technology industry
The knowledge platform for the financial technology industry

A-Team Insight Blogs

Breaking New Ground with Ultra Low Latency Messaging

Subscribe to our newsletter

Inter Thread Communication (ITC) is not a new concept. Developers have been doing it for quite some time. However, doing it fast and well has been tricky. This mimics almost all of the modern day messaging communication primitives, such as IPC and reliable multicast. That is why I believe that ITC is the new ground for ultra low latency messaging.

Pushing the high performance messaging model to its ultimate, present day, conclusion of message passing between threads in a single process is an exciting concept. The same messaging API used as communication between threads within a process, between processes within a box, and between machines separated by an entire planet is a quantum leap in location transparency. The ability that this abstraction allows is quicker time to market for high performance use cases and much more flexibility to messaging users. It simply makes sense for messaging systems to embrace ITC.

ITC is easy to do, but tough to do well. It has its own challenges from an implementation perspective. To do well, it requires intimate knowledge of many low level components, such as CPU caches, cache consistency models, and instruction pipelines. The payoffs are quite impressive.

At Informatica Ultra Messaging, one of our most aggressive customers in the area of high throughput and ultra low latency is LMAX. Martin Thompson, Michael Barker, and others at LMAX have done, what I believe to be, world leading work in pushing the boundaries of how low you can go with latency and how high can you go with throughput in communicating between two threads. In short, 25 million messages/second and 50 nanoseconds latency on modern commodity CPUs.

An even better part is that they and LMAX are extremely open to discussing the approach and the core ideas. The Disruptor code project is more than worth your time.

Ultra Messaging plans to support not only the notion of ITC as a transport, but to adopt the core ideas of the Disruptor pattern in many ways in our products to further provide world class low latency solutions to our customers. Stay tuned for more in this area and more from UM.

See also:

View a replay of a webinar about LMAX.

Martin Fowler has written this great article about LMAX’s architecture.

Here’s a case study on LMAX (PDF).

Subscribe to our newsletter

Related content

WEBINAR

Upcoming Webinar: Agility as Alpha: How Trading Infrastructure Determines Who Wins in Volatile Markets

Date: 21 May 2026 Time: 10:00am ET / 3:00pm London / 4:00pm CET Duration: 50 minutes Tariff shocks, geopolitical realignment and macroeconomic regime shifts are redrawing the investment landscape faster than most firms’ technology stacks can keep up. For hedge funds and asset managers, the ability to move quickly into new asset classes, geographies or...

BLOG

FCA and Turing Institute Collaborate on Synthetic Data to Advance AML Detection

The Financial Conduct Authority has published a research note from its synthetic data anti-money laundering project, an initiative that began in autumn 2024 and was developed with the Alan Turing Institute, Plenitude Consulting, and Napier AI to create a synthetic dataset for AML detection testing. The paper marks the culmination of that work to date...

EVENT

Eagle Alpha Alternative Data Conference, Spring, New York, hosted by A-Team Group

Now in its 9th year, the Eagle Alpha Alternative Data Conference managed by A-Team Group, is the premier content forum and networking event for investment firms and hedge funds.

GUIDE

Hosted/Managed Services

The on-site data management model is broken. Resources have been squeezed to breaking point. The industry needs a new operating model if it is truly to do more with less. Can hosted/managed services provide the answer? Can the marketplace really create and maintain a utility-based approach to reference data management? And if so, how can...