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

A-Team Insight Blogs

Citihub Pushes Down IPC Latency with Hardcore/Tibco FTL Combo

Subscribe to our newsletter

Trading technology consulting firm Citihub has released highlights of performance testing it completed on Tibco Software’s FTL middleware running on Hardware Computer’s over-clocked, liquid-cooled Detonator workstation. Not surprisingly, the results are the fastest seen to date for inter-process communication.

According to Citihub associate partner Ben Newton, the tests were driven by ongoing consulting engagements with trading firms that are looking to reduce latency for applications such as algorithmic and high frequency trading, but in a cost effective way, that’s also straightforward to deploy.

Enter Hardcore Computer, which manufactures a range of servers where the CPU is over-clocked – that is run at a clock rate faster than its design parameters – and so produces higher performance. Hardcore servers can achieve this, while keeping the CPU cool (and so reliable) by immersing the CPU and motherboard in dielectric liquid, which dissipates heat far better – 1,350 times better says Hardcore – than traditional air-cooled servers.

The Citihub tests were on a Hardcore Detonator workstation server (the company also manufactures blade models), with an Intel Xeon X5698 microprocesor. That chip is actually an ‘off roadmap’ design by Intel, which as standard runs at a fast 4.4 GHz. But when over-clocked inside the Detonator (on a motherboard from Supermicro Computer), it is pushed to 4.93 GHz on a continuous basis.

Running on the Detonator was Tibco’s FTL middleware, configured for inter-process communication between applications. Citihub’s tests determined that 16 byte messages – for the marketing folks – were passed with average latency of 237 nanoseconds, while more typical messages of 128 bytes took 271 nanoseconds.  Previous benchmarks published by Tibco show 128 byte messages being communicated in 354 nanoseconds.

Moreover, latency was “massively predictable” says Newton – less than two nanoseconds at one standard deviation. For its testing, Citihub employed embedded assembler code modules that count the clock cycles of the CPU – accurate to a fraction of a nanosecond.

Newton reckons that deploying Hardcore servers represents “an easy speed up” with a low “cost per microsecond improvement” for firms since no trading application software changes need to be made.

Configuring Citihub’s test application to use FTL was less than a man week of effort, says Newton, who also praises the middleware for its maintainability, with a good user interface – factors that impact cost of operation.

By complementing Hardcore’s liquid immersion (within the server) with water cooling within a data centre also pushes operations costs lower, since heat can be dissipated outside of the data centre itself, reducing the need for air conditioning. The potential exists, notes Newton, to improve performance while reducing costs.

Subscribe to our newsletter

Related content

WEBINAR

Recorded Webinar: Modernising Equities Trading Infrastructure: Architecture, Resilience and the Economics of Change

Equities markets are entering a more distributed, always-on phase. Liquidity is fragmenting across venues and geographies, trading hours are extending toward 24/5 and beyond, and market data volumes, investor expectations and competitive pressure are all rising at once. Many of the platforms that still underpin the equities industry were built for a different era –...

BLOG

Building for the Next Big Event: What Prediction Market Operators Need from Exchange Technology

By Ian Salmon, Head of Product Marketing, Adaptive. Prediction markets have moved from the edges of the financial ecosystem into a space that increasingly resembles regulated market infrastructure. What began as a retail phenomenon around political events and sports outcomes has evolved into a sector attracting institutional capital, established exchanges and serious regulatory attention. The...

EVENT

AI in Data Management Summit New York City

Now in its second year, the AI in Data Management Summit NYC explores what changes when AI becomes part of the data management operating model.

GUIDE

Trading Regulations Handbook 2022

Welcome to the third edition of A-Team Group’s Trading Regulations Handbook, a publication designed to help you gain a full understanding of regulations that have an impact on your trading operations, data and technology. The handbook provides details of each regulation and its requirements, as well as ‘at-a-glance’ summaries, regulatory timelines and compliance deadlines, and...