Tag Archives: IPC
Last week I had the opportunity to attend the A-Team Group Low Latency Event in New York City where as you would expect, the talk was about the continued push for lower and lower latency but also the massive growth of data. If there’s ever an example of Big Data in action, Capital Market are it!
Jitesh Ghai from the Ultra Messaging Business Unit participated in panel discussion on market data and how enterprises are dealing with the incredible volumes. He highlighted how some of our customers have gained massive performance improvements and lower latency while reducing hardware infrastructure. They did this by combining marketing data hardware solutions with Ultra Messaging software for cross enterprise distribution. (more…)
Recently, 29West-Informatica announced the general availability of a variety of offerings from the Ultra Messaging® line of low-latency messaging products.
First is LBM 4.0, featuring big performance improvements with the IPC transport, for applications on the same host. These performance gains are highly scalable, as well. As a result, with LBM 4.0, you can now run a complete trading system on a single server and eliminate excess latency from the system.
In addition to LBM 4.0, 29West-Informatica also announced these two other Ultra Messaging offerings, also generally available:
- Ultra Low-Latency Load Balancing (ULB) – Fully leverage parallel processing for demanding front-office messaging applications, such as high-frequency trading, with the nearly-horizontal scalability across multiple CPUs or servers provided by Ultra Messaging.
- Ultra Messaging Queuing Edition (UMQ) – A brand-new product, UMQ extends low-latency, once-and-only-once messaging to back- and middle-office applications, along with the performance and reliability advantages of Ultra Messaging. UMQ enables more flexibility in application design with fault-tolerant queuing, low-latency load balancing, and multiple dissemination models.
See links below for more, or contact your 29West-Informatica sales representative.
- LBM 4.0 Features Include Latency as Low as 660 Nanoseconds! – LBM 4.0 blog post
- IPC Performance Report Shows Ultra-Low Latencies of 660 Nanoseconds – IPC performance report blog post
- Informatica Releases New Ultra Messaging Offering; Latency Busters Messaging 4.0 Delivers Breakthrough Performance for Financial Services Operations – Press release
Experts agree that in the future CPU scaling will go out, not up. Clock cycles seem to have plateaued around 3 Gigahertz, and while five years ago people got excited about dual-core machines, today the major commodity CPU manufacturers deliver 4, 6 and 8-core processors and system vendors link these CPUs together in multi-socket servers, as Cisco does in their UCS line.
Our Ultra Messaging customers - financial players like exchanges and investment banks – are always interested in gaining every possible performance advantage in the ultra competitive world of high-frequency electronic trading. (more…)
- 35 Million messages per second using 29West IPC
- 89% cost reduction over current-generation servers
Last week, 29West had a couple of engineers in downtown Chicago, working with a good customer of ours on a new multi-core server they have, doing scaling tests with our streaming messaging product, LBM. This happens all the time; the more interesting part is that the box is a 4-socket Boxboro server, using Intel’s soon-to-be-released Nehalem-EX 8-core CPUs. That’s 32 cores, with the ability to run an additional 32 Hyperthreads; for a total of 64 simultaneous threads on a single machine. Needless to say, this customer was very interested in seeing what real-world system consolidation savings and performance gains they could realize with a machine like this – the kind that we’ve been talking about as imminent for several years now.
What we found was very exciting. We were able to deliver 35 Million 100-byte messages per second between 32 publishing processes and 32 receiving processes using LBM’s IPC transport. After reviewing these numbers, our Chicago customer reported they will be able to use this single 4U machine (which they quoted as costing them $20,000, obviously there is no official pricing yet, as the CPUs have yet to be officially released) to replace an entire rack of hardware, normally costing $180,000. This is the kind of immense cost savings with server consolidation that the industry has been talking about for years: 89% cost reduction with no performance penalty, in this case.
Intel and AMD agree that future CPU scaling will go out, not up. While clock cycles seem to have plateaued around 3 Gigahertz, and five years ago we were all excited about dual-core machines, the major commodity CPU manufacturers are now cranking out 4-core CPUs every day, with 6 and 8-core processors (like the Nehalem-EX chips we tested last week) coming very soon.
29West is extremely excited to work with our customers to best deploy our next-generation messaging inside these new densely multi-core systems, and integrating them into larger deployments using our full range of message delivery (streaming, persistent and queued) and connectivity features (global WAN routing, multi-hop forwarding, desktop distribution).