Category: LL-Benchmark Results
On 07 Jan 2009 in A-Team Delivery Terms, A-Team Free, LL-Benchmark Results, LL-Messaging Middleware, Low-Latency.com, MDI-Messaging Middleware, Market Data Insight 
Referencing a benchmark it conducted in September 2008 for the Thomson Reuters Market Data System (RMDS), Stac has released an update after applying certain optimisations.
For the update, Stac added two more 10GigE ports (bringing the total to four) to reduce contention among the cores competing to consume and transmit data, and moved the burden of TCP processing to the NIC, using Chelsio’s TCP Offload Engine (TOE). Stac then re-ran the Producer 50/50 tests on an equivalent 24-core Intel 7450-based server from IBM running Novell SUSE Linux Enterprise (SLES) 10.
In this optimised configuration, RMDS was able to output 19.1 million updates per second.
On 16 Dec 2008 in LL-Benchmark Results, Low-Latency.com
Stac has released a second report on the performance of the Reuters Market Data System (RMDS) using Chelsio 10GigE NICs that perform TCP offload. This time, the platform was Novell Suse Linux Enterprise Real Time running on HP c-Class blades.
READ THE REST
On 12 Dec 2008 in LL-Benchmark Results, Low-Latency.com
Stac has released a follow-up report on the performance of the Reuters Market Data System (RMDS) using Chelsio 10GigE NICs and Red Hat Enterprise MRG Realtime. In a previous Stac report, a solution consisting of Chelsio 10GigE NICs, Red Hat MRG, Intel processors, IBM HS21 Bladecenter servers and Blade Network Technology (BNT) switches demonstrated record low latency and very high throughput with RMDS.
READ THE REST
On 10 Dec 2008 in LL-Benchmark Results, Low-Latency.com
Securities Technology Analysis Center (Stac) has benchmarked Novell’s Suse Linux Enterprise Real Time at the lowest mean and maximum latencies ever recorded at high rates with the Reuters Market Data System (RMDS), as well as the highest RMDS throughput for a two-socket server. Novell says it was able to achieve simultaneously high throughput rates and extremely low latencies because of close collaboration with its technology partners – HP, Intel and Voltaire.
READ THE REST
On 11 Nov 2008 in A-Team Delivery Terms, A-Team Free, LL-Benchmark Results, LL-Hardware Acceleration, LL-Messaging Middleware, Low-Latency.com, MDI-Delivery Technologies, Market Data Insight
Solace Systems, which provides hardware-based messaging middleware, has collaborated with Arista Networks and Intel’s NetEffect unit to demonstrate an ultra low-latency data delivery solution using hardware platforms.
READ THE REST
On 11 Nov 2008 in LL-Benchmark Results, LL-Messaging Middleware, Low-Latency.com
Stac has released its second report involving real time Linux technology and a market data workload. In this evaluation, Novell, HP and Intel asked Stac to measure the performance of the Reuters Market Data System (RMDS) 6.x with the following other stack layers:
• Novell SUSE Linux Enterprise Real-Time 10 SP2 update 3 (SLERT)
• HP BladeSystem with HP BL460c Blades
• Quad-Core Intel Xeon E5450 (”Harpertown”) CPUs
• HP 4X DDR InfiniBand HCA
• HP DDR InfiniBand switch
• Voltaire Messaging Accelerator (VMA) InfiniBand Software
READ THE REST
On 10 Nov 2008 in A-Team Delivery Terms, A-Team Free, A-Team Free Access, LL-Benchmark Results, LL-Messaging Middleware, Low-Latency.com
Stac has released test results involving real time Linux technology and a market data workload. Red Hat, IBM and Intel asked Stac to measure the performance of the Reuters Market Data System (RMDS) 6.x with the following other stack layers:
• Red Hat Enterprise Real Time MRG
• IBM BladeCenter with HS21 XM Blades
• Quad-Core Intel Xeon E5450 (”Harpertown”) CPUs
• 10 Gigabit NICs from Chelsio Communications, without client-side acceleration enabled
• 10 Gigabit Ethernet switch from Blade Network Technologies
READ THE REST
On 25 Sep 2008 in LL-Benchmark Results, Low-Latency.com
Stac has completed tests to find the maximum Point-to-Point Server (P2PS) Producer 50/50 throughput on a single server using Intel Dunnington processors over a Cisco Nexus 5000 switch.
READ THE REST
On 24 Sep 2008 in LL-Benchmark Results, LL-Data Fabrics, Low-Latency.com
Sun Microsystems and GigiSpaces have set a new benchmark performance result that address the scalability needs of capital market customers at High Performance on Wall Street this week. In performance tests, the energy-efficient Sun SPARC Enterprise T5240 Server processed 1.3 million transactions per second with 60 concurrent clients to achieve an improvement of up to 40 per cent over competing multi-threaded platforms. The average throughput per client thread was sustained at about 21,000 transactions per second.
READ THE REST
On 24 Sep 2008 in LL-Benchmark Results, LL-CEP/Event Stream Processing, LL-Order Book Management, Low-Latency.com
Two weeks ago, Stac released a Stac Report on order book consolidation using the Aleri Streaming Platform running on a single quad-socket server using Solaris 10 and the Intel Xeon 7350 (Tigerton) processor. This server handled 180,000 order book updates per second with a mean latency of approximately 1.5 milliseconds and 99th percentile latency of approximately 3 milliseconds. On September 15, 2008, Intel released the Xeon 7400 series of 6-core processors, codenamed “Dunnington.” Intel therefore asked Stac to re-run the exact same Aleri configuration after upgrading the processors in the server.
READ THE REST
On 22 Sep 2008 in LL-Benchmark Initiatives, LL-Benchmark Results, Low-Latency.com
The Securities Technology Analysis Center’s (Stac) Benchmarking Council – a group of 40 leading trading firms formed a year ago – has approved the Stac-M1 Benchmark specifications, the first industry-standard benchmark for trading technology.
READ THE REST
On 17 Sep 2008 in LL-Benchmark Results, LL-Feed Handlers, Low-Latency.com
Stac has released benchmark results for the InRush embedded ticker plant from Redline Trading Solutions. In addition to the InRush software, the stack under test consisted of five HP DL380 servers that consumed directly from the network that was delivering the OPRA UDP multicast stream from the playback mechanism (Stac Hi-Fi). Each server hosted a feed handler that ran on a board containing a Cell Broadband Engine processor, along with a test client that ran on the server’s native Intel processor via Linux. The Cell board contained its own memory and network interfaces.
READ THE REST
On 04 Sep 2008 in LL-Benchmark Results, LL-CEP/Event Stream Processing, LL-Order Book Management, Low-Latency.com
Stac has released results from tests of an order book aggregation solution built on the Aleri Streaming Platform. The Aleri solution consumed streaming, full-depth order data for two US equity exchanges via the Reuters Market Data System version 6 and aggregated the orders by price to produce a single, consolidated order book stream as the output. The consumer of the consolidated output stream was a single instance of a test client that Stac wrote to the Aleri API and Aleri helped to optimise. The consumer subscribed to one quarter of the entire set of consolidated order books.
READ THE REST
On 10 Jul 2008 in LL-Benchmark Results, LL-Messaging Middleware, Low-Latency.com
Sun Microsystems has announced new benchmark performance results for Thomson Reuters market data platform. The new benchmark demonstrates that running the Reuters Market Data System 6.0.3 (RMDS) platform on Intel-based Sun technology breaks the million-messages-per second barrier, as well as offering the lowest available network latency on a 1Gb Ethernet.
READ THE REST
AMBook Intel Quad-Core (90 KiB)
Loading ...