- Application Performance
- Cyber Range Deployment
- Data Center Performance
- Load Balancer Testing
- Low Latency Networks
- Mobile Network Security
- Mobility Performance
- Network Performance
- Network Security
- Network Visibility
- Product R&D
Part 1: Finding latency in financial transactions, where every microsecond counts
The latency problem
When it comes to network performance, bandwidth gets most of the attention. But bandwidth has a twin brother that also affects performance, and his name is latency. This more complex element refers to delays that occur when data is processed by computers and transmitted across networks. As little as 10 microseconds of latency can cause financial losses, especially for time-sensitive financial applications such as high-frequency trading (HFT). To gain a competitive advantage, financial institutions are using specialized ultra-low latency (ULL) test systems to root out every last second of latency possible in their networks.
In financial transactions, latency affects the time between the initiation of a transaction and its execution. For example, when a brokerage’s automatic trading system decides that the price is right for a buy- or sell-operation, it initiates a trade. Any latency between the initiation of a trade and the match of the buy/sell to an appropriate sell/buy order provides an opportunity for the price to make an unfavorable change. Additional latency can literally cost money.
According to the TABB Group “... estimates that if a broker’s electronic trading platform is 5 milliseconds behind the competition, it could lose at least 1% of its flow; that’s $4 million in revenues per millisecond.” (“The Value of a Millisecond: Finding the Optimal Speed of a Trading Infrastructure,” The TABB Group report, April 8, 2008).
Improving latency of all types is a desirable objective. To know where to focus improvement, it’s essential to understand the latencies of the network’s component systems and the complete network itself. Testing provides the means and measurements for identifying sources of latency.
This is the first of a five-part blog. Still to come are testing scenarios for:
- Part 2: Component selection and characterization testing
- Part 3: Pre-deployment testing
- Part 4: Full system test
- Part 5: Live system test
For more informationblog comments powered by Disqus