Test Results
Lower Latency and Lower Cost – a Winning Combination
In comparative tests, infoDyne wins every time! Tests, conducted by our customers against their previous consolidated market data vendor, all showed dramatic improvement. Such tests were designed to measure the performance of InfoDyne's Ticker Plant and Middleware² against their previous vendor's consolidated feeds and Middleware² solution.
In specific customer pilots where the InfoDyne solution was chosen over other alternatives, specific factors that went into the decision went beyond latency and included:
- Ease of use of market data APIs
- Compatibility/interoperability
- Integrity/reliability of market data and delivery solution
- Capability of fault-detection, redundancy and fail-over
- Installation, configuration, monitoring, and maintenance
- Ability to customize and add required functionality
- Scalability/extension throughout an enterprise
- Support, responsiveness, and training offered
Most importantly, InfoDyne's solution proved to be less costly all together. The cost of ownership, including data charges, development, maintenance, and support, were reduced when the customer deployed the InfoDyne Ticker Plant System.
A Specific Test Scenario
This customer tested InfoDyne's Ticker Plant System with Direct Feeds against Reuters RDF and OPRA (OPDN) feeds. Tests were conducted to measure and compare update latency for each of the feeds as well as comparing update latency through InfoDyne's Middleware² vs. Reuters' Triarch system.
These are some general results derived from the detailed analysis conducted (see also NASDAQ, CTA, and OPRA Test Summaries):
- Tests designed to compare update latency between InfoDyne's TPS direct feeds to Reuters IDN feeds clearly shows an average improvement of ~350 msecs over ~98% of the trading day.
- Tests designed to compare update latency between InfoDyne's Middleware² to Triarch (i.e. ITES running on both systems) clearly shows an average improvement of ~35 msecs over ~99% of the time.
- At peak update periods of the day, update latency for the Direct Feeds compared to Reuters IDN feeds showed the updates arriving on the Direct Feeds an average of 500 msec sooner, with numerous events showing as much as an 8 second advantage for the Direct Feeds.
Request/Response Latency — 8 seconds vs. ?? minutes
InfoDyne's Ticker Plant System processes the direct exchange feeds at the customer's site, maintaining an in-memory cache of the entire volume of data available from the exchange source. This means that any symbol required from a given exchange source is immediately available and delivered to the customer's applications. The only limiting factor in this case is the speed and efficiency of the hardware, network, and applications at the customer's location.
This differs from consolidated feeds from third party data vendors, where the direct exchange feeds are processed in the vendor's data center. Through remote connectivity, data is then provided through a "selective" request service. These selective requests are typically processed at a lower priority than the update traffic and must be "throttled" to protect the central systems that are serving many customers. Once a symbol is cached at the customer location, it is then available for distribution, unless of course the "cache limit" has been exceeded. In this case, requested symbols may be rejected, or other symbols may be pre-empted. This causes further delays since the data center must be sent another request to unsubscribe from the pre-empted symbols.
It is no wonder that in numerous comparative request/response test scenarios, InfoDyne's Ticker Plant far out performed interactive third party feeds. In one typical test to request 10,000 symbols, InfoDyne's TPS delivered all responses in less than 8 seconds while the same requests sent to a third party consolidated feed took over 10 minutes to deliver the items.