SpeedLink - Low Latency GridServer Services

Delivering Mission-critical Services with Real-time Response

DataSynapse SpeedLink is an ultra high throughput addition to GridServer. Based on standard GridServer capabilities, SpeedLink can share engines seamlessly with other GridServer services. It uses the same simplified programming model and deployment approach as traditional GridServer-based applications, but it employs a different communications architecture and protocol to support extreme throughput and low-latency service operation.

With request-response latencies of less than 1 millisecond and unbounded throughput (100,000s of messages per second across 100 cores, for example), SpeedLink is a highly scalable and simple SOA approach to distributed computing that’s applicable for many data and process intensive applications.

SpeedLink Architecture
SpeedLink offers a simplified SOA approach to building applications. A SpeedLink client is a traditional SOA client – it can be an application server, rich client, or other component that loads the SpeedLink dynamic libraries and invokes the simple DataSynapse Service API.

The host computers that make up the grid environment need only to have the DataSynapse Engine installed to take part in the distributed processing architecture. All service deployments, software upgrades, and configuration changes are managed by the central Broker. A fault tolerant configuration employs a failover Broker that communicates with the active Broker to perform automated failover. The overall grid architecture has no single point of failure and supports an always-on, always-responsive computing paradigm.

SpeedLink and Application Design Considerations
Adapting your application integration layer with SpeedLink is very similar to application integration with standard GridServer services. While SpeedLink has lower latency and higher throughput, standard GridServer services still have their own advantages such as greater horizontal scalability and more data handling options. An example of a suitable SpeedLink application is a trading application, where portfolios, trades, and strategies would need to be constantly re-evaluated based on market conditions delivered from a market data feed. Because rates might change dozens of times a second, low latency is preferred over horizontal scalability.

Faster Services in Data and Process Intensive Applications
SpeedLink allows services to be created quickly in any language and deployed and accessed by applications with ease and simplicity. The unbounded throughput and low latency makes SpeedLink and ideal choice for creating new distributed computing systems where speed, scale, and performance are critical factors