To achieve resiliency at the feed process level, TPS+Plus runs multiple hot-hot copies of all feed handlers (i.e. Exchange Servers) which are then load balanced with the system. Feed handlers are all designed to manage services, route requests and responses, and provide load balancing, fan out, and fail-over. Within a given system, two or more Exchange Server instances are typically deployed, each processing both connections and performing line arbitration. While all server instances for a given feed are available, the TPS+Plus middleware infrastructure load balances client request for data across the number of Exchange Server instances providing the data being requested. In the event a connection or Exchange Server instance becomes unavailable, the middleware infrastructure automatically re-routes requests to the remaining Exchange Server instances providing the same data.

 

The distribution tier can also be implemented as a flat multicast architecture, with Exchange Servers directly publishing to reliable multicast channels. These multicast channels can be configurable and are load balanced to allow for partitioned distribution of multicast subjects. InfoDyne provides its own multicast capability or can integrate natively with several vendor middleware products based on customer requirements.

 

Load Balancing allows exchange feeds to be processed multiple times within a given system providing redundancy, failover, and request/response scaling. For redundancy purposes, data feeds are typically sourced through multiple connections (e.g. CTA's A&B networks). While all server instances for a given feed are available, the TPS+Plus middleware infrastructure load balances client request for data across the number of Exchange Server instances providing the data being requested to the user. In the event a connection or Exchange Server instance becomes unavailable, the middleware infrastructure automatically re-routes requests to the remaining Exchange Server instances providing the same data.

 

Furthermore, for all applications in the system, including any custom applications clients may develop and wish to integrate within the system, TPS+Plus can run multiple hot-hot copies

 

In regards to resiliency across multiple installations, or a regional deployment, of locally sourced data feeds, execution venues, infrastructure, and client applications, each regional deployment is capable of operating in a completely independent manner, with all components deployed in a redundant and fault-tolerant configuration. Regional deployments communicate with other regions through interactive Gateways (ITOD/GW). These gateways will reflect services from a remote region on to the local system as if they were running locally, providing seamless application integration. Interactive Gateways only send/receive data that is necessary to/from the remote region, dramatically reducing the required bandwidth necessary between regions. Interactive access to data within a regional deployment is also supported, such that Exchange Servers only publish the data requested from them and customer applications only receive data they have requested, dramatically reducing the processing load on both the Exchange Servers and the customer’s applications.

 

Finally, TPS+Plus boasts a system architecture that allows for both horizontal and vertical scaling, both within the feed handlers and within the middleware domain, ensuring full throughput and a flat sub-millisecond latency response regardless of message rates.