Unleashing the Performance and Scalability of Distributed Enterprise Applications
A distributed application's performance and scalability are fundamentally limited by the capacity of its underlying messaging infrastructure. A system cannot support higher throughput or lower latency than its messaging middleware supports.
RTI's real-time messaging infrastructure overcomes the constraints inherent in prior generation technologies, providing applications with unprecedented levels of responsiveness and scalability. Benefits include:
- Virtually unlimited scalability with no inherent limit on aggregate messaging capacity
- Extremely high application-to-application throughput in excess of 1,000,000 messages per second
- Very low latency - as low as 30 microseconds over Gigabit Ethernet
- Green operation with an extremely low total cost of ownership (TCO)
Architectural Innovations
RTI's 10x to 1,000x improvement in performance, scalability and cost-of-ownership are made possible by several architectural innovations:
Peer-to-peer messaging
Instead of routing messages through brokers or servers—the limiting choke point in other enterprise messaging and Enterprise Service Bus (ESB) solutions—RTI implements true peer-to-peer messaging. Messages are sent directly from producers to consumers with no intermediate processing.
- Scalability and Throughput - Without brokers, the network is the only point through which all messages pass. Thus, aggregate capacity is limited only by the network infrastructure. Hundreds of millions of messages per second can be supported.
- Latency - Latency is minimized since gratuitous network hops, software layers, and operating system context switches are eliminated.
- Green/Cost-of-Ownership - Significantly fewer servers are required because there is no need to host message brokers. In addition, network load is cut by at least 50% because messages are sent over the network once instead of twice. By reducing server and network infrastructure requirements, power consumption and rack space utilization are also minimized.
Reliable multicast
Messages can to be sent concurrently to multiple recipients. The switching and routing infrastructure delivers the message to those nodes (and only those nodes) that subscribed to the content. This enables extremely efficient broad data distribution.
- Scalability and Throughput - Publishers only have to send a message once regardless of the number of subscribers, maximizing the number of unique messages that can be sent over a connection.
- Latency - Messages are replicated and routed at wire speed by the switch, not serially in software. This allows every subscriber to benefit from equally low latency, no matter how many subscribers there are.
- Green/Cost-of-Ownership - Message routing and filtering are performed by the switch instead of by software, reducing processing overhead and server requirements.
Wire efficiency
The size of serialized messages can be reduced by over 90% compared to solutions that rely on self-describing, map and XML messages. Message fields can be pre-defined, eliminating the need for metadata (field names and types) to be embedded within each message. Data itself is serialized using a compact binary representation rather than as text.
- Scalability and Throughput - By reducing message size, more messages can be sent without increasing bandwidth requirements.
- Latency - Smaller messages minimize transmission time.
- Green/Cost-of-Ownership - Lower overhead reduces network capacity requirements.
Integrated data caching
Publish/subscribe topics can be linked to database tables, allowing applications to subscribe to database updates or publish data into databases (and to other subscribers). Subscribers can receive updates asynchronously or cache them for access when needed.
- Latency - Data can be accessed at memory speed rather than via a database query. In addition, updates can be received as asynchronous events for near instantaneous processing.
- Green/Cost-of-Ownership - Load on backend databases can be reduced significantly because updates are pushed once to interested subscribers (including via reliable multicast when there are multiple subscribers). Individual applications interested in the same data don't have to separately query the central database.
Learn more
RTI's real-time messaging solutions unleash the performance and scalability of your applications while reducing your server and network infrastructure costs.
RTI Message Service - The world's highest-performance JMS compliant messaging for distributed Java applications
RTI Data Distribution Service - Extreme performance messaging and distributed data caching for C, C++, Java, .NET and Ada applications
