RTI Data Distribution Service Outperforms Other Messaging and Integration Technologies

RTI Data Distribution Service provides significantly higher performance than other messaging and integration technologies such as JMS and Web services. Compared to JMS, RTI Data Distribution Service provides up to 25x faster performance (see the benchmarks).

RTI outperforms other middleware solutions in several ways:

• Low latency – the amount of time it takes a message to be delivered from an application running on one node to an application running on another
• High throughput – the aggregate amount of data or messages that can be sent in a given time period, e.g., messages per second.
• Low overhead – the burden placed on network and processor resources

RTI Data Distribution Service performs so well because it has an efficient implementation that was designed specifically to meet the requirements of high-performance real-time applications. It has been optimized based on ten years of real-world experience with hundreds of demanding applications spanning a broad range of industries and platforms.

Following are some of the architectural characteristics and features that provide RTI Data Distribution Service with its extraordinarily high performance.

True Peer-to-Peer Communication

RTI Data Distribution Service is built on a true peer-to-peer architecture, with direct end-to-end communication between message senders (publishers) and receivers (subscribers).

Unlike most other messaging solutions, RTI Data Distribution Service does not depend on a central server to relay messages. The use of a server adds latency, since all messages must go through an intermediary. It also introduces a potential bottleneck and single point of failure, since all messaging traffic depends on the network links and processing resources allocated to the server.

In addition to minimizing latency, peer-to-peer communication minimizes network overhead. Messages do not have to be sent first to the server and then to their intended destination. They can be sent only once—directly to the ultimate recipient(s), reducing network traffic.

Peer-to-peer communication also provides much more cost-effective scalability. As new nodes are added to a system, there is no central server that has to be upgraded in order to accommodate the added traffic.

Efficient Networking with Multicast

Most messaging products use TCP for their underlying network transport. RTI Data Distribution Service gains a performance edge by using uses its own highly efficient messaging protocol built on UDP.

RTI’s wire protocol is higher performance than TCP, requires less network traffic, and does not incur the overhead associated with establishing and managing TCP connections. In addition, RTI’s protocol can take advantage of multicast when available, for both reliable and best-effort messaging. TCP only supports unicast.

With multicast, messages that are destined for multiple nodes can be sent across the network just once. They are received by all recipients simultaneously. RTI Data Distribution Service allows multicast addresses to be assigned on a per message-topic basis. As a result, the underlying network packets used to send a message are only received by nodes on which there is a corresponding subscriber.

Multicast is superior to broadcast, since broadcast messages have to be received and processed by all nodes regardless of whether they host a subscriber or not. This introduces gratuitous processing overhead to all nodes that are not interested in the message being received.

Multicast provides much more scalable performance than unicast. With unicast, messages have to be individually sent to each subscriber. For messages that are being broadly distributed, the need to send the same message multiple times with unicast puts a substantial demand on the network. It also introduces significant latency for the unfortunate final recipient.

Highly Configurable Quality of Service

RTI Data Distribution Service supports over 20 configurable Quality of Service (QoS) options that allow its performance to be optimized based on the requirements of individual applications or message topics. For example, performance can be optimized to maximize throughput or to minimize latency, or based on whether best effort or reliable messaging is required.

Filtering

RTI Data Distribution Service is message-content aware and allows subscribers to specify filter expressions that are applied to each message before it is sent. By filtering unneeded messages in the middleware layer on the sending side, overhead is reduced on both the network and receiving node. In addition, application logic may be simplified if message recipients no longer have to analyze each message for relevance.

Performance-Optimized Core

The RTI Data Distribution Service core is implemented in “C’ for high performance with minimal processor overhead. Other language interfaces, such as Java and C++, are implemented on top of the C core. Applications are fully interoperable across languages; that is, C applications can seamlessly communicate with Java applications. This allows performance-critical code to be implemented using the highest performance language available.

Efficient Data Path

Unlike most other messaging interfaces, RTI Data Distribution Service allows messages to be sent and received without any copying of data between the application and middleware. By eliminating a copy, RTI Data Distribution Service reduces latency and processor overhead, particularly for large messages such as high-bandwidth streaming data.

Request more information