Introduction to WebSphere MQ Low Latency Messaging
- WMQLLM is peer to peer messaging product. Messages directly flow between publisher and subscribers. There is no daemon that is performing routing or broker task. WMQLLM is designed to deliver ultra latency with very high volume of message.
- WMQLLM product was originally generated for financial market like stock market or any type of financial market where there is demand for Ultra low-latency & high-throughput of messages. In financial market, there will be large volume of market data that has to be handled. Market data allows traders and investors to know the latest price and see historical trends for instruments such as equities, fixed-income products, derivatives and currencies. Delivery of price data from exchanges to users is highly time-sensitive. Here there is need to handling large volume of data with predictable latency. WMQLLM can handle large volume of data with very low latency.
- Based on patented technology from IBM Haifa Research Lab
Capabilities of WMQLLM
Highest Performing Messaging Middleware:-
WebSphere MQ Low Latency Messaging technology provides highest message throughput and lowest networking latency. WebSphere MQ Low Latency Messaging technology is proven in industry benchmarks, such as the STAC M2 benchmark, to provide ultra low latency and high throughput messaging.
WMQLLM uses RDMA technology for InfiniBand and 10 Gigabit Ethernet support. It uses shared memory support for interprocess communications.
Performance benchmark on throughput shows 98 million messages per second on native infiniband and shared memory. On 10 Gigabit Ethernet it provides throughput of 75 million messages per second.
Performance benchmark on latency shows 770 nano seconds for shared memory and 2 microseconds on Native Infiniband. On 10 Gigabit Ethernet it provides latency of 3.6 microsecond.
High Availability with No Performance Compromise:-
Without sacrificing the performance, WMQLLM provides reliable and consistent message streaming (RCMS) for high availability feature.
WMQ LLM supports Reliable Multicast Messaging(RMM) which provides reliable and high performance for one-to-many data delivery or many-to-many data exchange for publish/subscribe model of messaging. RCMS provides a layer of high-availability and consistent ordered delivery on top of the qualities of service provided by RMM layer.
WMQLLM has capability for Fast failure detection i.e in sub-second and recover the failure with zero message loss.
Ease of use features for rapid development and deployment:-
Following are few features listed that makes it easier to develop application and deploy them quickly in heterogeneous environment:-
- WMQLLM API's support different types of network and different types of message distribution like unicast, multicast, shared memory. WMQLLM allows applications to send or receive messages to multiple network over multiple transport within a single transmit or receive instance. For example, a transmitter application can create a single multi-protocol transmit instance that sends packets over TCP and UDP. Or a receiver application can create a single multi-protocol receive instance that receives packets over UDP and the InfiniBand verbs transport.
- WMQLLM has Flexible centralized configuration and monitoring support.
- WMQLLM provides a wide range of configuration for reliabilities ranging from unreliable to Assured Messaging.
Focus on Monitoring that Matters
WMQLLM has built-in monitoring service which provides rich set of performance and latency statistics. This statistical information can be used for troubleshooting and identifying the application bottlenecks or any problem in customer environment.
WMQLLM product also provides the framework for a monitoring plug-in. A loadable module can be developed that can be used to integrate the monitoring facilities of the WMQLLM with any third-party applications or in any existing monitoring infrastructure.
Congestion Management to Ensure System Stability
WMQLLM provides Congestion management features that allow the WMQLLM transmitter application to automatically detect and react to application that are not keeping up to the message rate environment either due to slow receiver or network congestion
WMQLLM provides configuration parameters associated with retransmissions. A retransmission is the resending of packets which have been lost, damaged, or received out of order. Using advanced configuration provided by WMQLLM will help in controlling Retransmission of packet. So congestion management will help us reduce the impact of poor performing applications or network components in the environment.
Message Filtering to Help you Scale
WMQLLM provides message selctor features using which message can be filtered. The two common methods provided by WMQLLM for filtering message is through use of message properties and turbo flow. Message filtering gives control over performance.
Supported Platform and languages
Platforms:- Linux and Windows on x86 and x86_64,Linux on pSeries & zSeries, Solaris (x86, x86_64,SPARC), HP-UX on IA64, AIX on ppc64.
Languages:- WMQLLM supports C, java and .net API’s .
The performance numbers listed in this article are based on the IBM’s test environment setup.