Featured Customer: Merrill Lynch

Scaling business with existing IT investment

Intro

One of the biggest problems with the data management of today is ongoing cost of hardware and software to scale business. As business throughput grows, the need for more hardware and software increases capital expenses as well as operational expenses and complexity. Conventional data management stack traditionally used to support low latency, high availability operations are not only expensive and complex, but fail to meet SLA requirements for mission critical applications. As an example, a traditional data management architecture for a global trading system consists of:

• An instance of the database in each major location (ie. New York, London, Tokyo)
• A second instance of the database in each location for business continuity
• SAN clustering for High Availability
• Storage level replication for disaster recovery
• A store-and-forward messaging product for global distribution
• A high speed messaging product for communication between components

Managing hundreds of gigabytes of operational data deployed to support 10s to 100s of applications on large clusters, grids and or private clouds while managing costs and minimizing administrative overhead can only be achieved through the deployment of technology like GemFire. GemFire enables enterprises to streamline their application architecture to take advantage of low cost commodity hardware solutions and to preserve existing investment in legacy infrastructure. Here’s is an example of how Gemfire was deployed to mitigate the cost of a conventional solution.
Benefit of GemFire

GemFire can achieve better performance, scalability and resilience, and simplify the architecture in the process above.

GemFire partitioned cache takes advantage of a pool of commodity compute resources to:

• Evenly load-balance across the cluster to improve performance
• Access data at memory speeds instead of database speeds
• Implement local replication (striping) for high-availability
• Sends notifications about changes to data directly to the interested components
• Provides lazy write-behind to the legacy data store for archival

GemFire’s WAN Gateway provides store-and-forward messaging and thereby:

• Distributes a consistent global view of trading activity around the globe
• Pushes data to the disaster recovery site with low latency, and without affecting the main trading flow’s performance
• Feeds the legacy data store in a single region for archival purposes