Citigroup Data Service System Based On Distributed Cache

There are lots of financial products for all kinds of investors in Citigroup. Before an investor starts to buy a product, he usually evaluates the risk of the investment which is provided by Citigroup risk systems. Originally these risk systems have dependency on traditional DB technology, but DB can’t support so huge concurrent access from grid computers, so it’s necessary to development new data service system to satisfy the grid computation requirement. It’s named Data Service Platform (DSP), which is leveraging the latest advanced distributed cache technology. In DSP, data is stored on distributed caches in different JVMs on different machines, which totally provides TB level in-memory data storage, and what’s more, the distributed cache nodes also provide load balance and fault tolerance functionalities.First of all, it’s illustrated that the importance of risk calculation and incapability of current risk systems and data service systems, which draws the conclusion that it’s very important and necessary to develop DSP. Next, functionalities of Data Service Platform are discussed, including data publish, data subscription and data retrieval as well as system management. Then, the main working flow of data publishing, data subscription and data query are analyzed. And Gemfire Enterprise Fabric based Data Service Platform is analyzed also to provide load balance and fault tolerance capabilities. Finally, the most importance subsystem including distributed cache partition subsystem and data service subsystem are designed in detail, and at the same time, the system monitor console and system management console are introduced as well.