The RAP repository uses a vectorial representation of data and supports multi-user and multi-application workloads by scaling up as well as scaling out in multi-processor, clustered configurations. RAP captures real-time data flow in both the cache database and the repository and maintains a cached-copy and a disk-based copy; the repository stores the union of both the historical and intra-day data as one complete time series. The latency of the disk-based repository relative to the cache is kept in seconds.
RAP supports different applications, such as pre-trade analysis, post-trade analysis, quantitative modeling, and scenario-based back testing against a shared database, and distributes the query workload across the cache and the repository based on the time-criticality of the data access requirements of each user application.
The design is based on the requirements of high performance and concurrent retrievals by a large user population, as opposed to the design needs of a high rate of concurrent updates as in the case of RDBMS. The internal data structure and the manner data is served to user applications renders the RAP repository particularly suitable to store large amounts of time series data. The next generation automated trading infrastructure in institutional trading and prime brokerage firms is characterized by a limited number of inbound data streams (that represent market data delivery channels) and a high number of concurrent reader processes that access indexed columns to retrieve large data sets for analysis. The architecture of the Risk Analytics Platform meets these requirements.
| Copyright © 2005. Sybase Inc. All rights reserved. |
|
|
