Leveraging HBase for a scalable distributed In-Memory OLAP/HTAP solution

The financial industry has been looking for a new technology that is able to continuously deliver risk figures accurately. After 2008 crisis, stricter regulation is being imposed on the financial industry, including intraday calculations and higher flexibility.

In this context, most tier-1 banks have identified Hadoop as a good candidate for delivering these high speed risk analytics. In reality IT departments face a too rich offer and too high costs of integration for building a consistent system that is actually able to deliver continuous Risk Computation.

Leveraging HBase

HBase offers a standard, consistent storage solution that gives very powerful capabilities to Scaled Risk OLAP cube. The distributed architecture of HBase provides true scalability that has been experimented up to several thousands of servers. Adding servers will enable more parallel processing capabilities to build cubes and then maintain up to date in real-time.

Automated data sharding and load balancing

Data sharding is the ability to split tables into several subsets of rows that are managed in parallel on several machines. Scaled Risk uses dynamic and flexible data sharding, which means that adding extra resources is as simple as adding extra nodes to the existing cluster.

The distributed capabilities that Scaled Risk offers must not get mixed up with distributed solutions that rely on static sharding, which always require huge deployment and administration efforts.

Distributed computation

Beyond data storage distribution, data processing is executed on each node for maximum parallelization and minimum network usage. Unitary computations and global aggregation never use locks thanks to the innovative architecture.

Caching capabilities

While data are persisted on disks, Scaled Risk enables huge cache on reads, which are enabled by each node RAM. It is actually a distributed cache that can be filled with various policies depending on data usage: cache on read, cache on write, mixed approach. A small cluster of 10 mid-range servers enables 1.2 TB of distributed cache. 

 

By leveraging HBase, Scaled Risk offers a disruptive and powerful OLAP/HTAP solution, at the crossroad of transactional data management, Business Intelligence, Big Data and real- time data processing.

Compared to other available solutions, Scaled Risk offers a unified platform:

  • Easy connectivity and built-in storage
  • Full scalability, so OLAP cube capabilities can be extended by adding extra resource (CPU, memory or HDD/SDD)
  • Low latency real-time architecture based on push technology from data input to the end-user desk
  • The best service continuity on the market with no need for active-passive architecture
  • Advanced built-in aggregations functions and an open API to integrate custom functions
  • Off-the-shelf grids and charts
  • An open and unified service layer to access all Scaled Risk Features 

Read more and discover how Scaled Risk leverages HBase to offer the best scalable distributed In-Memory HTAP solution on the market. Download our complete technical paper on Scalable Distributed In-Memory Analytics Platform.