SAP HANA as an Operational Data Store

Posted by Robert Klopp on August 31, 2012

More by this author

I was asked today if I had “any view on whether the Operational Data Store (ODS) is a dedicated node or is a just a set of processes within a shared machine (RK: shared with EDW and data mart workloads) and prioritised using workload management?

An ODS is an Operational Data Store which is sort of an intermediary between OLTP transactions and data warehouse queries to handle transactions that are too long-running for the OLTP system and too quick for an EDW. So if a single system could handle both OLTP and data warehouse transactions then there would be nothing to distinguish an ODS.

In a perfect scenario there would be one set of tables solving OLTP, ODS, data warehouse, and analytics queries. There would be no redundant data (except to solve for high availability). Figure 1 shows three workloads querying against one database instance on a single server or cluster.

In a nearly perfect scenario there would one set of processes on a shared platform with redundant data for each workload. Figure 2 represents three database instances, each with basically the same information, running on a single server or cluster. Note that it is often the case that each DBMS instance may run a different, specialized, DBMS product which is adding to the complexity.

In the real world these things tend to run on separate systems. Figure 3 depicts each workload on a separate server or cluster with separate database instances and with all of the costs associated with data movement between systems, floor space, power, and systems administration.

Blog20120829 Perfect World.png

I think that the near-perfect world is nearly achievable, but the workload management problem has not been solved to everyone’s satisfaction.

SAP HANA is designed to create a perfect scenario. In Release 1, engineering targeted data warehouse and data mart workloads. In subsequent releases we will support bigger and bigger data and OLTP. We already see less data redundancy and reduced schema complexity as aggregate tables, which hold derived redundant data, are eliminated.

It is an important distinction. No other DBMS is designed to support all of these workloads simultaneously. SAP HANA is very special in this regard.

VN:F [1.9.22_1171]
Average User Rating
Rating: 0.0/5 (0 votes cast)