Given the performance, capacity, and scalability of the Exadata platform, it is tailor made and an ideal platform for the Database Consolidation. Exadata platform allows consolidating databases with heterogeneous database workload on the Exadata Database.Database consolidation is a lucrative approach as it reduces the IT costs, reduces infrastructure complexity. However after database consolidation, it also introduces some challenges with regard to the following:
1) Can the system handle peak workloads, and can workloads co-exist together.
2) How to balance system resources among multiple workloads?
3) How to ensure that the critical system resources like CPU and I/O will be available to the most important applications?
4) How to guarantee the service level I/O within and among databases?
5) How to minimize consolidation risk?
Exadata implements three features to reduce such challenges in consolidating the databases.
- I/O resource manager (IORM)
- Database resource manager (DBRM)
- Instance caging
This paper focus mainly on the IORM feature which is unique to Exadata, and has the following characteristics:
1) IORM extends intra-database Database Resource Management (DBRM) features, categorizes and segregates I/O for individual databases as well as different I/O workload.
2) Enables storage consolidation for multiple databases regardless of their I/O consumption rates.
3) Manages and prioritizes the I/O at the storage cell, whereby each Exadata storage cell enforces these limits to prevent I/O resource starvation.
4) Uses Resource Plan to determine the order of I/O requests, maps the I/O request based on the attributes of the request and controls order that I/Os are issued to disk.
5) Automatically determines the best I/O optimization depending on the Database workload type (OLTP or DSS) by striking the balance between the disk latency and throughput.
Featured Speaker: Sameer Malik, Oracle Exadata/Oracle Database, Technical Architect, Accenture