Complexity of Oracle platforms

 

The performance of Oracle platforms is not predictable

Oracle platforms have a very high complexity. Engineers and system architects have a lot of options and alternatives to implement each layer of an Oracle platform. There is an almost endless number of combinations possible for technologies, components and configurations. Each decision may have impact to the Oracle database performance.  But nobody know exactly how much. The only way to calibrate the performance of an Oracle platform is to run a benchmark.

Weaknesses of some benchmark tools

 

Weakness of industry benchmarks and proof of concepts

A lot of benchmark tools have been developed in the past. Some of them are too complex to run them in the environment of the customer. Some of them provide metrics which are not readily understandable. Even proof of concepts with 5 or 10 years old application software have a limited informative value. Proof of concepts may be helpful to become familiar with new technologies but they are in general inappropriate to test performance limitations of new technologies and therefore are inappropriate for price performance considerations.

Key Performance Metrics

 

 

 

 

Key Performance Metrics for Performance Calibration

The Benchware methodology is based on readily understandable key performance metrics. Benchware runs performance tests to identify key performance metrics for processor, server system, storage system and database system. The benchmark load is generated by representative Oracle database operations. Key performance metrics are useful in different situations: performance calibration (quality assurance), platform evaluation and capacity planning.

Requirements to Benchmark Tool

 

Requirements to Database Platform Benchmark Tool

The Benchware Performance Suite (BPS) is an Oracle platform centric benchmark tool. It generates readily understandable key performance metrics for CPU, Server System, Storage System and Database System. This presentation describes the requirements to database platform benchmark tools in more detail.

CPU performance benchmark tests

 

How to benchmark CPU performance

CPU performance has a huge impact on the performance of most database operations, and of course on Oracle license and maintenance cost. Benchware uses Oracle data types, simple operations on these data types and Oracle SQL built in functions to calibrate CPU performance of Oracle database server.

SRV performance benchmark tests

 

How to benchmark Server performance

Applications tend to operate in memory as much as possible to avoid slow I/O operations.  Benchware uses representative in-memory SQL transactions for OLTP and analytic applications to calibrate server performance of Oracle platforms.

Benchware server performance tests are useful to verify, whether 2 or 4 socket server with Oracle Standard Edition and lower license cost deliver sufficient performance.

STO performance benchmark tests

 

How to benchmark Storage performance

Storage performance is essential not only for overall Oracle database performance, but also for system management tasks like backup, recovery and archiving. Oracle uses all kinds of I/O patterns, and different o/s calls dependent upon operating system and system load. Database I/O is much more expensive than native Storage System I/O (buffer cache management, concurrency). Benchware developed representative Oracle SQL performance tests to measure fundamental key performance metrics of storage systems in Oracle environments.

DBL performance benchmark tests

 

How to benchmark Database load performance

Database load performance is important for both: OLTP nd DWH systems. System Architects need understandable key performance metrics for capacity planning, e.g. the commit rate respectively the transaction rate or throughput of bulk loads to estimate the necessary time window. These kind of benchmark tests are useful to verify the impact of the disaster recovery infrastructure to application performance, when using DataGuard (database mirroring), ASM (host based mirroring) or an storage based mirroring solution.

DBX performance benchmark tests

 

How to benchmark Database OLTP performance

Database OLTP performance is important for all ERP, CRM, Core Banking and similar kind of applications. These benchmark tests deliver representative key performance metrics for capacity planning:  throughput and service times for light weighted transactions.