EclipseLink/DesignDocs/298985

Design Specification: Performance and Concurrency

ER 298985

Feedback

Document History

Date	Author	Version Description & Notes
2010-01-06	James	0.1 Draft
2010-01-206	James	0.2 Updated CacheId

Project overview

This project groups several smaller performance related bug fixes and enhancements into a single unit. Its' goal is the improve the performance, concurrency and scalability of the product.

Concepts

Performance is concerned about reducing CPU usage and finding more optimal methods of processing operations.

Concurrency is concerned with reducing contention and improving multi-threaded and multi-CPU performance.

Scalability is concerned with clustering, large workloads and data.

Requirements

The goal of this project is to ensure that our product remains the leading high-performance persistence solution. Areas of improvement are determined through performance comparison with other persistence products and benchmarking.

Specific performance investigations desired for this release:

• JPA performance comparison with EclipseLink 2.0
• core performance comparison with EclipseLink 2.0
• JPA concurrency comparison with EclipseLink 2.0
• JPA provider and app server comparison through SPECjAppServer ® benchmark.

Design Constraints

The goal of the project is to improve performance of common usage patterns. Fringe features and usage patterns will not be specifically targeted unless found to be highly deficient.

Any optimization must also be weighed in its' impact on usability, and spec compliance. Optimizations that may have a large negative impact to usability may need to be only enabled through specific configuration.

Functionality

Each specific performance improvement is discussed separately below.

Building objects from ResultSets

There is currently an old prototype of building objects directly from ResultSets. The goal of the feature is to allow "simple" objects and queries to be able to bypass the intermediate DatabaseRow objects build from JDBC used to build objects. Also to avoid a lot of the checks for non core features and simplify the object building process.

This will introduce a second path on queries and object building for these optimized queries, that should avoid a lot of the general overhead required to support advanced features. The feature will only be used on a query, or perhaps a class through configuration or if the class/query are determined to be "simple".

Initially simple will only include direct mappings, but hopefully be expanded to include single primary key relationships, or perhaps composite primary keys. It will not include inheritance, events, complex queries, fetch groups, etc.

Singleton cache keys and cache refactoring

Currently cache access can be an expensive operation. This will be improved by simplifying the CacheKey. For singleton primary key objects, the Id value (Integer, Long, String) will instead be used as the cache key. This will have a very broad impact as it changes the usage of Vector for the primary key, to be of type Object. CacheKey may then need to be refactored into a CacheValue. CacheKey will still be used for composite primary keys.

This could also allow for the JPA IdClass to be used as the cache key through configuration.

This would affect a lot of internal API, as well as some external API that currently is typed to Vector. The public API taking Vector could still be supported, but the API returning Vector would either need to be changed, or new methods added and old ones deprecated.

Testing

Both the existing performance and concurrency tests and pubic benchmarks will be used to monitor and evaluate performance improvements.

Specific performance testing desired for this release:

• JPA performance comparison with EclipseLink 2.0
• core performance comparison with EclipseLink 2.0
• JPA concurrency comparison with EclipseLink 2.0
• JPA provider and app server comparison through SPECjAppServer ® benchmark.

API

Singleton cache keys and cache refactoring

(old API is still supported, but deprecated)

• IdentityMapAccessor
  • *(Vector, Class) -> *(Object, Class)
• Session
  • keyFromObject(Object) -> getId(Object)
• ReadObjectQuery
  • get/setSelectionKey(List) -> get/setSelectionId(Object)
• ReportQueryResult
  • getPrimaryKeyValues() -> getId()

Open Issues

Issue #	Owner	Description / Notes
1	Group	What is the impact of the cache refactoring on Cache interceptors integration?
2	Group	What is the impact of the cache refactoring on backward compatibility?

Decisions

Issue #	Description / Notes	Decision

Future Considerations

Continually improve performance.