A Predictive Performance Model to Evaluate the Contention Cost in Application Servers
- CSIRO
- BATTELLE (PACIFIC NW LAB)
In multi-tier enterprise systems, application servers are key components that implement business logic and provide application services. To support a large number of simultaneous accesses from clients over the Internet and intranet, most application servers use replication and multi-threading to handle concurrent requests. While multiple processes and multiple threads enhance the processing bandwidth of servers, they also increase the contention for resources in application servers. This paper investigates this issue empirically based on a middleware benchmark. A cost model is proposed to estimate the overall performance of application servers, including the contention overhead. This model is then used to determine the optimal degree of the concurrency of application servers for a specific client load. A case study based on CORBA is presented to validate our model and demonstrate its application.
- Research Organization:
- Pacific Northwest National Lab. (PNNL), Richland, WA (United States)
- Sponsoring Organization:
- USDOE
- DOE Contract Number:
- AC06-76RL01830
- OSTI ID:
- 15004489
- Report Number(s):
- PNNL-SA-37133; TRN: US201015%%667
- Resource Relation:
- Conference: 9th Asia Pacific Software Engineering Conference , 435-440 ; IEEE Press,Piscataway, ,United States.
- Country of Publication:
- United States
- Language:
- English
Similar Records
Reaching bandwidth saturation using transparent injection parallelization
Reaching bandwidth saturation using transparent injection parallelization