Recently we have compared Application Server Middleware performance on Solaris SPARC S7-2 Server and x86 servers. I am not going to share the model and brand of the x86 system. But it is for sure that the two compared servers are in the same segment/class.
Here is the quick SPARC S7-2 CPU Benchmark Results on clients Datacenter:
SOLARIS SPARC S7-2 Server CPU Load Test Results on client Datacenter
Socket Load Test Results:
Thread execution times increasing continuously which means server response times are increasing and this is a bad situation for Middleware systems under load.
SPARC S7-2 Socket Load Test Result (Tuned and Fixed Load Test (3 seconds))
Our scenario is consuming socket for 3 sec. with 32 thread at the same time. Below chart shows that there is no slowness and every thread executed as expected with 3 sec.
Real Application Load Test with JMeter
SPARC throughput is 1312,3/sec and x86 is 789,3/sec which means SPARC has 1.66x better TPS performance.
CPU Utilization on Application Server (WebLogic) Load
x86 CPU usage increased to about 85%, SPARC S7-2 CPU usage increased to max 15% usage.
According to M7/T7 SPARC series, minimized version of SPARC S7-2 server could be used as consolidated server for Application Server infrastructure.
Conclusion
In general, your hardware configuration could be best, but it is for sure your software and Application Server container must be tuned for better performance. If you’re having performance problems about your application, upgrade your infrastructure with an expert with Application and Application Server tuning view.
The whole hardware configuration is not listed in this benchmark research because of NOT to target a brand or model as a weak performance criticism. If you want the whole hardware infrastructure configuration, please contact to me and drop a mail.
Tags: Java Server Loader (JSL)
Sparc S7-2 , Sparc T7/M7 Server, x86 Servers, Solaris Sparc, WebLogic Performance Tuning, WebLogic Load Test, WebLogic Migration, WebLogic Consultancy
Leave a Reply