Friday May 09, 2008

DTrace Probes for Databases

Robert Lor has listed potential standard DTrace probes for databases and some PostgreSQL-specific probes.

Monday Jul 09, 2007

Database Price-Performance on Sun All Open-Source Platform

More good PostgreSQL and open-source news! 

In the first week of July 2007, Sun announced a very attractive SpecJ2004 result for an all open-source Sun stack, including PostgreSQL on Solaris on Niagara.

Josh Berkus and Jignesh Shah have already written about the recent SpecJ benchmark results

The highlights are already given by Josh and Jignesh's blogs: Josh notes the importance of the results in proving the suitability of the Niagara architecture for DB applications and the importance of this result as a proof of SMP scalability. He also notes the significant price difference between Sun and the competition and looks forward to even better SMP performance by PostgreSQL database on Solaris. Jignesh gives some details regarding the DB tuning strategy used. If you want more of the tuning strategy details, you should probably leave him a comment.

Here's a summary of other highlights based on other sources:

  • This is the second all open source SPECjAppServer2004 benchmark result and Sun is the only vendor to publish all open source results.
  • It demonstrates Sun's commitment to use open source software at all levels of the stack -- including open source databases -- and to bring these price/performance benefits to users.
  • An all-Sun, all open source stack comprised of PostgreSQL on Solaris (built off OpenSolaris) on T2000s (with OpenSPARC) with Glassfish gave 89% the performance at 34% the cost of a comparable HP benchmark with proprietary database, application server, and hardware. (Tom Daly provides further details regarding price-performance results.)

If you do not know about SpecJ 2004, refer to spec.org. In summary, SPECjAppServer2004 heavily exercises all parts of the underlying infrastructure that make up the application environment, including hardware, JVM software, database software, JDBC drivers, and the system network. The primary metric of the SPECjAppServer2004 benchmark is jAppServer Operations Per Second ("SPECjAppServer2004 JOPS") in either @Standard or @Distributed mode.

Look, also, at Tom Daly's blog for more information on these performance benchmarks and more.

Disclosure Statement:
Sun Fire X4200 (6 chips, 12 cores) 778.14 SPECjAppServer2004 JOPS@Standard.
HP rx2660 (2 chips, 4 cores) 874.61 SPECjAppServer2004 JOPS@Standard.
SPEC, SPECjAppServer reg tm of Standard Performance Evaluation Corporation.
Results from www.spec.org as of 7/10/07.

About

MortazaviBlog

Search

Archives
« April 2014
SunMonTueWedThuFriSat
  
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
   
       
Today