Open Source RDBMS - Seamless, Scalable, Stable and Free

한국어 | Login |Register

Current Events
Join our developers event to win one of the valuable prizes!
posted 7 years ago
viewed 5489 times
Share this article

NBD Benchmark: CUBRID Performance Test Results


Recently I came across to a wonderful report created about two years ago right after CUBRID announced to open its source code and join the ranks of the global open source projects. The report is about the results obtained after the CUBRID Database System had been tested against the NBD Benchmark by NHN Corporation, the largest corporate user of CUBRID Database with a farm of over 10,000 servers. So, I decided to publish this article which would summarize those test results.

The NBD Benchmark stands for the NHN Internet Bulletin Board Application Database Benchmark. The test had used one of the first open source releases of CUBRID version R1.1 to compare with the three other database systems: two were commercial, and one - open source. This was the first Benchmark test when CUBRID had to assert its rights to be called an open source relational database management system optimized for Web applications.

The test scenario covered the characteristics of the Bulletin Board System type of web services. The results had to reveal the following variables:

  • the Performance in TPS (Transactions Per Second)
  • the Relationship between Database Sizes and the Performance
  • the CPU Usage for each Database System
  • the Disk I/O Performance
  • the Memory Usage
The performance results were measured on identical hardware configurations, using variables such as database size, workload increase, and cache functionality provided by each DBMS.

At that time as a result CUBRID R1.1 illustrated the second highest database system performance. It consumed below 30% CPU resources in the small DBMS benchmarking, which was less than the commercial database systems did. This indicated that CUBRID had a whole lot of a potential for even higher performance results if suggested 70% of the server load is maintained considering the same database size.

Follow the link below to read the full report

comments powered by Disqus