This page last changed on Mar 21, 2007 by david.soul@atlassian.com.

The most significant factors affecting server load are peak user load, the editor to viewer ratio and total content.

  • Peak user load is the maximum number of client browsers accessing Confluence simultaneously
  • The editor to viewer ratio is how many users are performing updates to Confluence content versus making those only viewing content
  • Total content is best measured by counting total spaces

Minimum Requirements

Under 25 Online Users

  • 1GHz+ CPU Pentium 4 or equivalent
  • 256MB RAM

25 Online Users Or More

  • Dual 2.4GHz CPU Pentium Xeon or equivalent
  • 512MB+ RAM

Example Hardware Setups

These are actual hardware specifications for non-clustered Confluence instances.

Users Spaces Pages CPUs CPU (GHz) RAM (Meg)
150 30 1,000 1 2.6 1,024
350 100 15,000 2 2.8 700
5,000 500   4 3 2,024
10,000 350 16,000 2 3.8 2,024
10,000 60 3,500 2 3.6 512
21,000 950   2 3.6 4,048

Notes

  • It was not recorded whether the RAM refers to either total server memory or memory allocated to the JVM
  • Total pages is not a major consideration for performance. For example, instances hosting 80K of pages can consume under 512 meg of memory
  • Blank settings indicate that the information was not provided
  • Always use an external database

Maximum Reported Usages

Most Spaces 1700
Most Internal Users 15K
Most LDAP Users 100K
Most Pages 80K
Document generated by Confluence on Mar 22, 2007 20:56