What should we measure?All applications should have the following metrics in common:Number of Users - by community (department, MIT affiliation, etc)
- total unique users
- number of user accounts
- number of user sessions/logins
- by time period (peak usage, average usage)
System Utilization - CPU
- memory
- disk space
- network
Application Specifics (metrics here vary from applications to application) - wikis (automated)
- DW (automated)
- Roles (automated)
- Jira
- Jabber
- RT
- Web Servers (Apache, etc)
- Web Services (UAWS)
- Thalia (manual)
- Stellar (manual)
- Touchstone unique users by app and logins by app (should merge with DW to report by community)
- Developer Tools
- Kerberos (via RT, bug fixes, commits to tree)
How do we measure them?Getting information into the Data Warehouse - in a standard format
- except system utilization
SAMM would provide a continuous view of these stats - show trends
- how app behaves (learn more about environment)
- blending of stats (users, memory, spikes)
What do we have right now?What requires more work?- see grid above for whats missing
- how do we allocate resources?
|