Repository Analytics: Helping to make informed decisions
By ckamps on Sep 15, 2009
Over the past several months our Update Center project has been rounding out our overall pkg(5)-based content delivery solution by providing usage reports to adopting project teams: GlassFish Application Server v3, Web Space Server, Web Stack and soon, Message Queue.
You may be familiar with the term web analytics, defined as the "analysis and reporting of internet data for purposes of understanding and optimizing web usage" by Wikipedia. In the context of delivering content via network package repositories, the Update Center team has deployed the beginnings of a reporting solution that helps product teams improve their offerings by providing greater insight into how add-ons and updates are being adopted by end users.
Over the past several months, our toolkit project has:
- Defined a database schema and deployed a MySQL database to manage the usage data
- Developed and deployed scripts to tranform raw repository access log data and to load it into the database
- Developed an overall dashboard and an initial series of product-specific reports
- Started tweaking reports based on feedback from users
Project teams can currently obtain the following information via the reports:
- Active unique installations (i.e. "active users")
- Add-on (new) packages installed
- Updated packages
- Operating system usage
- Client tool usage (Update Tool GUI, Software Update GUI, pkg(1) CLI, pkg(5) Java API)
- Add-on activity by registered installations
Additionally, we report on the versions of the toolkit that are in use across all installations. This information is a real boon to our toolkit development project in that it lets us make informed decisions about how much we invest in older versions. It also helps us monitor the roll out of new releases of the toolkit.
Over the next several weeks we'll publish new blog entries that explain the reports and how projects are using them to improve their products and ultimately the experience of their users.