Monitoring and Accountingfor AAI
- Courtesy of RAPTOR
Nicole Harris (JISC Advance)but really
Rhys Smith, Cardiff University Phil Smart, Cardiff University
2
Most Important – Logo!
3
RAPTOR - Background
• In UK, Athens -> Shibboleth = loss of usage stats– and… stats increasingly important in this age of
austerity• Alongside this, institutions have multiple systems– Shibboleth / OpenAthens / Ezproxy / IP auth / etc
• Institutions typically require relatively in-depth stats– E.g. E-resource usage by Department
4
RAPTOR - Background
• Alongside this:– UK federation has a need for stats to prove value for
money to funders;– Typically needs only simple “headline” stats;– What opportunities are there in shared data vs. local
data?
5
RAPTOR Goals
• Easy to install & configure;• Not intrusive - must not touch existing systems
except on log file periphery;• Web front end designed for non-technical users;• Scalable;• Standards based where possible;• Free to use, open source, community driven.
6
Raptor Components
• Client– ICA (Information Collector Agents)• Java log-file parser• Sits alongside the service to monitor
• Server– MUA (Multi Unit Aggregator)• Java front-end to event database
– Web• Java servlet, connects to MUA and allows user to
interact with events
7
8
Technical Details
• Java enterprise edition software• Web container
– Embedded Jetty for standalone deployment on all components
• Web application (RaptorWeb)– Embedded Jetty for standalone deployment
• Java Server Faces, with Richfaces components (AJAX)• ICA and MUA are self contained web services configured
using Spring• Uses Apache CXF• Springs inversion of control feature allows config in XML
files
9
Technical Details
• All communication between components over TLS, mutual authentication for security & integrity, encryption;
• Currently manually configured keys (but installer helps take out the pain);– Future – use SAML metadata to ease this.
10
Supported Platforms
• Platforms– Linux
• YUM repository through openSUSE Build Service• Suse, RHEL, Mandriva variants, Debian/Ubuntu soon
– Windows• MSI installers
• Packages for– ICA / MUA / Web– MUA & Web (preconfigures MUA-Web)– ICA & MUA & Web (preconfigures everything)
• Requires only Java– Postgres optional (can use built in HSQL DB) but highly recommended
11
12
13
14
15
16
17
Activity Data overview• What resources are involved?
– Whatever you like! Typically aimed at library resources for now. • How is the data extracted or collected?
– As per previous slides.• What can the data tell us?
– High level analysis of what is being used. • Is it useful?• Can it be aggregated with other activity data?
– Issues with standardising on resource names / entityIDs.– Will be aggregated with Monitoring Unit information. – Linked to LDAP for more detailed analysis. – Local vs. aggregated view.
• Are there legal issues (DPA or rights)?– Depends on level of information aggregated and whether it is PII ‘free’.