This page is an explanation of the technology behind Sakai at Rutgers.

Sakai software

Sakai is an open-source collaborative environment for education and research. It was originally developed with support by the Mellon Foundation, at the Universities of Michigan, Indiana, Stanford and MIT. Since expiration of the grant, development work is being done by institutions throughout the world. For more information about the project see the Sakai project web site.

New versions of the Sakai software are currently being released once a year, timed for implementation for July or September. Rutgers is currently running version 2.9, an update of software that will be released in late Fall 2012. Rutgers moved to 2.9 in May, 2012.

Rutgers moved to 2.9 before the official release for two reasons: (1) The two major new features of 2.9 actually come from us. (2) The developers believed that even the early version we adopted was in better shape than the previous release. Our experience so far supports that judgement.

Sakai hardware at Rutgers

Sakai is written in Java, using current open-source tools such as JSF, Velocity, and Hibernate.

At Rutgers it is run on a cluster of 6 systems, of which 3 are in operation at any given time. The systems are actually "virtual machines" running on a set of 3 blades in a single HP C7000 blade chassis. Each blade has 72 GB of memory, and two 6-core Intel Xeon processors (X5650). Citrix Xenserver is used for virtualization.

In addition there is a database system running the Mysql database. (The software is MariaDB 5.1.12.) That system is another virtual machine in the same C7000 blade chassis. It uses two 6-core Intel Xeon X5675 processors, with 10 cores and 32 GB of memory  assigned to the database system. A second blade with the same configuration is available as a backup.

Storage for the database (as well as the virtual machines) is on a Dell md3600i iSCSI array.

User files (other than the database) are stored on a Network Applications file server, which is shared with other OIT systems. Both the Dell array and the Network Applications file servers have redundant controllers.

sakai.rutgers.edu is redundant pair of "load balancers," systems that accept connections from users and distribute them across the 6 front end systems, trying to keep the load evenly distributed. If one front end fails, users will be moved automatically to one of the other systems. However they will need to login again.