This is an update as to what is happening with the 1.4 Perceus tree.

There has been a lot of focus on scalability (speed and handling of very large numbers of nodes), master node fault tolerance as well as a back-end rewrite allowing a much more abstract interface to the back-end database.

Using a multi-threaded/pre-forking Perceus daemon we are able to achieve very high node connection throughput. Initial tests show that Perceus should be able scale upwards of tens of thousands of nodes as long as there is a scalable architecture for Perceus to leverage (we can be and should be consulted for large scale implementations). There is also fault tolerance and load balancing built into the Perceus node client daemons (provisiond) as well.

The current development tree can also support multiple database back-ends (Berkeley hash, Berkeley BTree, and MySQL). Each has their pros and cons, and more importantly is the ability for the administrator to pick the model that best fits their environment and needs (facilitating the architecture, not dictating it is something Warewulf and Perceus has been about since day one).

If anybody is interested in early testing of Perceus 1.4, being included in announcements and status updates, or partnering with Infiscale.com please contact us directly using the Contact Us form on this page.

Thank you for supporting Perceus and Infiscale.com!