What is PAGG ?

How is PAGG different from beancounters ?

Beancounters are attempting to provide resource limits on a per-user or per-group basis. This will ensure that a specific user or group cannot abuse the system and disallow other users from gaining fair access to system resources. Or, put another way, it politically manages the resources on the system.

PAGG is just a way to implement generic process containers. We are proposing PAGG so that we can provide a process container in the form of a job. The job provides a process group type that can be used for accounting and resource limits. And that is the difference. Job based resource limits are used to provide better workload scheduling on a system. This results in better throughput. So, jobs allow more efficient scheduling of work on the system.

Taken together, beancounters and jobs could be complementary technologies. But, using PAGG allows us to minimize the impact jobs will make on the core Linux kernel. In addition, PAGG allows other developers to leverage the use of process containers for different applications. Now, hopefully if I made any mistakes with regards to beancounters you will be willing to email me any corrections. I will update this FAQ with additional information as I learn more about beancounters.