You are viewing an old version of this page. View the current version.

    Compare with Current View Page History

    « Previous Version 2 Next »

    Qube! makes use of a concept it calls "clusters". The clusters in a Qube farm are laid out in a tree-like structure, much like a directory tree where files are kept on disk. The familiar view of a directory tree is that it's "rooted" at the left, and branches out to the right.
    Clusters are organized like a directory tree, but instead of branching left to right, it's convenient to imagine them branching upwards, away from the root.
    Clusters are designated in the same way as a directory tree; the root is written as '/', and clusters in the root are written as /A, /B, and /C/D (/C is in the root, and D is "above" /C)
    In the same way that folders in a directory tree contain files and other folders, clusters can contain other clusters. Clusters can also contain Workers and jobs. This is the key to Qube's priority scheme.

    • Workers belong to 1 and only 1 cluster.
    • Jobs are submitted into a cluster; you can change a job's cluster after it's submitted, but it will always be in 1 and only 1 cluster at a time.

    The cluster membership of both a Worker and a job is considered when determining the overall priority of a job to run on a particular Worker. The basic rules are: 

    • A Worker will run the lowest-priority job that is in its own cluster before it runs the highest-priority job from any other cluster.
    • If there are no jobs in its own cluster waiting to run, a Worker will then run jobs from other cluster in decreasing priority.

    So jobs from "other" clusters have less priority than jobs from the same cluster. But there is even a precedence ordering for all the "other" clusters; jobs in clusters nearer a Worker's cluster are considered before jobs from clusters farther away. It "costs" a job in priority to run in clusters other than its own, and the cost increases as the distance between the job's cluster and the Worker's cluster increases. 
    Climbing "down" the tree (toward the root) is usually at no cost. However, a job submitted to /A/B/C has higher priority in /A/B/C, compared to /A/B, but that's just because the former case is in its own cluster. It's when a job starts climbing up the tree that it starts to lose priority.

    • A job has the top cluster priority in its exact cluster path.
    • It has 2nd cluster priority at all cluster nodes under its branch (towards the root).
    • So, a job submitted to /A/B/C/D/E has the highest priority in /A/B/C/D/E.
    • It has 2nd priority in /A/B/C/D, since a job specifically submitted to /A/B/C/D has top priority in there.

    When a job cannot find any hosts by descending down its tree branch towards the root and has to start "climbing" the tree to find hosts, it loses priority. The more level it has to climb up, the lower its priority. 

    • If jobA is submitted to /A, jobB to /B, then they both will have the same priority in /C.
    • If jobA is submitted to /, jobB to /B, then still both of them have the same priority in /C.
    • If jobA is submitted to /A, jobB to /B, jobA will have higher priority than jobB in /A/C because jobA only has to climb up 1 level, whereas jobB has to climb up 2 levels (over to / then up to /A/C).

    Another example that shows how jobs in different clusters are prioritized on the same Worker in /A/B/C:

    Job Cluster

    Calculated cluster priority order for a Worker in /A/B/C

    Cluster traversal count

    job cluster = /A/B/C

    1

    no traversal

     

     

     

    job cluster = /A/B/C/D

    2

    1 traversal down

    job cluster = /A/B/C/D/E

    2

    2 traversals down

     

     

     

    job cluster = /A/B

    3

    1 traversal up

    job cluster = /A/B/F

    3

    1 traversal down, 1 up

     

     

     

    job cluster = /A

    4

    2 traversals up

     

     

     

    job cluster = /

    5

    3 traversals up

    job cluster = /G

    5

    1 down to /, 3 up into /A/B/C


    All jobs are first sorted by the cluster priority order for the Worker, and then sorted by the job's numerical priority value within that priority order.

    • the lowest priority job with cluster priority order 1 will have a higher effective priority than the highest priority job with cluster priority order 2
    • the lowest priority job with cluster priority order 2 will have a higher effective priority than the highest priority job with cluster priority order 3
    • and so on…
    • No labels