Plumage - NoSQL Operational Data Store for Condor
.................................................

Schema for Data Samples
=======================

Plumage ODS captures sampled ClassAd data at regular intervals. ClassAd 
attributes typically have lengthy names. Plumage captures and preserves 
the raw ClassAd attributes in their entirety. However, there is a lightweight
schema used to record the time-sampled copies of these values from the raw
ClassAds. For the purposes of efficient indexing and storage, these key fields 
are abbreviated to a few lower case letters. The following is a mapping of 
these terse key names to their more familiar Condor ClassAd counterpart. 

In all cases, the type of the sampled attribute can be assumed to be the same 
as its ClassAd counterpart.

NOTE: 'ts' is used everywhere in the samples as timestamp (non-ClassAd 
attribute).

ACCOUNTANT
----------
'n':    Name
'ag':   AccountingGroup
'prio': Priority
'fac':  PriorityFactor
'ru'':  ResourcesUsed
'wru':  WeightedResourcesUsed
'bu':   BeginUsage
'lu:    LastUsage
'au':   WeightedAccumulatedUsage
'cq':   ConfigQuota
'eq':   EffectiveQuota
'sq':   SubtreeQuota
'sp':   SurplusPolicy

MACHINE
-------
'm':    Machine
'n':    Name
'ar':   Arch
'os':   OpSys
'req':  Requirements
'k':    KeyboardIdle
'la':   LoadAvg
'st':   State
'cpu':  Cpus
'mem'   Memory

SCHEDULER
---------
'n':    Name
'mjr':  MaxJobsRunning
'nu':   NumUsers
'tja':  TotalJobAds
'trun': TotalRunningJobs
'thj'   TotalHeldJobs
'tij':  TotalIdleJobs
'trem': TotalRemovedJobs
'tsr':  TotalSchedulerRunningJobs
'tsi'   TotalSchedulerIdleJobs
'tlr'   TotalLocalRunningJobs
'tli'   TotalLocalIdleJobs
'tfj'   TotalFlockedJobs

SUBMITTER
---------
'sn':   Name
'ma':   Machine
'jr':   RunningJobs
'jh':   HeldJobs
'ji':   IdleJobs


NOTE: The raw ClassAd data could just as easily be captured by a remote client 
and stored to an RDBMS such as PostgresSQL using a different custom schema. 
These samples are collected for the Plumage sample tools packaged for 
illustrative purposes.
