Hey,
Am 04.04.2012 um 23:30 schrieb Kostya Lutovich:
My name is Konstantin Lutovich.
I'm a student from ukrainian national university of Kyiv. Got bs degree last year and
now studying to get the master one.
I liked the idea of implementing RHQ storage backend using Infinispan and want to
contribute it as a GSoC project.
Now I would like to get familiar with RHQ and Infinispan because I have not used them
before.
For now I've build RHQ, looked a bit through the confluence and source and started
with some quickstarts on Infinispan.
Now I am going to continue working with documentation. Would like to ask what modules of
RHQ should I look at for now?
Any suggestions are highly appreciated :)
The relevant module is the server jar and here especially the MeasurementDataManagerBean
with
org.rhq.enterprise.server.measurement.MeasurementDataManagerBean#addNumericData
Then there are numerical Baselines and a few hourly jobs that also iterate on that data.
One of the key factors here is that a metric is always identified by a scheduleId - that
is it's primary key
(technically the PK is <scheduleId, timestamp>, but that does not matter for this
discussion). So in
order to be able to scale the backend Infinispan infrastructure, by adding more nodes, it
is important
to co-locate data with the same scheduleId together. When e.g. retrieving metrics for a
graph, the data
for one schedule and a time range is obtained. When doing the hourly compression job, data
for a schedule
is aggregated and so on.
Aside from simple storage and retrieval of data, I also expect the backend Infinispan
instance to do the
hourly jobs (aggregateion, baseline compression, deletion of old data).
Also if for a graph display the data needs to be aggregated, I expect that to be done on
the Infinispan instance.
Heiko
--
Reg. Adresse: Red Hat GmbH, Technopark II, Haus C,
Werner-von-Siemens-Ring 14, D-85630 Grasbrunn
Handelsregister: Amtsgericht München HRB 153243
Geschaeftsführer: Mark Hegarty, Charlie Peters, Michael Cunningham, Charles Cachera