[Ofmfwg] Updated Composability/Monitoring/FAM/Decision Layer slides

[Aguilar, Michael J.]

Everyone

Here is some updated slides that I intend to put up on our Github site.

I worked on the Janusgraph/Cassandra server set-up over the last week and got a successful design, last night.

My working design is as follows:


  *   Ubuntu 22.x
  *   2 cores of X86_64
  *   4 GB of memory

The choices were made as follows:


  *   Cassandra database back-end
     *   I came to the conclusion that we needed a NoSQL database, as opposed to a relational database when I was imagining out the database transactions.  I came to that conclusion due to the fact that the keys needed to point to a next jumping off point.  That just smells like a columnar format database instead of a primary key set-up with fixed data lengths.  I saw a need to address multiple paths and simultaneous connections, something that a relational database is going to struggle with.  I jumped on Cassandra instead of Mongo or HBase because of its column data formatting and peer-to-peer architecture, for scaling.
     *   Janusgraph can do BerkeleyDB, as well, but Cassandra is more ‘robust’ and is scalable.
     *   We can use Lucene for queries.
     *   For my testing, I chose 3.11.0 because that was the version that was verified with Janusgraph.
     *   I configured that back-end janusgraph configuration with cql and Lucene.
     *   For dependencies, I installed Python2 and OpenJDK-8.11-JRE.
  *   Janusgraph
     *   Configurations set to run the Lucene search libraries.
     *   We can create Vertices and Edges that mirror our Sunfish database.
     *   We can fill the Vertices and Edges to match the Redfish entries
     *   The edges can mirror our Redfish connections.
     *   We can leverage Lucene libraries to gather intelligence on the machines and components
     *   We can describe multiple Graphs, simultaneously, for ‘zoom in/zoom out’.
     *   We can write tools that create Redfish/Swordfish associations and simultaneously build and deconstruct database entries and associations.
     *   Do we want to do a start-up comparison of the persistent database entries and in-memory real-time entries?  Are we looking for a run-time comparison?
     *   We can run Gremlin code (Turing compatible code for graph database entries and searches) for database changes and Lucene searches
     *   Janusgraph and Gremlin work with Java and Python3 extensions.  So, we can code up Gremlin transactions from a consistent code base.
     *   We can leverage the Gremlin shell and remote connections to provide in-code development and testing.
  *
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openfabrics.org/pipermail/ofmfwg/attachments/20230714/5f4c6013/attachment-0001.htm>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: Reference_Framework_for_Composition.jpg
Type: image/jpeg
Size: 53105 bytes
Desc: Reference_Framework_for_Composition.jpg
URL: <http://lists.openfabrics.org/pipermail/ofmfwg/attachments/20230714/5f4c6013/attachment-0003.jpg>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: Reference_Monitoring_System_Framework.jpg
Type: image/jpeg
Size: 55874 bytes
Desc: Reference_Monitoring_System_Framework.jpg
URL: <http://lists.openfabrics.org/pipermail/ofmfwg/attachments/20230714/5f4c6013/attachment-0004.jpg>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: Reference_Composer_Framework_Components.jpg
Type: image/jpeg
Size: 57542 bytes
Desc: Reference_Composer_Framework_Components.jpg
URL: <http://lists.openfabrics.org/pipermail/ofmfwg/attachments/20230714/5f4c6013/attachment-0005.jpg>