Parallel Query from Cacti Directly

[TheWitness]

So, MySQL and MariaDB both have taken Parallel Query to the max, but there is one class of Map Reduce queries that neither vendor has decided to tackle, and that's with respect to.

So, here is the scenario. If you are a syslog or a flowview user, you know full well, the tables get real big and there are lot's of them. In today's MySQL or MariaDB, there is no concept of breaking the UNION queries required to gather data into smaller pieces and running each piece in parallel and then combining the results (you know map reduce). Both products go through the unions in serial, which leads to real long query times. It's awful.

So, I was thinking, can't we just do that in Cacti? Tell me what you think.

[TheWitness]

Well, I wrote it for FlowView, and it likely, with enough nodes, rivals Splunk performance.