Closed bmtcril closed 1 year ago
An update on where I'm at with this. All tests have been on my M1 Mac under Docker Desktop assigned 4 CPUs, 10GB memory, 1GB swap, with no other containers running.
ENGINE MergeTree ORDER BY (course_run_id, verb, emission_time)
PRIMARY KEY (course_run_id, verb)
100000000 rows in set. Elapsed: 19288.384 sec. Processed 100.00 million rows, 145.28 GB (5.18 thousand rows/s., 7.53 MB/s.)
Next steps:
I was able to complete a billion row test with remarkable success, and am writing up findings here: https://openedx.atlassian.net/wiki/spaces/COMM/pages/3593109505/Clickhouse+as+an+LRS+Data+Store
I'll close this ticket out when the details are all completed there.
This has been done for a while, ClickHouse is selected, Ralph integration is done. 👍
We're interested in running Ralph as an LRS, but are hoping to use a highly scalable analytic database as the backend. This spike is to get hands on with Clickhouse and see how well it suits our needs. This isn't intended to be a load test or scalability test, but rather to answer these questions:
AC:
Do the following and document the necessary steps, SQL, etc. as well as the outcomes...