Describe the current state/issue
Part of #666. Matomo's read-write operations for logging are connected to the primary database. This setup is nonviable relative to the increased DB operations from EZID assuming N2T's resolution functionality, post cut-over, and had to be fully disabled with a hotfix
Describe the desired state/solution
Should we continue to use Matomo, we would need to separate its read-write operations for logging into a different system, specifically using either RabbitMQ or Redis to:
Offload the write-intensive logging operations from the primary database.
Use a message queue (RabbitMQ) or in-memory data structure store (Redis) to handle the high volume of logging data.
Use an asynchronous process to periodically transfer the logged data from RabbitMQ/Redis to a persistent storage solution for long-term analytics.
Per review with Jing, we would pursue this relative to using Matomo exclusively for UI (vs. API) monitoring and also investigate whether this could be logged to OpenSearch instead.
Describe the current state/issue Part of #666. Matomo's read-write operations for logging are connected to the primary database. This setup is nonviable relative to the increased DB operations from EZID assuming N2T's resolution functionality, post cut-over, and had to be fully disabled with a hotfix
Describe the desired state/solution Should we continue to use Matomo, we would need to separate its read-write operations for logging into a different system, specifically using either RabbitMQ or Redis to: