Our product comprises a collection of microservices catering to a substantial customer base. Currently, we are in the process of assessing Grafana Loki as a logging solution. Within our application logs, the tenantId is a crucial component that I am extracting and establishing as the tenant identifier. It has come to my attention that Loki inherently supports multi-tenancy. However, I am uncertain about its feasibility in our scenario, considering the potential translation of this feature into managing thousands of tenants. I seek guidance on the most advisable approach for configuration:
- Employ Loki’s multi-tenancy functionality, segregating each customer’s logs into distinct tenants. Is this setup scalable given the volume of tenants?
- Opt for a single tenant setup, utilizing the pack/unpack/structured metadata feature to affix the tenant context alongside logs and subsequently filtering based on this context during queries. Would this method be more efficient?
- Are there alternative approaches recommended for our use case that might offer better scalability and efficiency?
I appreciate any insights or recommendations you can provide regarding the optimal setup for accommodating our extensive customer base within the framework of Loki’s capabilities.