🔭 First Impression: Exploring Grafana Mimir
... After Years with Thanos!
Not an observability( OE ) purist, but I do appreciate system(s) that tells me how my pets and cattle are doing - So , let's just say I do and have to keep my eyes on the metrics.
After relying on 'Thanos' for scalable Prometheus metrics for quite some time, I recently built a prototype using Grafana Mimir—mostly to satisfy my curiosity.
What I expected to be a familiar experience turned out to be a surprisingly strong contender for how we think about metrics at scale.
We historically leaned on Thanos for its modularity and open standards, but once your platform grows beyond a certain point, things like compactor bottlenecks, sidecar sprawl, and query performance start to hurt and you notice where the seams show.
So I took Mimir for a spin. Here’s what stood out—first-hand:
1. Query Performance - impressive
Mimir’s sharded querier and query scheduler did a great job slicing large queries, preventing slowdowns. In Thanos, we’ve had to be more tactical with query shaping—Mimir handles that more natively.
2. Cleaner Architecture, Fewer Moving Parts
The decoupling in Thanos brings flexibility—but also complexity. Mimir felt more “batteries-included,” with centralized compaction, consistent ingestion behavior, and better internal coordination.
3. Multi-Tenancy Built-In
We serve multiple product lines and environments. Mimir’s first-class tenant isolation and fine-grained limits gave us confidence we could scale without teams stepping on each other’s toes.
4. Operational Efficiency
Rate-limiting, resource usage policies, and native object storage integration? Much easier to configure and reason about compared to stringing together CLI flags in Thanos components.
❓The BIG Question :
Is Mimir better than Thanos?
In many ways— YES , if your scale and complexity demand it Or If you're hitting the "metrics ceiling," Mimir is worth your time. But I wouldn’t call it a replacement for every scenario. Thanos remains easier to plug into an existing Prometheus setup and is still a great option for teams just getting started or managing smaller environments.
PS: If you're evaluating observability platforms at scale, it’s worth giving Mimir a serious look. Happy to chat if you’re navigating a similar shift.
#GrafanaMimir #Thanos #Prometheus #Observability #SRE #DevOps #CloudNative #MetricsAtScale #EngineeringLeadership