Ask HN: What is the best solution to monitor your infrastructure in 2022?

14 points by misiti3780 ↗ HN
I have a large AWS deployment, Django based using celery, scrapy, DRF, redis, ECS, postgres. I was wondering what software solutions other engineers are using to monitor this type of thing?

I was looking to ask questions like:

- Number of requests per minute/hour/day - Number of celery workers working - Size of celery queue - Keys in redis - postgres stats - etc.

Basically what a macro-view of what is going on. I dont want to build my solution from scratch because I do not have the time.

11 comments

[ 2.8 ms ] story [ 41.4 ms ] thread
What's my budget and staffing? My solution would entirely depend on that.
Sounds like a reasonable question.

> Basically what a macro-view of what is going on. I dont want to build my solution from scratch because I do not have the time.

Seems like this person is in AWS already; is there not an AWS centric solution to this that can be spun up pretty quickly?

staffing is backend engineers (myself one of them) already overworked, budget is 1K/month right now
Prometheus and Grafana. Datadog. Managed services.
I tend to divide "monitoring" into two pieces: alerting ("tell me when XYZ is down or failed") and metrics ("tell me what's going on right now so I can look for patterns and bottlenecks"). Those are closely tied together, but they're not quite the same thing: there are plenty of products that are really good at one but not good at all at the other.

Looks like you're looking mostly for the latter, in which case Prometheus + Grafana would be the free-to-play option I'd reach for first. The new Grafana Alerts looks interesting for covering the "alerting" piece. Cloudwatch is the logical choice for AWS stuff and it does both pretty well (although I don't know whether you need a separate service like PagerDuty to really handle alerts), but it's not free (especially as you do more with it), and some of the other options like a managed provider (Datadog, e.g.) become more attractive as you put more and more money into it.

For database I love DPA. For New Relic is great for pricing for the rest. I prefer it to Datadog as NRQL is awesome. Datadog has an events API which is cool also.
Depends on your budget. If you have a shoestring budget, then I recommend Honeycomb or Sentry if you don’t want build your own solution
Atatus provides dashboards and alerting for monitoring your servers, software and more in order to understand how healthy your systems are.