HN Explorer

Ask HN: What tools do you use for a cloud service?

2 points by esseti ↗ HN
Hi People,

I'm checking out what tools people use to maintain up and running cloud based services. Those are some that I know, it would be great to know what you use.

Cloud: AWS/GCP/Azure/Other Here I need something for Docker, probably GCP with kubernets suits better. Price wise GCP is better. AWS has more tools. Still, I can't figure out what I really need. Any suggestion for a system that helps in filling in the complex AWS pricing system or that has a comparison among the services?

Log: Graylog/ELK/ I want to collect log for inspections, plus having the possibility to search for users and IP. Both can be deployed on our servers.

Error: Sentry/OpBeat I need a system that reports expecptions and error when they happens, if they have insights on the performance is a big plus. Not sure if sentry or opbeat can work as logger aggregator. Sentry seems to allow self deployment, which is good.

Monitoring: DataDog/statsd+graphite+grafana Need to check if machines are up or down, load etc. Datadog seems to be very good altought it cannot be selfhosted.

Stats: Prometheus/Graphite Right now I've statsd sending to graphite etc a pain. I need to monitor VMs and containers. Plus exposing my own stats such as how many calls a user do.

1 comment

[ 1.3 ms ] story [ 9.4 ms ] thread
tracker1
You could also use Digital Ocean or Linode with or without Docker Cloud tools. Of course, then you don't get the option for some of the hosted DB and other resources. Depends on your needs, expertise and time to setup and configure really.

I prefer GCP and Azure's tooling to AWS (which just feels like there are too many options). Azure in particular is really easy to get started with some of their hosted services compared to the others, but compute nodes seem a bit overpriced and slower than the others. GCP seems to be best in terms of raw compute/memory and reduced latency (within the dc). Amazon is somewhere in the middle, with the biggest list of service options (should you want them).

Unfortunately there's no clear winner, and likely never will be. What tooling are you wanting to use, that may guide your answer a bit.

At work now, we're using VisualStudio.com for source hosting as well as some CI pipelines in there, combined with GoCD for production deployments. Our dev services are in Azure, we have a combination of in-house (3 phx/nyc/lon colocations) and Amazon. While I prefer Azure and GCE to AWS, it looks like AWS has entropy, so heading in that general direction moving forward.

Splunk for general logging (daily log shipping from various servers), and a combination of MS-SQL, Cassandra and Dynamo for different work loads and DCs.

We are heading towards containers for newer projects, which will at least make things more portable moving forward.