Often, or at least to what I've been exposed to, the Startups have a reduced budget for infrastructure and they are in constant delivery pressure.
This two constant implies that you have some issues in production that you might no realise easily that they are happening.
I would say, that when the Startup is an early stage, the website traffic is still low. So we don't need a huge infrastructure support a lot of logs per minute.
That's when Affordable Monitoring comes into play. I would define Affortable Monitoring to a small budget solution is at least able to:
- Centralized Logging with Search capabilities
- For Backend, Frontend and Mobile
- At least 7 days retention time
- Alerts when the service is down or in degraded service state
- Some metrics like:
- Request Per Second
- 2XX, 4XX, 5XX errors in a period of time
- Mean Response Time
We should also define what a small budget means. I would say 20-30EUR/month for all the services, not per host.
Here what we are using π β
We have been using CloudWatch because we are using AWS as a Cloud Solution, and most of the requirements are in the free tier so it's a cheap solution for us.
It's true that it doesn't support, as far as I know, frontend logs. That's why we were considering Rollbar for the React WebApp.
And you? What do you propose as a good stack with those constraints? Would you adopt Grafana, Prometheus that early? What about ELK Stack? Is it worth compared with CloudWatch? Self-managed or third party hosted? Which metrics do you monitor?
Top comments (2)
Hello Aleix,
If you need something for monitoring your website's and APi's downtime try using
~ Uptime Robot (free)
~ Logic Monitor
~ Cloudradar ( 1.50$ per month ) my choice.
Apart from all these, you wanted to go pure free, try using Zabbix and Nagios which has great options inbuilt. And as same using webhook divert alerts to SLACK.
Note : Even if you are okay and in need of getting alerts to your personal mobile, try MSG91 tool ( startups affordable ).
Hi Allerick, thanks for sharing! We definitely could use some of those solutions you share :)