2. Similarly, monitoring is a one time task too. And if this isnt yet a project making $xx,000/month, you probably dont need monitoring yet.
3. Similarly, if your project isnt making a ton of $$$, you can forgo OS updates and only take care of critical secure updates.
In summary, the things you think you need to do, you can delay until youre making non trivial $$$ from it.
Many managed databases are a commodity, there is a ton of competition trying to sell managed postgres cheaper than the last company. More 'exotic' databases have less competition and they can make a higher margin.
If you are trying to min max your cloud cost you need to focus on cloud products that are heavily commoditized.
Disclaimer: I work for Altinity.
• Idling - your cluster will scale to zero during periods of inactivity. If this is a relatively ad-hoc data analysis project, you'll benefit from the usage-based model (ie - if you're only actively running queries 10% of the time, you won't pay for the other 90%)
• Autoscaling — similar to above, if your workload is fairly bursty, your cluster will scale up and down based on resource demand
• 'Development' tier clusters — given the OVH specs listed, likely this isn't a good fit, but depending on your workload, the 2x8GiB setup can be a very cost-effective option
Happy to chat further about your workload and potential cost optimizations that can be made in ClickHouse Cloud if you're interested — email in my profile.
Disclaimer: I work for ClickHouse