Thoughts on the new pricing
I just saw the new pricing announcement (BTW you should start a thread with each announcement so we can comment on it directly) and had some thoughts.
I always wondered about the "execution cost" of my operations and how that would affect not just my cost but also you guys. Before you didnt tie how I write my code to the cost of running it. So effectively the incentive for writing high-performance code (using indexes and whatnot) wasnt there, now it is because it contributues to my quota.
So in general I think this is a good thing as it incentivizes you to code in a more efficient manner.
Now what I think is missing from Convex to complete the circle is better tooling to estimate and analyze the performance of your queries, mutations and actions.
The graphs that you give on the dashboard are nice but I think there's a lot more that can be done to help here.
For example; a big issue I have with Postgres and other databases is that my queries work great locally when I only have a few tens of records in my database but as soon as it hits production then that JOIN or AGGREGATE that you did suddenly grinds things to a halt. Or worse it starts slowing down over time and unless you have some APM hooked up like Elastic then you wont know that.
Setting up and understanding Elastic is non-trivial so I think there is definately a lot of room here for Convex to do what you have done for application development but for application monitoring. I want to know if my queries or mutations slow down over time and why they might be slow (ANALYZE EXPLAIN). I want to be emailed / notified when response times are above 100ms for example.
Overall I think the new pricing is fair and good but IMO it would be good to provide further tooling in the future to allow us to know WHY some operations are slow and expensive and if they change over time.
Cheers, keep up all the hard work!
I always wondered about the "execution cost" of my operations and how that would affect not just my cost but also you guys. Before you didnt tie how I write my code to the cost of running it. So effectively the incentive for writing high-performance code (using indexes and whatnot) wasnt there, now it is because it contributues to my quota.
So in general I think this is a good thing as it incentivizes you to code in a more efficient manner.
Now what I think is missing from Convex to complete the circle is better tooling to estimate and analyze the performance of your queries, mutations and actions.
The graphs that you give on the dashboard are nice but I think there's a lot more that can be done to help here.
For example; a big issue I have with Postgres and other databases is that my queries work great locally when I only have a few tens of records in my database but as soon as it hits production then that JOIN or AGGREGATE that you did suddenly grinds things to a halt. Or worse it starts slowing down over time and unless you have some APM hooked up like Elastic then you wont know that.
Setting up and understanding Elastic is non-trivial so I think there is definately a lot of room here for Convex to do what you have done for application development but for application monitoring. I want to know if my queries or mutations slow down over time and why they might be slow (ANALYZE EXPLAIN). I want to be emailed / notified when response times are above 100ms for example.
Overall I think the new pricing is fair and good but IMO it would be good to provide further tooling in the future to allow us to know WHY some operations are slow and expensive and if they change over time.
Cheers, keep up all the hard work!
