Performance testing and credentials from FS
Has anyone done performance testing of Convex? What is the expected system response for medium complexity of mutation? What is maximum TPS?
I could not find any information on that topic anywhere.
Do you know any significant financial services institution that is using Convex on production? Specifically I am interested if the system ever went through detailed security assessment.
Thank you!
4 Replies
Thanks for posting in <#1088161997662724167>.
Reminder: If you have a Convex Pro account, use the Convex Dashboard to file support tickets.
- Provide context: What are you trying to achieve, what is the end-user interaction, what are you seeing? (full error message, command output, etc.)
- Use search.convex.dev to search Docs, Stack, and Discord all at once.
- Additionally, you can post your questions in the Convex Community's <#1228095053885476985> channel to receive a response from AI.
- Avoid tagging staff unless specifically instructed.
Thank you!
The end user will call LLMs API deployed on external server, that will be fed financial information stored in convex database. The financial data is hundreads of parameters per one asset, that will be provided from DHW and trading systems.
What we want to measure now is what volumes convex can handle, first for POC second for expected production volumes.
Do you have any credentials from major financial institutions?
There was a thread on here a while back, from what i recall it was suggested that no customer has had a need for more than 10k functions a second but there are some in the thousands. As with anything its highly dependent on design.
soc https://discord.com/channels/1019350475847499849/1019350478817079338/1286445732269789187
@Marcin hi!
so to expand on this a bit
Convex is in general as fast as it needs to be, where we're constantly scaling it to keep up with our busiest customers. right now, if someone has ~1000 mutations a second and < 10k queries a second, we can work with you and support that load with good performance. we'd just have to make sure your project was fairly optimized. but if someone showed up right now and said "I need to do 10k mutations a second with good latency", we're probably 9-18 months away from an architecture that can support that. so the answer on scale is "it depends", and scaling is sort of a process not really a state. lmk if that makes sense
on the security front, we're HIPAA, SOC 2 all that good stuff. we have good security policies and automation, we do periodic pen testing, use 3rd parties to audit designs of critical systems (like convex auth), etc etc. Convex is a very security-minded platform