Newbie question 🙂 : How does using Pinot compare with a situation say using a "snowflake warehouse with connectors" for real time analytics? One obvious difference is being open source. Are there any more advantages (like operational simplicity, cost, etc?)
Dhiwakar N A
02/21/2023, 3:33 PM
Another difference i can think of is being able to "install" the "datastore" in our own cloud
m
Mayank
02/21/2023, 3:52 PM
Yes, Pinot shines when you have to power user facing analytics (internal or external), including all the above you mentioned
d
Dhiwakar N A
02/21/2023, 4:37 PM
Is user facing analytics not possible by using a warehouse like snowflake with connectors? In their documentation they seem to claim to support “near real time analytics”. Making it user facing is just a step away from making the backend services fire sql queries to snowflake
m
Mayank
02/21/2023, 4:40 PM
Going by experience of folks who tried it, it does not scale for qps/latency and become very expensive.
d
Dhiwakar N A
02/21/2023, 4:59 PM
Got it, thanks. Probably because it's centrally managed.
m
Mayank
02/21/2023, 5:10 PM
Actually, it is more about Pinot’s architecture, indexing techniques among other things, that contribute towards that.
d
Dhiwakar N A
02/21/2023, 5:12 PM
Yes, I did read up about the star tree index, can be very powerful if used well