# Client interface to use for faster data ingestion **URL:** \ **Category:** CrateDB\ **Created:** [April 29, 2020, 9:05am UTC](https://community.cratedb.com/t/client-interface-to-use-for-faster-data-ingestion/384 "2020-04-29T09:05:49Z")\ **Posts on this page:** 2\ **Page:** 1
**Author:** ![Vivek](https://avatars.discourse-cdn.com/v4/letter/v/77aa72/32.png) [@Vivek](https://community.cratedb.com/u/Vivek)\ **Post date:** [April 29, 2020, 9:05am UTC](https://community.cratedb.com/t/client-interface-to-use-for-faster-data-ingestion/384/1 "2020-04-29T09:05:49Z")
Is there a difference in ingestion speed depending on the end point (Http or Postgres) we use, Is there a benefit in terms of data ingestion over the selection of the endpoint, i.e. to use Http or Postgres end point to communicate with CrateDB. I have gone through Insert method documentation, my question is more on the choice of the endpoint. Any suggestions? Thanks ---
**Author:** ![Walter\_Behmann](https://sea2.discourse-cdn.com/flex020/user_avatar/community.cratedb.com/walter_behmann/32/117_2.png) [@Walter\_Behmann](https://community.cratedb.com/u/Walter_Behmann)\ **Post date:** [May 13, 2020, 7:17am UTC](https://community.cratedb.com/t/client-interface-to-use-for-faster-data-ingestion/384/2 "2020-05-13T07:17:09Z")
hi Vivek, from the various tests I have run in the past and I explicitly had a look at the endpoint, it is a very small difference, **if** at all. Assuming you already have batch inserts on the picture, you are good to go. The implementation on the client side (json en-/decoding, libraries used, etc. ) can have hugh impact. We also published some blueprints recently about ingesting via an Azure Function. Which reaches easily about 100.000 events/sec. on a 5-7 cluster. Side node: insert hits the CPUs, make sure to have a few “arround” 😉 … but not too fancy neither. To reach 100k insert/s we typically use 8-12 cpus / node. It makes sense to use what is the easiest to implement in your case.