question Re: Kudu increment update support in Support Questions

Kudu increment update support

wumin — Fri, 16 Sep 2022 11:31:06 GMT

hi,

Dose Kudu support increment update?

For example, we have a table with two field, browser and visit count :

borwser visitCount

firfox 10

And then, we got a new visit from firfox, the table will be update to:

borwser visitCount

firfox 11

we want to do this update by increment operation, ranther then read the visitCount, add one and write it back.

Thanks in advance.

Re: Kudu increment update support

J-D — Thu, 27 Apr 2017 14:50:27 GMT

Hi,

Kudu currently doesn't support atomic increments or any kind of operation that reads a value and makes a decision based on that value with the guarantee that it won't change.

Thx,

J-D

Re: Kudu increment update support

linhao — Mon, 19 Jun 2017 08:19:54 GMT

Hi, We also found Increment as a very useful operation in production. It can reduce row number. So why increment is not supported, is it on the plan?

Re: Kudu increment update support

Tomas79 — Mon, 19 Jun 2017 10:21:33 GMT

I think it is just because how Kudu is implemented. The visit count is not the primary key of the table. Nor the browser column. For atomic incrementing you have to have the data in one location on one server.

Re: Kudu increment update support

hindog — Fri, 25 Aug 2017 21:20:31 GMT

"Nor the browser column" -- What if the "browser" column WAS the primary key? Would increment be possible in similar fashion to upsert, the difference being that internally Kudu would read the existing value and add it to the increment value before saving?

Re: Kudu increment update support

J-D — Mon, 28 Aug 2017 16:03:11 GMT

@hindog wrote:
"Nor the browser column" -- What if the "browser" column WAS the primary key? Would increment be possible in similar fashion to upsert, the difference being that internally Kudu would read the existing value and add it to the increment value before saving?

The other Kudu limitation that this runs into is that row key columns cannot be modified. It would be somewhat easier to implement atomic increments on non-key columns, but then that means doing a lot of updates and it's not something Kudu excels at e.g. it supports modifying data but the higher the rate the slower the reads are.