Using SingleStoreDB as a storage engine of ClickHouse

Help
1

Hi,

To use aggregate functions provided only by ClickHouse, I want to connect SingleStore using ClickHouse’s storage engine feature. The table storage engine feature seems to work well in a simple test, but I would like to hear if there are any problems or limitations that occur while using it for work.

Thanks,
Jaenyun

1 Like
2

I don’t have any information on that other than to say that we’re MySQL-compatible so any MySQL interface they have may work. I’ll ask somebody who knows this area better to take a look. If you learn anything else on your own, please do post. Thanks!

3

out of curiosity, which aggregate functions are you interested in?

-Adam

4

Thanks to Hanson for the reply.

Hello Adam,
Clickhouse is used in several services, and one of them is considering applying S2DB. And it was discovered that the retention and windowFunnel functions do not exist in S2DB.

I know it can be implemented with UDAF or Wasm. I already implemented retention with UDAF, but it showed faster processing when using S2DB as the storage engine in ClickHouse.

Therefore, I am reviewing whether it is possible to let S2DB handle most of data ingestion, ad-hoc and join queries, and to handle only some aggregations that S2DB does not provide for ClickHouse. I’m looking for examples of this structure.

Thanks,
Jaenyun

5

Jaenyun,

I’ve asked around and did not find anybody with experience using SingleStoreDB as a storage engine for Clickhouse. Please let us know if you learn anything about this.

I opened an internal feature request for “retention” and “windowFunnel” functions (no commitment at this time, but we will track this request). You can implement those as UDAFs in SingleStore – it sounds like you have done at least one of them. In case you have not seen it, see our enhancements in 8.1 related to "handle state" Wasm UDAFs which are faster for aggregates with large state.