[derN3rd]

My company recently started to invest more into analytics and we had to find a good solution on which backend/database we want to decide. After some research we settled on ClickHouse and we couldn't be happier.

- Super easy to setup - Easy to backup - Fast configuration (documentation could be better at some parts) - Similar SQL dialect as our devs use in MySQL

Only negative points I could find so far: - No 'good' management GUI as e.g. phpMyAdmin, pgAdmin, Mysql Workbench - Caching/Batching layers not directly implemented, but through external software

As we are a fairly small company all other analytical databases would have cost us a large amount of money/time more. Friends of us recently hired a group of data engineers/analytics who also brought all their AWS knowledge and toolings with them, which basically brought them to the same outcome as us, while we only have 5% of their costs and all our devs are able to either ingest or query some data

EDIT: Does anyone have some recommendations on what GUIs I could give our PMs to work with ClickHouse instead of writing queries? All SaaS I found didn't support ClickHouse yet or would cost us a newborn. Also what tools do your devs use when they work with ClickHouse data?

[glogla]

I like a lot of things about Clickhouse but one thing I'm afraid of is what happens when your data won't fit on a single machine. The replication and sharding seems pretty difficult and from reading the documentation feels like it might be pretty fragile.

I think once you reach that scale, systems that completely separate data and compute (like snowflake or trino+s3) are much less of a pain to run since even if you completely blow up your compute the data stays.

> Does anyone have some recommendations on what GUIs I could give our PMs to work with ClickHouse instead of writing queries? All SaaS I found didn't support ClickHouse yet or would cost us a newborn. Also what tools do your devs use when they work with ClickHouse data?

I think both Superset and Metabase were interesting choices - if you want to save money (at the expense of engineering time) you can self-host them.

[hodgesrm]

> I like a lot of things about Clickhouse but one thing I'm afraid of is what happens when your data won't fit on a single machine.

ClickHouse sharding and replication is not that hard to master--it's simple and the parts are visible. If you don't want the headache of distributed system management, run it in Kubernetes or a managed service. ClickHouse-as-a-service is widely available from multiple vendors.

ClickHouse is also fast and cost-efficient at scale. It's a very good fit for multi-tenant SaaS analytics where you need fixed latency on responses to users.

Disclaimer: I work for Altinity, who run a cloud platform for ClickHouse.

[nojito]

Ebay wrote a great blog post about clustering clickhouse

https://tech.ebayinc.com/engineering/ou-online-analytical-pr...

[lk888]

--- The replication and sharding seems pretty difficult and from reading the documentation feels like it might be pretty fragile.

That's why trip.com replaced Clickhouse with StarRocks, a free (open source) OLAP database that handles sharding beautifully to give you linear scalability, and better query performance for single table or joined queries.

[gaploid]

We built a Managed ClickHouse service to help exactly these difficulties with that technology. We are handling sharding, clustering, zookeeper, patching, updates without downtime, and Hybrid storage based on S3. https://double.cloud

[AdamProut]

How many Clickhouse as a service offerings exist now? I stopped counting at 7 a few months ago (double.cloud was not on my list).

[gaploid]

Could you share a list of them?

[AdamProut]

  - Firebolt (Hard fork of clickhouse)
  - Altinity
  - Gigapipe
  - Hydrolix
  - Bytehouse.cloud
  - https://clickhouse.com/ ("coming soon")
  - TiDB (Their columnstore is a fork of clickhouse)

I stopped tracking after this. I saw a few press releases go by announcing a few others as well which I lost now.

The official Clickhouse Inc. is surely going to be under pressure to pull features out of their open source offering over time to differentiate themselves.

[zX41ZdbW]

ClickHouse can run in a classic shared-nothing setup and in "cloud-native" setup with shared storage. Setting up a distributed system with hundreds of machines can be difficult... but it's actually not more difficult than for any other distributed system at this scale.

[blinkov]

Just in case, here's built-in batching https://clickhouse.com/docs/en/operations/settings/settings/... and a list of mature UIs https://clickhouse.com/docs/en/connect-a-ui (that all have SaaS offerings I believe) and all third-party UIs https://clickhouse.com/docs/en/interfaces/third-party/gui

[derN3rd]

Thanks!

I remember trying the built-in batching but we had some trouble with it, so we just switched to <https://github.com/nikepan/clickhouse-bulk> which works without any issue since then.

Will have a look at the UIs listed there

[ceeplusplus]

Tableau is second to none for dataviz imo. Much more versatile than the competition. You can do interactive visualizations, set thresholds, color code/label data, etc. Of course it costs a lot, but if dataviz is something that is delivering real value the cost is more than worth it.

[gbrits]

Clickhouse support for Tableplus landed 2 months ago. https://github.com/TablePlus/TablePlus/issues/670