ClickHouse, a column-oriented DBMS to generate analytical reports in real time

Clickhouse has some weird quirks when you think of it as a SQL Database, but its astounding to use it. It&#x27;s faster than one would think, it can do some really cool data modeling, and provides a wealth of features for the average user out of the box.<p>The most important thing, and the thing that makes it attractive to me is that it is almost stupidly simple to setup and get running. It&#x27;s quite simple (when you wrap your head around it) to do sharding or replication and scale up. The zookeeper stuff takes a bit more effort, but most of that is due to zookeeper and not ClickHouse.

I remember playing with it a while ago and here are my 2c:<p>- It&#x27;s ridiculously fast, like, I didn&#x27;t know where the performance was coming from.<p>- Getting it up and running was a bit clunky (Docker saved me), but I hear it&#x27;s better now.<p>- It has non-standard (I mean, everyone is non-standard, but this is way off ANSI) and case-sensitive (???) SQL syntax. This annoyed me again and again.<p>- It seemed (and still seems) like a project that lives and dies with one developer - and no matter how brilliant he may be, I&#x27;m not willing to invest in a technology that has this risk and it&#x27;s so hard to migrate off of (because of the non-standard SQL).<p>I&#x27;m sad about the last two points, because the database is rather brilliant otherwise.

Discussion from a couple years ago: <a href="https:&#x2F;&#x2F;news.ycombinator.com&#x2F;item?id=11908254" rel="nofollow">https:&#x2F;&#x2F;news.ycombinator.com&#x2F;item?id=11908254</a>

The basic techniques for implementing a fast column-store data warehouse have been well-known for 10 years. There are several excellent commercial and open-source implementations of these techniques:<p><pre><code> - BigQuery - Snowflake - Redshift - Presto </code></pre> ClickHouse is not one of them. It doesn&#x27;t have:<p><pre><code> - Transactions - Distributed joins - Separate compute from storage - UPDATE - User management </code></pre> I don&#x27;t mean to be a jerk, I&#x27;m just trying to save people some time. Columnar DBs is well-trod territory and ClickHouse is way behind.

ClickHouse Indeed does not do &quot;Separate Compute from Storage&quot; yet it is architectural decision not a feature gap. Running ClickHouse with directly attached storage and built in replication can be super fast and cost efficient. It works best for stable workloads

CloudFlare is using ClickHouse. That does say something<p><a href="https:&#x2F;&#x2F;blog.cloudflare.com&#x2F;http-analytics-for-6m-requests-per-second-using-clickhouse&#x2F;" rel="nofollow">https:&#x2F;&#x2F;blog.cloudflare.com&#x2F;http-analytics-for-6m-requests-p...</a>

Can it run on a cluster? Or single server only?<p>Can you update specific rows? How fast are updates?<p>How does it compare to monetDB.