[wilbeibi]

The thing I really don't get is why CockroachDB is avoid benchmarking with it's rival tidb (https://github.com/cockroachdb/docs/issues/1412). tidb already pretty mature, used in many big companies (Let's say, Didi, which on the similar scale data with Uber, and banks).

Even if I like CockroachDB's pg sql more, it would be helpful to have the comparison/benchmark to show something more.

[atombender]

TiDB looks promising, but it doesn't have serializable transactions at all, which makes it something of an apples-to-oranges comparison at the moment when it comes to OLTP.

TiDB has a weird kind of variation on "read committed" where you get phantom reads (though they're not called that in the documentation, which is actually ambiguous on this point). This is a problem for apps that expect consistency.

[siddontang]

Siddon Tang from TiDB here. Thanks for your interest in TiDB! TiDB uses Snapshot Isolation (https://en.wikipedia.org/wiki/Snapshot_isolation), which is similar to Repeated Read in MySQL. It doesn’t allow Read Phantom but can’t avoid Write Skew. IMO, this can work well for OLTP in most of the cases. If you really care about serializability, you can use `select for update` to promote the Read as Write explicitly like other databases do.

TiDB supports READ COMMITTED isolation which is not the same as MySQL, but it is just designed for some special cases for TiDB itself and it is not recommended for external users.

[atombender]

Thanks, that's helpful!

[gregwebs]

The documentation states that phantom reads are not possible (under the default isolation level or repeatable read): https://github.com/pingcap/docs/blob/master/sql/transaction-...

[manigandham]

TiDB is much more complicated to run with several moving pieces. Also missing a lot of standard relational features as you can see from the roadmap: https://github.com/pingcap/docs/blob/master/ROADMAP.md

[jinqueeny]

As a distributed database system, the highly-layered architecture of TiDB is not a disadvantage because it makes TiDB easier to debug and diagnose when issues occur. The complicity resulted from the separated modules can be easily tackled by automatic deployment tools and microservice framework. For example, TiDB can be easily deployed either using Ansible: https://pingcap.com/docs/op-guide/ansible-deployment/ or Docker Compose: https://pingcap.com/docs/op-guide/docker-compose/

TiDB has been widely adopted by many users (https://github.com/pingcap/docs/blob/master/adopters.md) in production because it support the best features of both RDBMS and NoSQL. It is quickly evolving and iterating based on users’ requirements which are prioritized and listed on the Roadmap.

[manigandham]

Sure, I get the advantages of layers in architecture, especially the growing trend of separating compute and storage.

However there's a difference between architecture and deployment, and having everything in a single package makes operations much easier. CockroachDB also uses a KV storage layer (using RocksDB) with SQL on top.

[gregwebs]

TiKV uses Rocks also which is not distributed. TiKV is distributed and used by TiDB. TiSpark also uses TiKV directly which starts to demonstrate some of the advantages of the layered architecture.

[manigandham]

I get it. Both databases are layered, but I'm saying deployment is still easier with a single package/binary instead of having TiDB, TiKV and Placement Driver instances.

[siddontang]

TiDB provides many some ways to let the deployment easier. If you just want to have a try on the single machine, we provide docker-compose (https://github.com/pingcap/tidb-docker-compose). But if you don't like docker, the single binary may be easier.

For the business users, we support ansible and K8s deployment, both can help them run the TiDB easily and quickly in production. The deployment is not a problem so far.

IMO, for the distributed system, if you need to operate many instances (10, 100 or even more), what you pay attention to most is not whether the instance is one single binary or not, but is how to operate all these easily. At that time, there is no much difference.

[minxomat]

CockroachDB's SQL implementation is still heavily unoptimized for actual business queries (subqueries, unnest, (recursive) CTE, schema changes). Features like this either take way too long or are not even implemented yet. Some of them are in the 2.1 milestone though. Let's hope.

I'd really love some kind of distributed-for-performance database using the exact optimizer and query planner of SQLite plus std plugins (FTS5, JSON, transitive_closure, spatial). Something like a mix between Bloomberg's comdb2 (which uses a modified SQLite frontend) and rqlite (distributed-for-safety).

Note: You can save most of the shortcomings of CDB on the SQL client side today, but don't underestimate the time it takes to implement CDB-specific workarounds...

[jimbokun]

"I'd really love some kind of distributed-for-performance database using the exact optimizer and query planner of SQLite plus std plugins..."

Why do you think the identical optimizer and query planner would work in a distributed environment, with no changes from the single server implementations?

[minxomat]

Sure there have to be (minor) changes and some removals, but at the core that's exactly what Comdb2 did, and it works pretty well.

If Comdb2 7 ever reaches stable, and BLP invests some time into deployment, ops and monitoring tools/docs, it'll be a strong competitor.