They should use SQL Server (which has great replication abilities, although horizontal scale out is still difficult) or MemSQL (which is distributed, scalable, and can do everything they need).
Or use Cassandra which is a perfect fit (or ScyllaDB which is a better version of it).
This all sounds like an aversion to just paying for or using better products when the problem is easily solved.
SQL Server needs a couple hits with a cluebat to even satisfy the basic three requirements for a database system they listed (I'm talking about the MVCC line of course).
Honestly, if a commercial database provides what you require and you have the budget? Sure, why not. But I'd always try to avoid that myself, because the serious players (MS, Oracle, is anyone still using DB2?) are really, really expensive.
SQL Server works just fine and has lots of concurrency control to do whatever they need. And the way they're using the database doesn't seem to really make this an issue outside of their replication.
> are really, really expensive
This whole janky setup they have sounds even worse. None of the commercial relational databases are really that expensive considering what they offer, and we're talking about Uber here. We're a small startup that pays for both.
Money for (a better) working product with support is the right call, not build it yourself. This is just poor tech decision (outside of using a RDBMS in the first place).
Those are all massive scale tech companies that actually needed to invent many of the datastore technologies used for big data today. Uber is not among them.
But I'm sure you know all this as it seems you work for an enterprise database company that actually makes the exact product Uber should use.
Or use Cassandra which is a perfect fit (or ScyllaDB which is a better version of it).
This all sounds like an aversion to just paying for or using better products when the problem is easily solved.