As a precursor to this (excellent, in depth) post I also recommend Martin Kleppmann's Designing Data-Intensive Applications, which is (to date) the definitive 101 book on the topic.<p>He also did these awesome Tolkien-esque maps of the database engine ecosystem: <a href="https://martin.kleppmann.com/2017/03/15/map-distributed-data-systems.html" rel="nofollow">https://martin.kleppmann.com/2017/03/15/map-distributed-data...</a><p>Anyway, I inject this sort of stuff directly into my veins, so thanks very much for the post!
Good post! May sound more credible if you have mentioned that YugaByte started as a fork of Apache Kudu and some of the design decisions and technical challenges that you mention were addressed by the code that was already there before your fork.
The article makes mention of Aurora and Spanner as inspiration; did CosmosDB not exist yet? I realize Cosmos is really more of a document database but given its ability to be queried with SQL in addition to massive horizontal scale and global replication seems like it should be more or less in the discussion.<p>Reference to querying Cosmos with SQL: <a href="https://docs.microsoft.com/en-us/azure/cosmos-db/how-to-sql-query" rel="nofollow">https://docs.microsoft.com/en-us/azure/cosmos-db/how-to-sql-...</a>
Following research papers from academia also explored similar problem:<p>1) BESPOKV: Application Tailored Scale-Out Key-Value Stores
<a href="http://people.cs.vt.edu/~butta/docs/sc18-bespokv.pdf" rel="nofollow">http://people.cs.vt.edu/~butta/docs/sc18-bespokv.pdf</a><p>2) ClusterOn: Building Highly Configurable and Reusable Clustered Data Services using Simple Data Nodes
<a href="http://people.cs.vt.edu/~butta/docs/hotstorage16-clusteron.pdf" rel="nofollow">http://people.cs.vt.edu/~butta/docs/hotstorage16-clusteron.p...</a><p>DynomiteDB from Netflix allowed to scale singe server NoSQL stores. <a href="https://github.com/Netflix/dynomite" rel="nofollow">https://github.com/Netflix/dynomite</a>.
That wasn't immediately relevant to me, but is definitely a good read since I imagine I'll be dealing with these problems within a year.<p>Thanks!