>Transactions in no way imply durability. I don't think that's true. Transactions are commonly defined as durable (e.g., in https://msdn.microsoft.com/en-us/li

>Transactions in no way imply durability. I don't think that's true. Transactions are commonly defined as durable (e.g., in https://msdn.microsoft.com/en-us/library/aa366402%28VS.85%29.aspx https://msdn.microsoft.com/en-us/library/aa366402%28VS.85%29...). The term "transactional database" is used synonymously with "ACID database". >If you mean that you should compare them, then I agree. If you mean that you should benchmark them against each other, then no, you're going to get junk results because they're very different. You're right. One could benchmark a non-ACID datastore against SQLite but only if each ran in an in-memory database mode. >That benchmarks seems to have fallen in that trap, as it seems to be comparing durable with non-durable datastores with no mention of the persistence setting. I agree with you per the above, however, the results from SQLite and Vedis in particular appear to be meaningful to compare. Edit: One thing I've just noticed about Vedis is that it is dual-licensed under the Sleepycat license and a paid commercial use license. The Sleepycat license is a strong copyleft license that requires that you ship the code to "any accompanying software that uses the DB software". It's something to have in mind if you want to use Vedis commercially. By contrast, rlite has a 2-clause BSD license.