I entirely agree the vendors are (partially) to blame here.
Locking RBAC and TLS behind a paid subscription is a sure way to force companies with security teams to pay for it (or not to use it).
This particular lesson comes relatively cheap--dropping your data is not the worst an attacker could do with it. Hopefully, more people will research what I'd call "security 101"...
Don't make excuses for their shitty business practices.
This probably isn't an excuse for Elastic anymore, but it's how Elastic was born ("easier than Solr!" -- they didn't invent full-text search, after all) and it's how whatever supplants Elastic will be born. Why is MySQL dominant over Postgres even though "referential integrity" didn't come to the game until v5? Why Docker over jails or OpenVZ? etc. People adopt technology because it's fashionable and it becomes fashionable in part because it's perceived as easy to use. Security and ease-of-use are not quite true opposites, but there's definitely some intrinsic tension.
We have a lot of hapless practitioners in the space and the root problems here won't go away until we get some standards. Better solutions usually lose because crappy stuff focuses more effort on marketing and an "easy" onboarding process, where better stuff focuses on the operational complications of the real world.
It _can_ be. At the same time some of the most used software in the world manages just fine without a company running paid subscription plans and locking free users out of critical security components.
ElasticSearch/MongoDB/Redis et al are trying a new model for how to create OSS with a company behind it funding all/most of the development. That's OK, and I'm super interested to see how it works out long term. But there are many many counter examples of similar sized or way bigger software projects that never needed to do this. Pretty much everything that those three databases depend on to be used in applications is OSS that never had a "paid subscription" locking access up. How useful would any of them be without Linux, or Apache/Nginx, or Ruby/Python/PHP/Perl.
My fear is that half-assed-OSS that does a _great_ job of "capturing developer mindshare" but a lousy job of securing free use of their software - is one day going to be the root cause of some _spectacularly expensive_ data breach, after which pointy haired bosses and less technical C suite suits are going to feel the full power of Oracle's golf-course-and-expensive-restaurant marketing army, and nobody in a company bigger than 10 or 12 people will ever be able to use any database with less that a half million a year license because "due diligence!" and "risk mitigation!" (and "Waygu steak with expensive whiskey" and "dirty free software hippies exposing you to data breaches!!!")
That said, I've incidentally heard a lot about Mongo in the last half a year. Might be my bubble. Might be MongoDB actually maturing and getting really good. I hope it's the latter.
For everyone else, JUP ("Just Use Postgres")
Snark aside, Postgres should hold you over for quite some time.
With Postgres you probably don't need Redis, Elastic Search, Mongo or Kafka.
An example of an issue I am dealing with currently: while you can create a gin index to speed up containment queries, Postgres doesn’t keep any statistics about jsonb columns. This means the query planner will sometimes do stupid things, like using the index even for very non-selective overlap conditions, which is a lot slower than just doing a sequential scan.
Less of an issue for me but worth considering: the size of the gin index in my use case seems to be about 5x bigger than the size of the unindexed data. I was surprised by the size increase. I only use the containment operator so I could make a smaller/faster index using the jsonb_path_ops operator class. This is on my todo list :)
Like all non-btree indexes in Postgres, the index is unordered. That means sorting by values in the jsonb column will always be slow. This doesn’t matter for selective queries, but exacerbates my already slow non-selective queries that return large result sets.
That said, if your queries are selective, jsonb + gin indexes are surprisingly performant (in the 0.5-10ms range for small result sets). My use case is a mix of structured relational data with jsonb for user-defined values (which of course they want to use for querying/sorting and I was dumb enough to say “sure, why not?”)
In terms of the magnitude of data, there’s roughly 10 million rows. Each team using this service has the query scoped to about 500k-1 million records, and then additional filters (on the jsonb column) will scope that down to anywhere between 60k-0 results.
I ask because where I work we sync postgres to a secondary store for search, but the way it's done in a piecemeal, application-specific way gives me the heebie jeebies. It almost certainly will result in that secondary store drifting. Unfortunately we can't use something like zombodb [1] as we're on amazon RDS. It seems like you know your stuff, and seeing non-deterministic consistency irritates the heck out of me!
But I'll be happy to replace it by something else, my load is extremely small and only single requirement is to have DB as network daemon, not as embedded storage as it will be used by 2 applications (main daemon an API).
RethinkDB was really nice candidate for it but it's not alive anymore: https://rethinkdb.com/blog/rethinkdb-shutdown/
- Proper replication in its first-class, default configuration
- Automatically managed cluster membership
- Seamless automated failover
- First-class async client libraries in many languages
- A non-awful query language
Focusing on the JSON is beside the point (though it is a convenience). Wake me up when you've got a properly distributed database.