Did they ever publish the rationale for why they used MongoDB in the first place? They mention having 2.3 million content items. If we assume that they have 100 DB entries for each content item, that's still only 230 million DB items. In that case, was it important to run a sharded cluster vs a typical primary-secondary HA setup? (which they ended up switching to)
I was reading the article waiting to get to the part where they handle their shards can scale. It ended up just reading like a migration from one db to another.
I'm guessing their original choice for MongoDB was more for the schemaless development flexibility rather than horizontal scaling capabilities. This seems to have worked well enough for them. DynamoDB seems like a natural fit coming from a document store and they did very well not to choose it.
I think AWS is due for a managed NewSQL store comparable to Google Cloud Spanner or MS Cosmos DB.
I wondered about that, yeah. That's just Not That Much Data; too, they're clearly able to sustain multiple-minute downtimes (the blog asserts as much). The original design reeks of over-engineering.
Great blogpost, though; well written, which is always a surprise.
Considering they've been using it from 2011 or something, there can't be any good reason but a dumb choice by tech team or management bought Mongo's sales talk without good verification.
