I also know that there are some specific technical areas where MongoDB can pose a serious threat to Snowflake in the future…
I don’t see that so much:
-
MongoDB is a NoSQL (non-structured or semi-structured) database only. Snowflake is primarily a structured database that also handles semi-structured data.
-
MongoDB does not provide an SQL interface, Snowflake provides SQL access to both structured and semi-structured data.
-
MongoDB is primarily a Transactional (OLTP) database while Snowflake is primarily an Analytical (OLAP) database. Now, both can handle the other type of load, but MongoDB’s analytical performance suffers greatly while Snowflake’s cost suffers greatly when doing the other.
-
I should probably do a deep-dive into the differences between, for instance:
• Cloud native
• SaaS (Software as a Service)
• Software as a Managed Service
• Software provided as a set of Serverless APIs
But, for now just know that MongoDB’s Atlas today is cloud native and SaaS, but it’s arguable whether it’s truly a Managed Service. This is why a third party, Digital Ocean, has a MongoDB as a Managed Service offering! And don’t forget, MongoDB itself has a Serverless version of its database in Preview/Beta today, so that’s going to change (Digital Ocean investors or potential investors need to bone up on this ASAP).
Snowflake is only offered via Serverless APIs. Digital Ocean doesn’t provide any Snowflake services because, well, Snowflake is already a fully managed service that automates just about everything for you.
So, I don’t see MongoDB as any kind of threat to Snowflake. I do, however, see Snowflake as a partial threat to MongoDB, as MongoDB has been trying to push itself as an Analytical solution, for which it’s not ideal. Matter of fact, there are documented workflows that use the Transactional capabilities of MongoDB to act as a “front-end” to a Snowflake instance. MongoDB is great for capturing the data from thousands of IoT devices, for instance, and then batch copying that data over to Snowflake where analytics are performed.
And note that MongoDB is even limited in the Transactional realm, as it’s not a structured database and so isn’t really suitable for many kinds of applications (financial for one) that use structured data and are typically implemented using an SQL interface. This isn’t to say that MongoDB doesn’t have its uses, but when looking at the TAM (Total Available Market), I have to believe that Snowflake’s is much much larger. Very very few companies will migrate from a Teradata or Oracle database to a NoSQL database like MongoDB, for instance.
That said, there are many use case for semi-structured data and the proliferation of things like IoT devices means that such workloads are expanding. So there is definitely a need for something like a MongoDB, and I did take advantage of the sympathy drop in MDB earlier this week to (re)start a small position.