MongoDB talks up its AI chops by talking down PostgreSQL

At the risk of protesting too much in the shifting database landscape, NoSQL-based MongoDB has attempted to trash the competition by claiming PostgreSQL systems lack scalability to keep up with the demands of AI workloads. CEO Chirantan Desai took the opportunity of largely positive earnings results for the quarter ended Oct. …

[JLV]

MongoDB is AI scale

https://www.youtube.com/watch?v=b2F-DItXtZs

[Autonomous Mallard]

I was just about to post that...

"If /dev/null is fast and web scale I will use it."

[Charlie Clark]

Scaling problems?

"not able to scale with PostgreSQL... and they moved completely to MongoDB."

Sounds a bit like "couldn't drive in the lake with my car, so I smashed the windows out and started waving my arms".

Postgres scales fine, but what do they want to do with the data? And scaling is fine, but don't you also want to keep the integrity of the data?

[theblackhand]

Re: Scaling problems?

"And scaling is fine, but don't you also want to keep the integrity of the data?"

Data corruption allows AI to be extra innovative....

[Anonymous Coward]

Re: Scaling problems?

You can perfectly maintain data integrity with Mongo. Does occasional data duplicates to meet scalability requirements breach the data integrity principle to you? I am not here to trash Postgres which I agree works well in a lot of cases, but I did see clients hitting scalability issues on Postgres once they had more than 2000 writes per seconds to handle. What did these clients do? They denormalised with various techniques like adding caching layers, materialised views, tried to use jsonb columns, split their dbs into microservices… all these techniques are fine but do involve data duplication of some sort and breach the “strong data integrity principle” that a lot of devs seem so worried about. The truth is that building distributed systems will involve some compromises including data duplication, whether you use Posgres ot Mongo. MongoDB’s approach is to make denormalisation the norm. This isn’t wrong and can work very well if handled carefully. As someone who worked in enterprise software implementation for a decade, I saw large banks, telcos and big retailers adopt Mongo for

critical and acid compliant workloads very successfully.

[Anonymous Coward]

Scale out early, and scale out often

Mongodb scales, but why should a cluster need sharding - adding cost and complexity for the customer- at such tiny data volumes [70% of max 4TB from the M50 tier upwards]?

[Ken G]

Re: Scale out early, and scale out often

About to say, maybe the AI is needed to manage the sharding.

[DrewPH]

Mongo comparing itself to SQL?

In other news, a zebra compares itself to a bungalow.

[Tim99]

Pause for hysterical laughter

MongoDB and AI together, what could possibly go wrong? A new example of write-only data?

[Charlie Clark]

Re: Pause for hysterical laughter

Just don't mention DuckDB to them… consistency and transrelational…

[werdsmith]

Some years back I built a good multi- resilient shard test lab with MongoDB freely downloadable for evaluation.

This year we tried to repeat the exercise and came up against a minefield of monetisation.

[Bebu sa Ware]

"super-high growth AI company", "AI·native companies" …

the move fast, break things crowd ?

Their avoidance of PostgresSQL is probably a selling point for the latter.

The imminent demise of RDBMS must rank with fusion power and flying cars—none of which is likely soon and the first the least likely.

[AMBxx]

Re: "super-high growth AI company", "AI·native companies" …

>> create governance as they go,

What could possible go wrong?!