MongoDB's CEO: Expect aggressive investing as biz aims at Oracle et al Dev Ittycheria, chief executive and president at MongoDB, took a moment out from the company's European event to tell The Register about its strategy for the future. The NoSQL business is approaching its tenth birthday, but plans for neither profit nor exiting are on the immediate horizon. Due to the hundreds of millions in …
It seems that he's saying that their popularity is money-fuelled?
Out of curiosity, anyone here with experience using MongoDB in production for at least a year or so? Happy with it? Thoughts?
Any other NoSQL users being able to share a success story?
As for me, I evaluated and discarded NoSQL solutions a few times over (I really am open-minded about it, but over the years one has also developed a nose for bullshit), with two exceptions: Memcache between 2008-2012 or so, and currently using Redis on a pre-production stage project, and so far I'm very happy with it: it's flexible, stable, uncomplicated, very well-documented, prioritises performance over features, and the code is very clean. If I were to invest in a NoSQL business, I know which one I would chose, and it would not be the ones saying "well, the other guys took decades to get it right, so what's the rush?"
Anyway, honestly interested in others' experiences.
I have used MongoDB in production and I am considering to become a certified developer. Not because I love MongoDB, but because I good make a good living as a consultant by redesigning badly thought out schemas or migrating MongoDB back to a SQL based environment. Personally I only use MongoDB for storing JSON retrieved from external APIs. You only need a few lines of Python for this and can be very useful if you want to scrape additional fields at a later stage. However, it is less useful for storing big data, especially as I found the integration with Apache Spark (beginning of 2016) not very good.
Postgres provides you with enough options for when you need to keep this kind of data around, and, as a bonus, it won't magically lose it for you. This is the "o" in NoSQL I was recently at a talk on Map/Reduce and the volume of transient (OLAP) data you need for non-relational systems to have an edge is staggering.
Otherwise stick with Redis for the transient stuff if it's working for you.
Thanks Oxford and Charlie for your insights.
The "back to SQL" thing I've been hearing a lot from my local contacts, so not surprised you've made your niche out of it!
JSON processing appears very solid in PostgreSQL indeed, although I have no production experience with it, but since we do use PostgreSQL already, this is likely to be carefully considered next time we need a document store type of thing. Plus, I like JSONB.
> the volume of transient (OLAP) data you need for non-relational systems to have an edge is staggering.
This! I think this is what really pissed me off last time we looked at Mongo (apart from the horrible documentation). We know that disk is cheap these days, but Mongo seemed to be taking the piss. I don't recall the exact numbers but it looked like we would be having to provision an order of magnitude more storage than needed to actually hold the data¹.
I do not want to knock it down too much, since we just run an evaluation (well, two) and we felt we were not comfortable with it, but we could have well made the wrong decision or it could have been down to our lack of expertise or other extrinsic factors, not the product itself. However, still not too thrilled by a company whose CEO only ever seems to talk about how much money they have to burn. That's cool, but how about the actual product?
¹ In the end, our solution relies on raw data written on disk with just the indices in Redis. This has opened up many more possibilities that we hadn't considered before and things are running extremely fast.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
