Suck on this, Larry: NoSQL pair hit the G-spot

There has been plenty of talk about open-source databases growing faster than those belonging to tech dinosaurs like Oracle, but nothing tells a story quite like cash. MongoDB and Basho, pioneers in NoSQL, yesterday separately announced that they are each receiving further wads of cash from the venture capital sources. Open- …

[DreadPirateRobot]

"Basho reckons to have displaced Oracle at the NHS in the UK – although it doesn’t say where or how much"

It will be the DB running Spine2. That was rolled out earlier this year and it has a NoSQL backend, it was Oracle SQL formerly. I used to work for a medical record software company that integrated with Spine and now Spine2.

[A Non e-mouse]

Statistics, statistics, statistics

There has been plenty of talk about open-source databases growing faster than those belonging to tech dinosaurs like Oracle..

New open-source database sells one license in year 1, and two licenses in year 2. That's 100% growth in one year.

Oracle sells 1 million licenses in year one, and 1.1 million licenses in year two. Oracle "only" has a 10% growth.

[Anonymous Coward]

Open Source crackware.

1. Get people using the product extensively.

2. Wait for the investment money to come in via big business.

3. Change the licence or sell it to a big IT vendor.

4. Retire.

[Andy 102]

"Basho reckons to have displaced Oracle at the NHS in the UK – although it doesn’t say where or how much.."

They probably heard about it from here maybe....

http://www.theregister.co.uk/2014/09/09/nhs_spin2_rips_out_oracle/

[disgruntled yank]

As the man said, above

Venture capitalists pouring buckets of cash into this or that company proves only so much.

[gary27]

300m and counting yet slower than postgres

Test with Postgres 9.4 new jsonb datatype and faster indexing showed postgres to be faster than mongo at processing / storing json data.

So I'm confused as to why any sane person would pick the slower, less functional, less robust, no acid Mongo over Postgres!

Also much less risk with Postgres, since when developers finally come to their senses and realise that the relational model is far superior, to ancient hierarchical model ( think IMS, XML) they have the option to, much more easily switch, via joins within the same db.

[Anonymous Coward]

Re: 300m and counting yet slower than postgres

Indeed. Please, before using an immature and unproven tech, try to make Postgres sweat. It will lilely be worth it.

[Charlie Clark]

Re: 300m and counting yet slower than postgres

Add reliability to the list of reasons as well. As opposed to MongoDB's ability to vote as to which node is the master as the fastest way to lose data.

[Anonymous Coward]

Re: 300m and counting yet slower than postgres

I guess I dropped Mongodb when I discovered your objects can't be larger than 16mb, then you have to use a different, quirky API called GridFS. It's still that way. Defeats the purpose of having one humongous db for everything.

Then we learn Postgres has faster I/O, and it's added a nosql-esque flexible data type, and it has full text search... on top of a nice robust relational model. I would love to ditch Solr (Java, grrr) and use Postgres for everything -if IT lives up to these expectations.

[Paul Hovnanian]

Statistics

"Mongo says it is used by more than 2,000 customers, including 34 of the Fortune 100, with nine million downloads."

My Debian distro laptop ran mongodb. I don't know why. I discovered it one day when it seemed to be running a bit slow. So I ran a process list and found 'mongodb' to be hogging CPU cycles. I killed it and haven't noticed anything important not working.

So I'm not really sure what those 9 million downloads means other than it's a depenency of some little used utility tht people pull in when they run a package manager.

[ben_myers]

Lifeblood of company in SQL

If the lifeblood and guts of your company are in an Oracle database, you need to make damned sure that everything works perfectly before moving to ANY other database. Unfortunately, very few enterprises probably adhere to sound development practices like strict separation of database manipulation logic from presentation, so you can't just replace the back end. And a lot of Oracle databases are legacy ones and the designers and developers are long gone, leaving few traces of documentation behind.

Oracle will be a tough nut to crack and replace. Now, if somebody came up with a mapping layer to transform SQL calls into whatever NoSQL uses (not even sure if feasible), and with performance comparable to Oracle, then Oracle and Larry would be sailing around in deep doo-doo.