Open to the core: MongoDB's enterprise push in 'joins' U-turn MongoDB grabbed headlines last week with the release of version 3.2 of its popular NoSQL database. Consistent with the company’s prescribed messaging, the tech media dutifully inserted “enterprise” into every headline, touting MongoDB’s new storage engines for better data security, among other things. But one thing was missing …
I agree.
Most of these companies take a really stupid approach, always trying to milk large enterprise, its not only MongoDB, another company that really annoys me doing this is SeveralNines with ClusterControl. Where the vast majority of users are probably SME's with very small budgets, up til a couple of years ago $10k was about 1/2 my annual software budget.
They are so short-sighted, sure its better to have 1000 customers each paying $100, than 100 customers paying $10k. The next effect is still $100k, and the damage from losing a few customers is far less impacting.
Hilarious. This might be an easier sell if Mongo (and the rest) had promoted their tools as large scale document management or file systems, but if you insist on calling the product a "database" and implicitly inviting RDBMS comparisons then you're simply asking for trouble if you try to present the single most basic relational feature as "enterprise".
If they want to charge for something, implement bulletproof ACID. Lots of freetards won't need it (or understand it in some cases) but backside covering management might pay up as part of their P45 avoidance strategy.
[*] I use Mongo myself as a storage engine for tens of thousands of JSON packets. It works great for that, but it is functionally indistinguishable from dumping INI files in a sub-directory - the database in this system is PostgreSQL.
I've never liked OpenCore software. Sun tried to do it with MySQL (I expect Oracle are probably continuing that theme), and the biggest issues are not cost, or having to have licenses for each developer instance (I've not run a dev stack on my laptop for years, its easier to have it on a VM - connectivity is a minimum requirement for any work these days).
It's really about how well tested those features are. With OpenCore, the newest and flashiest features are used by the smallest section of the user base. The OpenCore company will tell you that their Enterprise paid-for features are the most well tested of all their features, but this is bollocks - the main testing of all open source software is when all these different users start using it in all the different ways that they need to use it, and not just in the few ways that the company who developed it have envisaged. With OpenCore, there just aren't as many users exercising the "enterprise" features, and the bugs don't actually get fixed until the feature is in the core.
It's some fucked up version of enterprise software, the enterprise version is the least tested version, and has the most unstable and newest features in it. WTF.
You don't have to have this model - take Lucene/Solr. Most of the development of this takes place by people employed by a few companies, chiefly LucidWorks. LucidWorks offer consultancy and support, chiefly for people who need help customizing the software they have written. It works much much better IMO.
The same approach you suggest seems to be working for EnterpriseDB: helping companies to come down off their hideously expensive Oracle habit.
This whole article smacks of sour grapes from Mr Asay who has drifted from company to company (now with Adobe it seems) where he seems to have championed the open core approach.
I'm pretty sure that companies will pay something for software though more along the lines of support and maintenance than the mere privilege of being able to own something. The industry also has to adjust to years of over-charging for minimal added value: the core of lots and lots of enterprise stuff has always been open source.
Why don't old fashioned "enterprise wide" licenses - with or without tiers - solve this problem?
"Free for student and non-commercial use" gets the product into lots of hands for "testing".
In my experience "consultancy" is more often about finding ways around product bugs anyway in order to meet project timeframes. And I'm not talking about Open Souce products only. I've seen some... very awkward... iterations of commercial software that simply did not work as advertised.
Surely the point is to be "a lot" cheaper than MS or Oracle for Enterprise and massively cheaper at web scale (eg charge by #instances or charge by #employees whichever is the cheaper model).
4WIW Sun Microsystems in their naive, well-meaning, partly competent, partly delusional way did try this (eventually) with the software products inherited from Netscape but the sales force was naturally structured to focus on $100k-$10m hardware deals. Disrupting that in the mid- dot-crash era to add some focus on $10k-$100k software deals as well was never going to be easy. Particularly when the execs didn't have the balls to buy a bloody database to counterweight the business' overwhelming dependency on selling hardware to run Oracle (and Peoplesoft and BEA... see the pattern?). The execs closed their eyes and went "lalala nothing to see here... we are a giant of the industry and Larry really really wants to be our friend" until it was too late then overpaid massively and we know who bought who in the end.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
