December 30, 2004
@ 08:38 PM

In Adam Bosworth's post Where have all the good databases gone he asks the Open Source community to target some problems with relational databases that the Big 3 vendors have seemingly been unable to solve.

Krzysztof Kowalczyk has an interesting response to Adam Bosworth's post entitled Google - we take it all, give nothing back where he writes

Open-source - not working as advertised.

The popular theory ("myth” would be a better name) is that open-source works because of this positive feedback loop:

  • source code for product foo is released
  • it’s free so it gets used
  • if it doesn’t fully meet someone’s needs, that someone can code the functionality (since the code is open) and submit the changes back to project (something not possible if you use closed products like Windows or Office or Google)

  • those contributions improve the product for everyone else, so more people use it so more people contribute the code and so on. Sky is hardly the limit.

The good thing in this theory is that it doesn’t rely on kindness of strangers but on englightened self-interest of those who benefit from free software. The bad thing about this theory is that in theory it works much better than in practice.

It’s all because of a weblog post by Google’s Adam Bosworth. Read it yourself, but the gist of it is that, according to Adam, commercial database vendors don’t understand the needs of companies like Google or Amazon or Federal Express. Relational database rely on static schemas and there are no good ways to dynamically reconfigure databases without the disruption in service. Adam ends with a plea to open-source fairy:

My message is to the Open Source community that has, so ably, built LAMP (Linux, Apache and Tomcat and MySQL and PHP and PERL and Python). Please finish the job. Do for databases what you did for web servers. Give us dynamism and robustness. Give us systems that scale linearly, are flexible and dynamically reconfigurable and load balanced and easy to use.

This is why the theory of open source doesn’t work in real world. A multi-billion company has a clear need for software that works well for them but instead of investing in existing open-source projects like PostgreSQL or MySQL to make them do what they need, all they do is ask some magic, undefined entity they call Open Source community to do the work for them. For free.

Google - we take it all, give nothing back. Come work for us.

Let’s estimate how much money did Google save by using open source software that they would otherwise have to purchase. The operating system for tens of thousands of their computers. Web servers they use. All the Unix utilities they use. Editors, compilers and debuggers they use to write their code. E-mail smtp server. E-mail pop servers. Languages like Perl and Python. Databases like MySQL and PostgreSQL. It’s safe to say that if Richard Stallman was never born, the licenses for those kinds of software would cost them tens of millions of dollars.

And what does Google contribute back? Where are their patches to gcc, gdb, python, postgresql, sendmail, emacs?

Google - we leave open-source to Microsoft. Come work for us.

It’s very ironic that I can find more open-source code created by Microsoft and its employees ( RSS Bandit, IronPython, Windows Installer XML (WiX), FlexWiki) than by Google employees. Not saying that there aren’t any but they are certainly not easy to find, even when I use mighty search engine trying to find google open-source.

Google - we like our hardware cheap and our software free. Come work for us.

If you’re into this stuff you know that Google is known for it’s highly tuned process of selecting hardware components (i.e. all those thousands of computers they need to index and store the web) to hit the best price/performance ratio. In a way, they use the cheapest thing, when you define the cost as the total cost of ownership (as opposed to simply the cost of buying the hardware). Thanks to Adam’s admision:

Indeed, in these days of open source, I wonder if the software itself, should cost at all?

we also know, that they like their software free.

As a side note, it’s a surprising statement coming from Adam who knows very well that writing software costs a lot. Open-source doesn’t eliminate this cost, it just shift the costs and allows unlimited number of free-riders, like Google.

I’m picking on Google, but they are not alone. Amazon, yahoo, ebay, aol. Any large business that uses web as means of providing services and making revenues is enjoying enormous savings by using open source stack on their back end. And what do they contribute back? A good approximation of zero compared to benefits they reap.
...
But Adam’s example shows that there’s a fat chance of this happening. Adam is not a rank Google employee. He was not hired to give free massage to stressed Google employees. Before Google Adam was a high-ranked executive at Microsoft and BEA. He led teams that created successful products (IE, Access among them). He’s in position to influence what Google does. He understands technology, he understand the cost and difficulty of making software. He has a weblog and deep thoughts. If only he understood the strategic value of open source.

If someone like Adam cannot see further than the tip of his own nose and his ideas are as bold as asking others to write the software he needs for free, then I don’t have much hope for anyone at aol to get it either.
...

Google - do no Evil. Do no Good. Just like everybody else. Come work for us.
...
In those days of focus on corporate profits (where there any other days?), Google’s motto “Do no Evil” is refreshing.

Or is it? It’s a nice soundbite, but when you think about it, it’s really a low requirement. There are very little things that deserve to be called Evil. If a senior citizen is taking a nap outside his house on a sunny day and you kick him in the groin - that’s Evil. Most other things are bad or neutral. Not doing Evil is easy. Doing Good is the hard thing.

I doubt that this is the kind of response that Adam Bosworth was expecting when he posted his plea. The fun thing about corporate blogs is that it gives people more places to read between the lines and learn how a company really thinks. I suspect this is why Google doesn't have many authentic bloggers and instead has favored the press release page masquerading as group blog approach at http://www.google.com/googleblog/.