In Febuary of this year Google re-launched JotSpot as Google Sites. Google had acquired Jotspot some 16 months earlier, during which time Jot was only available to existing customers and closed to new signups. What happen during those 16 months and why did the process of integrating with Google take so long? Looking through the list of companies that Google has acquired, Jotspot would be considered lucky as many others have died, stalled or lost out to competitors because of the acquisition process.
Blogger was acquired by Google in Febuary of 2003, and at the time it was the leading blog platform by a wide margin. Within a few months, MovableType had taken over the self-hosting market, followed by Typepad and then WordPress and WordPress.com. In the interim Blogger had stalled at Google, with no new feature releases, no improvements and a lack of support.
In 2005 Dodgeball was acquired – a potentially early Twitter or cool location based service, and it died inside Google. In Febuary of 2006, MeasureMap, the blog analytics tool, was acquired and never heard from again. GrandCentral went to Google last year, for $45M, and since then the service has been frozen with no new users allowed to signup and sporadic periods of downtime (meaning users cant get any phonecalls, at all).
One of the first main challenges for a company that has been acquired by Google is adopting the proprietary technology stack used within the company. Google does use Linux and open source, but their core technologies are all internal to the company. I have heard that it can take a new engineer at Google anywhere from 3-6 months to become accustomed to using these tools and services. The table below sets out the Google stack and the technologies used:
Google Technology Stack
|C++, Java and Python||Core libraries and components in C++, web applications in Java (Google Web Toolkit) or Python (not as common)|
|MapReduce||Distributed computing library and cluster. Written in C++ can interface in Java or Python|
|Big Table||Distributed column-oriented data store with query language.|
|Google FS||Large-scale distributed file system. Used for object/file storage|
Because of the difference in technology, it can take a company anywhere from a year to three or more years to move over to the Google infrastructure and architecture. Blogger was still running their own infrastructure until their new release last year, and they have finally integrated Google ID’s. YouTube is one of the only recent acquisitions where full steam and emphasis were placed into getting the site moved over to run on the Google platform. YouTube managed to pull it off, but it is a rare case inside Google (and also a special case).
The problem isn’t one that is unique to Google, as the other big web and technology companies such as Microsoft have also struggled with or simply neglected some of the companies they acquire, but there are problems that are very specific to Google because of the technology they use. Microsoft develops their products using .NET and the Win32 API, there are millions of developers capable of developing in this environment. There are also millions of open source developers capable of swinging PHP, Python or Ruby. But building on MapReduce and BigTable at Google? There are only a few thousand specialists who are all either already employed at Google or former employees.
And what comes of former Google employees? They spend years building on a technology stack that nobody else is using. How useful are they to a company that is looking for MySQL, Apache, Python, PHP etc. experts? It probably isn’t as big an issue – as the developers can adapt to their environment, especially considering Google is hoarding some of the best developers out there.
The stack of technologies developed by Google has helped them build out their search engine and core technologies quickly and efficiently – not to mention at a fraction of the cost of what they would have paid using traditional clustering and hardware. But when it comes to bringing external technologies and companies in house, the Google technology stack is a tax on developers which slows down development to a point where it can kill a product.
The solution for Google is to either adopt a more open stack in parallel to what they currently use, or to open source their internal technologies (as Facebook and Yahoo! are doing) in the hope that they will spread and gain adoption from more developers.