Oracle buys Sun

Generally I don’t comment on events in the business world, since this blog is about web development and software engineering. However, the acquisition of Sun by Oracle, which was officially announced yesterday, is so large-scale that it is likely to affect the engineering halls in many subtle ways. Quick facts: the deal is $ 7.4 billion worth, it was unanimously approved by Sun’s board and it will be closed this summer. Sun and Oracle published identical press statements yesterday which sing the praises of the acquisition.

I am not sure whether this is good news. While it was apparent to most observers that Sun was past its zenith, one wonders what will happen to its employees and its innovations. Granted, an acquisition by IBM would have tipped the scales even more in favour of Big Blue’s dominance in the enterprise market and that might have distorted competition. But one may doubt that Oracle will uphold Sun’s commitment to the open source community. Sun’s market was driven by innovation and open source products. Oracle’s market is clearly not.

In particular, one wonders what will happen to MySQL which was bought by Sun earlier last year and which competes with Oracle’s core products. Pessimistic observers have already called it MyToast. Will Larry Ellison allow MySQL to compete in the enterprise market? Probably not. Other items in Sun’s portfolio once considered crown jewels, such Solaris and Glassfish, might also be on the endangered list. But that is pure speculation at this moment. Whether this acquisition will turn out to be a good move for Oracle is currently debated by the industry experts. Whether it is a positive turn for open source community may be reasonably doubted. Time will show.

The triumph of the lamp

It is human to feel satisfaction when one’s predictions come true. To predict the success of LAMP in 1998 wasn’t that difficult, but neither was it a no-brainer.At the time, the acronym had just been coined by the German c’t magazine and it wasn’t widely known in the corporate world. LAMP stands for Linux-Apache-MySQL-PHP, a set of open-source software that powers web servers with dynamic content. Occasionally, the ‘P’ in LAMP is switched for Perl or Python, although PHP is now by far the most popular scripting language.

carbidelamp.jpgI remember having suggested a LAMP architecture for the implementation of a geo information system and extranet for a large government organisation in 1998. This project was on a tight budget, so I proposed to invest the resources into software development rather than into hardware and licenses. LAMP seemed ideal for it. However, the committee was utterly surprised that the word “Microsoft” did not appear in the proposal and they did not seem to put too much trust into any of the letters of L-A-M-P.

Luckily, another of my then customers was more open to the suggestion. A mid-sized logistics company was looking for new way to do business on the web. Since the company wanted to run their own servers, the LAMP stack offered a perfect solution to do this cost efficiently. It turned out to be foresighted decision. LAMP quickly gathered momentum on the Internet and soon became one of the mainstream web development platforms.

One has to keep in mind that the four pieces of software that make up LAMP have not been designed as a unified platform. On the contrary, they have different histories, and they were not specifically developed to work together. This is what distinguishes them from their competitor platforms ASP/.NET and Java/J2EE.

Let’s briefly go back to the year 1998. It was the time of the browser war and the boom. The Internet then consisted of about 5 million websites, which is less than one tenth of today (2005). Linux was at kernel version 2.8.x, Apache was at 1.3.0 and the Apache Software Foundation was not yet founded; MySQL was at version 3.21, and Andi Gutmans and Zeev Suraski just released the crucial PHP3 release.

Linux and Apache were already strong at the time. Linux was fairly mature OS with a userbase of 7.5 million in 1998. Torvalds had just trademarked the Linux name and the corporate world started to take notice. The Apache web server -originally developed as an extension of NCSA httpd server- already commanded 50% market share. MySQL and PHP, on the other hand, were new kids on the block. PHP had a userbase of only several ten thousand users and MySQL was widely considered a toy database.

The combination of these products, however, did one thing extremely well: powering dynamic websites. Plus they were free. Anyone who wanted to run a web server could use them without paying a single cent for license fees. Although MySQL and PHP had several limitations at the time, it did not really matter for the purpose of serving web pages. MySQL delivered impressive performance and PHP 3 was “cleaner” and easier to program than Perl. Thus the LAMP stack offered a killer platform for web applications.

Today, the quartet is more successful than ever. Apache powers 70% of all web servers. MySQL is the most widely used database on the Internet. It has matured into a full-featured RDBMS with transactions, replication, clustering, and (as of 5.0) stored procedures, views and triggers. PHP is now installed on 20% of all Internet sites. It fully supports the OOP paradigm and it has grown an extremely large function library that makes programmers feel like boys in a candy store.

What is behind the success of LAMP? In the case of Linux and Apache this is fairly easy to tell. They are both free and they offer excellent performance. You can easily pack 50 to 100 low traffic web sites onto a commodity PC. This makes Linux + Apache very popular with hosting companies who take advantage of the low cost of ownership. Furthermore, it allows service providers to customize the server’s configuration and administration model and apply it to an arbitrary number of cheap boxes which translates to arbitrary scalability. Hosting companies love it.

The case is slightly different with MySQL and PHP, because their growth is driven mainly by developers rather than by hosting companies. From the start, MySQL was geared towards web applications, which means its main strength is concurrent reads. Besides, it is easy to use and administer. PHP has become the scripting language of choice because its learning curve is almost zero, which means that programmers can be productive with PHP from day one. In addition, PHP is integrated very well with Apache from a very early time. It uses resources efficiently and avoids the CGI model and its known security problems.

All of the LAMP components originated around 1995 or before, which means that all of them recently passed their 10th birthday. LAMP can now be considered a mature architecture. The case of LAMP proves that using an open source platform is not like buying a pig in a poke. It proves that open source gets the job done, and -in the case of web servers- that it does the job better than anything else.