The panel was moderated by Steve O'Grady (Redmonk), and its participants included Bill Hilf (Microsoft), Tim Bray (Sun), Mike Olson (Oracle), Mårten Mickos (MySQL), and Anant Jhingran (IBM).
Bill explained that Microsoft's "ulterior motive" behind its support of PHP is that, as a tools company, when 75% of PHP developers are devloping on Windows but deploying to UNIX, well, that's something they want to resolve, so that more people deploy PHP applications to Windows.
Tim pointed out that Sun machines runs Linux "just great", in the event someone doesn't want to run Sun's OS. Tim mentioned that Solaris is one of the most observable operating systems available, pointing out the integrated DTrace debugging tool.
Mike reiterated that Oracle last week announced full support for Linux for Oracle.
Mårten says that the strength of open source is that there are multiple projects at once, and that each can focus on making their tools faster - Zend can work on making PHP faster, while MySQL works on making MySQL faster. He expects that PHP and MySQL will continue to evolve to support the needs of Web 2.0 applications.
Anant briefly talked in general about how IBM iSeries servers are extremely reliable - and tat the difference between "Windows" reliabilitnd iSeries reliability is night and day.
Tim says that it's obvious that a large portion of the market has made it clear (with their wallets) that open source is the way people want to go, that there are strong engineering reasons for continuing to support open source development, and that corporate involvement with open source tools is only going to continue to grow.
Mike says that Oracle was led to open source applications - particularly PHP - by their customers, who wanted to use Oracle with PHP. He says that Oracle isn't trying to subvert open source applications, rather that they want to make their customers successful, and if that means supporting open source, then that's what it means.
Bill talked about how one of the key qestions of working with open source tools is, "does this help us make more money"? - And any company needs to ensure that they are.
Question from the audience along the lines of - "Buy Oracle, which is Fast, or MySQL, and then buy people to make it fast?" The answers from Mike and Mårten were essentially "given what you're trying to implement, it'll take engineering effort to determine what the best solution is". Tim (I think) said that he's never seen a large-scale application deployment succeed without a substantial investment in people. Bill chipped in that it's really all about the staff - people that aren't good won't result in good performance. Anant talked about how TCO was a strong consideration for IBM's clients.
Mårten pointed out the distinction of "stacks" from a customer and vendor perspective - from a venros's perepective, it's "how can we lock a customer in?". Tim points out that the the most important thing is freedom - esecpailly the freedom to leave your current vendor. Tim thinks that changing the database is the hardest part of the whole stack - changing OS and hardware is easy, and not hard (though, not necessarially easy, at that) to mix software environments in a given application. Mike actually said that - pragmatically, you don't want to change the underlying software or hardware because you need a stable foundation to work from, and no matter what you start from, it's always going to be hard to change. As soon as you build a system that has to work, you are locked in, to the platform you chose. Bill talked about how the important thing is that given an existing investment in tools, to ensure that there's an evolution path, and credits LAMP's success to this - the loosely coupled nature of the LAMP stack forced communication and open standards in each of the tools, allowing them to work better and be replacable if necessary.
Tim talked about how for many years, Sun's answer was "Java. What was the question?", although they're moving towards more dynamic languages in the future. He also thinks that there will be "no winner" between PHP, Java, .NET, Ruby, etc - they exist, and they'll have their own uses. The interesting problem, he thinks, is that given all the different languages and technologies, the key challenge will be integrating them all together.
IBM has invested significantly in PHP. Anant said that he wants to see more enterprise-class applications in PHP, because then it makes it easier to sell enterprise-class IBM servers.
In closing, the panelists talked about how the'd like to see PHP continue to evolve. Bill wanted to see PHP continue to use FastCGI, and sees potential for desktop PHP applications. Tim thinks PHP is pretty strong, but talkefd about the proliferation of passwords and identiy, and thinks that identity managment is something that needs to addressed, and pointed people at Pat Patterson on Identity Management. Mårten was happy that Zend was doing both ZendBox and Zend Core, and hoped that there would be additional ready-made downloadable stacks in the future. Anant thinks a perfect storm is coming regarding information managment.