Today, hot on the heels of Zack's interview for Drupal Voices, I'm happy to announce the latest update in the Project Mercury AMI series, a 0.51 release that for the first time includes both 32 and 64bit infrastructure, meaning you're free to run these images on Amazons larger and more powerful instances!
In addition to offering 64-bit support, this release also consolidates some important changes in the stack infrastructure. It utilizes the latest version of Pressflow (including Drupal Core 6.14) as well as keying of Pressflow's lovely new home on Launchpad (the perfect place to get involved w/that great project, btw). It also uses an updated version of Cacherouter, which we're hoping to work with Steve to bring to a 1.0 release in the near future.
The manifest paths for the new images are:
The easiest way to find them is still with a keyword search for Pantheon or Mercury.
In addition to making images available in the EU, a 64-bit version was one of the main requests I've received, and I'm glad to give people looking to run this stack on bigger (virtual) hardware the opportunity to see what Pressflow and Varnish can do together.
The big question is what's next for Mercury and the rest of the Pantheon project. Our grand vision is to develop a full roster (a real pantheon, if you will) of images, every piece you'll need to support first-class Drupal development, testing and production environments. However, in all the conversations we've had since launching getpantheon.com and in talking to folks at/around Drupalcon Paris, it also feels like there's possibly a more pressing need for a best-practice all-in-one system.
Our goal is to show that Drupal can take advantage of cutting edge hosting methods, current development/deployment techniques, and can perform quite impressively on a well-tuned stack. Basically, we want to prove to the world that Drupal is not only the best open-source CMS, but a world-class publishing tool for anyone looking to manage their content on the web.
While there are plenty of high-end hosts that make their living building production clusters for the top of the market, the remaining 99% of the Drupal universe is currently underserved. There are many people out there running on stock, un-tuned VPSs, or even on shared services like Dreamhost or Hostmysite. Most Drupal installs in that situation probably feel somewhat slow under normal circumstances, and quickly become victims of their own success at even modest levels of load. Plus there are many of Drupal admins out there who are pushed into the position of becoming "accidental sysadmins" in order to set up and support all these sites.
This feels like a big need, so while we're going to continue developing additional images (for Solr, for dedicated database host, etc), our current priority is developing Mercury as an all-around Best Practice starter stack for hosting Drupal with Apache, PHP and MySQL. We're also hoping to work out a pilot program with some VPS providers to allow people to try out Mercury on-demand and see how it works.
There's a ways to go before this will be the well-oiled and integrated process that we envision, but things are looking exciting. With a little luck, we should have some great solutions to many common Drupal hosting headaches, and by standardizing a lot of people around a common stack configuration, we can improve the quality of documentation as well as the pace of innovation as new techniques (Edge Side Includes, anyone?) emerge for Drupal 7!