Would you like to discuss your environment with a knowledgable engineer?
Drupal is a popular open source Content Management System with a dedicated developer community providing a nearly unlimited set of features and support ; free!
Drupal has a big issue with performance. Scaling a Drupal website is a challenging, time consuming activity. The architecture is designed for ease of use and flexibility, not for high volume. When you drive heavy traffic to a Drupal sites you frequently encounter slow page loads and an inability to connect to the site.
There are endless discussions in the drupal forums about ways to make Drupal scale. The bottom line is that even with exceptional technical skill and tremendous patience; segmenting databases, deploying clusters, and optimising code, you will at best increase performance three to five times from a default install. This is often a fraction of what is required by a busy site.
There are three main issues to scaling Drupal. aiScaler handles these with simple to implement, elegant solutions. The first issue is managing the network sessions from large numbers of concurrent users. In standard architecture a thread or process is opened to support each simultaneous connection. With a few hundred users connected, passively browsing, the number of connections begins to swamp system resources. Second, the amount of time to generate pages begins to suffer dramatically with each additional request, as Drupal must go to the database, application and web server each time to re-generate content. Finally with dynamic data such as forums, Drupal must continually re-generate the same page to make sure updates are instantly reflected. A process that simply does not scale.
aiScaler solves the first issue by completely removing session management from the Drupal server. aiScaler front ends all client communications via its right-threaded non-blocking, multiplexed IO design, nerd speak to some but this architecture allows virtually unlimited numbers of connected clients.
Second, aiScaler intercepts all client requests using it flexible caching configuration and serves 99% of requests directly from RAM, without hitting the internal infrastructure. How well does this work? Please review the PDF linked at the bottom of this article. With aiScaler deployed on a normal Drupal installation we increased the number of responses handled by 620 times, from 26 requests to 16,132 per second! Yes you read that right, aiScaler scales the requests per second on Drupal 620 times!
aiScaler relies on caching to achieve this performance so you might naturally ask what about content that changes such as user generated content. This is where you use aiScaler Header Driven Cache Invalidation. aiScaler gives you a simple way to tell the aiScaler server when content has changed. You can set the Drupal server to easily signal aiScaler to update its copy of content when a user makes a change. You get all the scaling benefit without sacrificing real-time updates, the best of both worlds. All this without changing the set-up of your site. We have partnered with some of the most respected developers and ardent supporters of the Drupal community. Please contact us to make an introduction and see Drupal scaling made simple with aiScaler.
aiScaler saves a lot of money We use aiScaler at Giant Digital in front of our load balanced AWS hosted servers on Scalr. Although occasionally we do experience high loads that spawn multiple DBs or web servers for the most part aiScaler deals with 99% of the traffic that's thrown at it and this is for 20 million+ pages a month servers. -www.gigwise.com
Easiest caching server we've found We were looking for something to help keep as much load as possible off our Windows origin servers. Our website is subject to extreme spikes in traffic, and there are make requests that can be cached. We've now used aiScaler during several large events, and the CPU is hardly pressed at all. The support from aiScaler had been excellent. They were able to create a new feature for us in under 24 hours. -ScribbleLive.com