Proxies.. My favorites subject..
A "Reverse Proxy" is typically an appliance-based device that sits between a users' browser and a web server and serves up static data they have cached in RAM or on their disks. Proxies (forward and reverse) are EXTREMELY efficient at serving up static web content. Most large websites use them as a relay tier between the actual web servers and the Internet. The proxies serve the static content and let the web servers handle the dynamic stuff and be the main "source" for any static content updates, etc.
In general, for a typical static content website (eBay is a great example with all the JPEGS and pre-built pages), the proxies handle 70+% of the content. Given that your average proxy is about 4X as efficient as a web server, it's a signficant cost savings and a big boost in performance, too.
Not having to mess with those pesky OSes and legacy file systems makes the big reverse proxy vendors pretty cool. From a security perspective, they are great - No clients actually talk to the web servers directly, they talk to the proxies, so that a lot of common vulnerabilities on the web server and OS are avoided.
We use a lot of gear from Network Appliance and BlueCoat in production and I've looked at equipment from several other vendors, as well. It's all good stuff, just dpends what you're looking for.
If you had a totally static website, you could leave it in one location, put reverse proxies in other data centers and use a global server load balancing tool like F5's 3DNS to distribute traffic to the closest available data center. Again, exactly how we do it, but we have content servers in each data center, all front-ended by reverse proxies.
- G
Facebook
Twitter