My application is not the next Twitter so why should I care about high availability?
| It often seems that load balancing and high availability are associated with only high traffic sites, like Twitter and Google. But load balancing and high availability isn't just for Web 2.0 phenomenons or web monsters; it can be an invaluable tool in your strategy to maintain service level agreements and customer satisfaction no matter how large or small your customer base - and data center - might be. Load balancing is integral to scalability, to being able to increase the capacity of your web and application servers. But it also just as inexorably linked to high availability through its ability to provide fail-over. Fail-over ensures that if, for any reason, one server in a pool/farm is unavailable that requests are redirected to a secondary or stand-by server. This ensures the site or application is available at all times. More typically, all servers in a pool are utilized at all times to improve performance and to maintain availability in the event that one or more server becomes unavailable. This is true whether you have two servers or two-hundred servers in your pool; whether you're Twitter or Bob's Widget Shop. | Where's F5?
|
High availability can be used in multiple scenarios to provide for continued availability regardless of size and reach of your site or application.
MAINTENANCE WINDOWS
Everybody has them, and they often result in downtime that, while understandable, may frustrate customers or users, especially if it's unscheduled. Patching, upgrades, migrations, hardware changes - these can all lead to necessary downtime. By implementing a high availability strategy through load balancing, you can ensure that applications remain available. This is accomplished by performing whatever tasks are necessary on one server, allowing the second (or more) to continue to serve requests. Because the load balancer (or application delivery controller) mediates between clients and your servers, customers see no interruption of service while you are working on any one one the servers in the pool.
JUST IN CASE
Unscheduled downtime is a nice way of saying "Things happen". And when those things happen that cause a server to fail - hardware, infections, licensing issues, bugs - it's nice to know that your application or site, and thus your overall availability, will not likely be affected. Unanticipated downtime can destroy your overall availability rating and cause users to go into a tizzy. A high availability deployment will prevent "things" from taking down your entire site or application, giving you time to focus on the problem at hand and solve it without fielding calls from any number of interested and angry constituents.
WIGGLE ROOM
Sometimes you develop an application and you're pretty sure you can serve the needs of your customers just fine. And then something happens and you're the next best thing on the web since Google. Maybe you got slashdotted or farked, or maybe you're the last retailer left in the country that's selling The Hottest Christmas Toy this year. Whatever the reason, a sudden spike in volume of users can leave your servers smoking.
Implementing a high availability infrastructure can ensure that even if you don't always need that second (or one hundred and twenty-second) server that in the event you do need it, it's there and immediately usable. There's nothing special you need to do, it just picks up the extra load by virtue of being part of the pool. And if you need even more wiggle room, you can add another, and another, and another server transparently. There's no need to interrupt service, just tell your load balancer or application delivery controller that the server is available and it immediately becomes part of the pool, serving up your application to hungry users.
High availability isn't just for huge sites and applications, it's a good strategy for anyone who delivers applications via the web. If your business might suffer from downtime, then you need to consider implementing a high-availability strategy sooner rather than later.
