How to Minimize Website Downtime

Website downtime inconveniences visitors and can be quite costly to an organization’s reputation as well as its bottom line. But, is there anything that can be done to prevent it? Are outages inevitable? There is a sense in which the eventual failure of any complex system is essentially assured. There are just too many variables. Fortunately, there are many things that can be done to reduce the risk and, if nothing else, minimize the duration and thus, limit the impact. Following, are some of the best ways to minimize website downtime.

Move to a Dedicated Server: This is probably the easiest way to improve the availability of a website. This is the “low hanging fruit” of minimizing downtime. There are millions of websites hosted on shared servers hosted by third-party hosting companies. Shared hosting is inexpensive but there is no way to control how other clients use the server resources except to let the hosting company take care of it. Some hosting companies are a lot better at it than others but even the best fail at one time or another. And when they do fail it can take days to get a hosting company to track down and address an issue such as a rogue, CPU-hogging process launched by another customer. In the meantime, every website hosted on the shared server can suffer from degraded performance or complete inaccessibility. In reality, these problems can be very hard to track down and can result in months of intermittently degraded performance.

Manage Server Resources: Once the website is running on its own dedicated server, make sure it has enough resources to perform properly. The resources to be concerned with are disk space, RAM and CPU. If the server runs out of disk space, it will likely just hang. If it runs out of RAM, the operating system will start swapping memory to and from the hard disk. The result will be seriously degraded performance. If the server needs to process a very large number of simultaneous requests, CPU power may be an issue. The goal of server optimization is to design processes and allocate resources in such a way that the system overall is so efficient that CPU becomes the bottleneck. Then, if performance is not acceptable, a CPU upgrade becomes the easy fix.

Maintain Software Upgrades: Since most web servers must necessarily face the public Internet, they are vulnerable to all manner of attack. The most common scenario is unmaintained web server software becomes out of date and is infected by an automated worm. The activity of the worm degrades performance or brings the server down altogether (and server performance is usually the least of the damage inflicted by such a worm). Keeping software upgrades current for all public-facing servers is a must.

Require Best Practices for Web Development: If the web application hosted by the server is not designed correctly, the door may be left wide open for attacks that can bring the server down or worse. An automated bot can submit forms so quickly that a database can fill the hard drive or the server may run out of CPU resources. A SQL injection attack could delete critical data causing the application to crash.

Insure Sufficient Bandwidth: Bandwidth can be expensive, but it won’t matter if the server is finely tuned if there isn’t enough bandwidth to process all the concurrent requests. Look for ways to conserve bandwidth such as compressing content and using efficient formats for media.

Use a Website Monitoring Service: Even if every area is addressed, there are still going to be unforeseen circumstances that will result in downtime. A log file can fill up a hard drive. A component can fail. Unexpected publicity can cause an extended spike in traffic consuming all available resources. When one of these things occurs, the key to minimizing downtime is finding out about the problem quickly. A quality website monitoring service can test the server’s Internet connection, web server software, front-end application and back-end resources for proper functioning once every minute, 24 hours a day. The best monitoring services use multiple monitoring locations and will notify the system administrator about a problem using e-mail, SMS text or phone call alerts. The sooner the administrator knows about a problem, the sooner the issue can be resolved and proper functioning of the website can be restored.