Many hosts advertise 99.99% uptime or anything above 99.9%. Is 99.9% any different from 99.99%. Does it make a difference to anyone. The difference is only 0.09%, for crying out loud!! Well, there is a difference
- A 99.99% uptime means 54.84 minutes of downtime a year.
- A 99.9% uptime means a total of 6.91 hours of downtime a year.
- A 99% uptime means 3.6 days of downtime a year!!
Web Hosting consumers need to seriously look at their Web Hosts uptime claims and compare the downtime they have faced with the figures above.
For a Web Host, does it take the same monitoring mechanism to sustain a 99.99% and a 99.90% uptime. Obviously not. But first a look at what is Uptime.
Uptime and downtime
Uptime is a measure of the time a computer system has been “up” and running. It came into use to describe the opposite of downtime, times when a system was non-operational.
Downtime refers to a period of time or a percentage of a timespan that a machine or system (usually a computer server) is offline or not functioning, usually as a result of either system failure (such as a crash) or routine maintenance. – Wikipedia
A Web Host has to actively monitor Downtime of all the services running on each Internet Servers. What services need to be monitored? An Internet server has services like Webservers, Email server, DNS, FTP, Control panel software, PHP and many other goodies. These services are assumed to be available 24/7. If a Web Host claims 99.99% uptime, all of the advertised services should be available 24/7 at 99.99% and available at promised speed.
Web Site owner and the Web Host can confirm that all these promises are being kept through Uptime Monitoring. This article deals with ways and means of measure uptime.
Uptime monitoring is basically Downtime monitoring. Some Web Hosts confuse uptime monitoring with the number of days the server was up. This is not what we aim to measure here. In uptime monitoring, Web Hosts measure the number of times the services went down, for what duration and at what speed the services are available.
Are some people assuming that a good server will never have downtime ? On the contrary all kinds of downtime are not bad. Some are required for maintenance activities, like installing new services, configuring and optimizing. Maintenance related downtimes have to be carefully scheduled to reduce the impact on customers. For example if customers are mainly located in the United States, the Web Host can decide to conduct maintenance related activities after business hours.
Downtime caused by System failures on the other hand are the ones to be checked. System failure downtime can be the result of Overloaded servers, security attacks such as DOS attacks, poorly configured or outdated software. Web Hosts should take preventive measures to ensure that these problems are plugged as much as possible. And to ensure that downtime are detected early on, Web Hosts should use Network monitoring services or software.
Network Monitoring services
A Network Monitoring service will monitor one or more services on servers that are open to the Internet. A monitoring service will have servers around the world which checks your server or website in intervals of 1 minute to 1 hour.
When you sign on to a monitoring service, you need to check the locations they have hosted their monitoring software. Typically they will have at least one or two in each continent.
The most important aspect of a monitoring service is the Monitoring Intervals. Do they check every minute, every 5 minutes or every hour? If the interval is 1 hour, a service can go down for half hour and no one would be the wiser. The service wouldn’t record it as a downtime. Obviously it makes sense to go for an Interval at least 5 minutes or lesser. The provider’s charges go higher as the intervals go shorter.
Also check the methods through which they notify you. Do you receive alerts via email, IM and SMS. Choose a service that provides you your choice of communication. If you are on the road most of the time, a SMS alert would be more useful than an email alert.
Alertra.com is a good paid service. They let you receive downtime notifications on email, Instant messenger and SMS. They also have some good free trials to check out their service. www.siteuptime.com is a good free monitoring service. For intervals less than 5 minutes, they charge nominal fees.
Monitoring with software
Its also useful to use software which you can install on another server and remotely monitor your Internet servers. There are Commercial and Free options as usual.
Nagios is the best Open source software out there in my opinion. Its highly configurable to include unlimited servers and regular and custom ports.
Another good Open source software is Zabbix. It has some excellent graphic visualization, generally available only in commercial software.
Human touch – 3rd party server human” monitoring service providers
If you have colocated servers or unmanaged servers, your Datacenter will expect you to monitor the servers yourself. Even if you have signed on a Server monitoring service or installed a Monitoring software, the best these can do is to alert you. The best that a Server monitoring service or a Monitoring software can do is to alert you. You need to work and get the server or service running back on its feet again.
What if you are on vacation. In any case you need your 8 hours of undisturbed sleep right ? There are human monitoring service providers, who will either monitor your servers themselves, or work with your monitoring providers to Take action when an alert comes in.
Here is how it typically works. The human monitoring service provider gets an alert when any service goes down. If the server is still up and only one service is down, they will login to your server and bring up the service. In case the server itself has gone down, they will call your Datacenter on their emergency numbers and ask them to reboot your server. Basically they take over the coordination work which you would otherwise do yourself. And all this happens when you are fast asleep in your bed.
Bobcares.com is a leading provider in this business. They staff their facility 24/7. These services are provided either standalone or clubbed with other technical support services. Other companies are touchsupport.com and supportresort.com
Uptime monitoring is an important way of measuring your Web Hosting operations performance. A way to know if you are keeping your promise to your customers.