High availability (or High Availability, has) is a design of computer systems and the implementation of the Protocol to ensure a certain degree of operational continuity in a measured time.
The availability of a system refers to the ability to access the system, realize actions or collect data. If a user cannot access the system it is said that is not available.
Some people say also that when a system, despite being available, has a long response time than usual and do more difficult, although slightly, the access to information either that there is high latency or the system is overloaded, it is said that the system is not available too.
The term used to express the time that a system is not available, it doesn’t matter for the reason, is downtime.
Downtime is the time, as we have seen before, whereby a network service is not available. We can have two types of downtime, according to his origin.
- Planned downtime: is that which is done periodically (usually) to perform maintenance on the system, They can be software updates, changes in hardware or something as simple as restart the server. This time ought to be prepared in advance and tend to occur at when the system load is as low as possible to avoid interruptions the service.
- Unplanned downtime: it is that which as its name suggests, has not been prepared, with which tend to be serious problems, such as hardware failures, power supply failures or network connections failures, attacks from DDoS, etc..
Many web hosting services, used servers leased from other companies and say that they have a high availability, when they actually does not control the high availability on their system
Calculate the availability of a system
Availability is usually expressed in a percentage of time (in minutes) in a year. The calculation is like that:
AV = (525600 – Mi) / 525600 * 100
Where: “Mi” are the minutes of downtime in a year.
This result should be as closer as possible to 100%, values most common tend to be:
- 99.9%, mean 43.8 minutes a month or 8,76 hours a year of non-availability
- 99.99%, implies 4.38 minutes a month or 52.6 minutes per year of non-availability
- 99.999% implies 0.44 minutes a month or 5.26 minutes a year of non-availability.
Operating time is not equal to the availability time. Because a system may be operating but not available due to a failure in the network services.