Best Practices for getting High Availability Web Hosting | 99.99% uptime
As organisations become more and more dependent on their online presence, they are dependent on the service level provided to their clients. That will include the availability of their web presence. Losing service could be disastrous for an organisation.
Managed Service Providers (“MSPs”) offering web hosting services need to make sure that they can demonstrate high availability of web hosting to give their clients comfort. It’s all very well to say 999.9999% availability’, but they also need to show that they can recover from an outage quickly and efficiently with the minimum of disruption to their clients.
High Availability and Business Continuity is the name of the game.
High Availability – The Definition
High Availability Web Hosting can be defined as the infrastructure, operational procedures and optimisations that are designed to ensure a specified level of accessibility for users to the hosting servers. The specified level is usually defined in a Service Level Agreement between the MSP and the client, set out as a percentage of total availability.
The best way to get High Availability Web Hosting – Requirements
High Availability Web Hosting needs an infrastructure with reliable power and network connectivity to allow the functional server or servers to operate.
Be aware that it is sometimes difficult to compare “high availability” claims. Take care to understand what availability measures mean in practice. As an example, 99.9% availability could mean an average of 10 minutes of downtime per week. 99.99% means 1 minute per week.
High Availability –The Environment
One simple definition of a High Availability Web Hosting environment is one with no single point of failure. Each infrastructure component is duplicated so that if it fails, there is a backup component to take over. It also means that for a total loss of service, there must be multiple failures in the infrastructure.
Resilient infrastructures include redundant power supplies and redundant network connections. Most MSPs operate datacentres with a Tier III certification, which by definition includes a high availability configuration. The objective of the MSP is to ensure that their clients are happy that power and network connectivity are provided with a very low chance of interruption.
As an example, the HostSailor High Availability Web Hosting infrastructure in Romania has been very carefully designed with availability as a key factor. The data centres are designed and equipped with the latest technology that will self-heal a network in the event of a network failure for whatever reason.
Which features should consider used for getting high availability?
To ensure nothing slips through the cracks, the highly secure and safe data centres are routinely monitored 24/7. Local and remote support technicians are always available, which means any incidents are dealt with quickly and with minimal delay.
The Romanian private data centres are powered by a dark fibre ring throughout Bucharest. This ring has currently seven points of presence, from where the data centres are securely connected to our Hostsailor servers, ensuring access unless all seven PoPs fail at the same time.
Power and network failures
It can be protected by redundant services, but equipment failure and software faults can still result in service losses. The answer is to have server-side measures to protect against equipment failure. In some high-performance environments, high-availability clusters of servers are used together with load balancing. Sometimes the cluster includes a redundant server.
Should any server in the cluster fail, its load can transfer automatically to the redundant server, or be spread among the other servers in the cluster. Without a loss of service.
In some environments such as e-commerce, downtime for routine maintenance is not an option. A high-availability cluster is the solution to this problem and allows high availability web hosting to be confidently offered by an MSP.
If you are selecting an MSP, you need to list your business requirements and measure the MSPs offerings against them.
Remember though, with the best will in the world, natural disasters will affect services levels. It is possible to guard against some external risks like power or telecoms failures, but some force majeure incidents cannot be predicted or avoided. They could be weather-related, a change in the political environment, physical attacks or even malicious or accidental human activities.
The SLA must indemnify the Web Hosting service provider against penalties for these types of outages.
High Availability Web Hosting – Risk Management
This is where business continuity planning comes into play. If a major disaster hits, say an earthquake takes out the data centre and the local infrastructure, it doesn’t matter that you were operating a High Availability Web Hosting service. It isn’t coming back soon at that location. You need a business continuity plan that brings your clients back into operation as soon as possible.
Two things are needed, a complete backup of all the client systems and data, and an alternate site to run the new service from. This can run from a full replication service to a remote site for critical applications, to off-site systems and data backups that can be reloaded to a new configuration at an alternate location.
Replication can be on-site or to an off-site location. Obviously, if there is a total loss of a data centre, off-site is significantly better than on-site.
An MSP with a multi-site capability like HostSailor can set up connectivity across multiple data centres so if the prime site fails, the processing can automatically switch to an alternate site with a minimum loss of service.
One point to note is that replication and backup are two complementary activities. Even if you have set up replication, it is still an essential activity to take regular backups of systems and data.
Finally, you can’t guard against everything. The steps above can help guard against equipment and external service failures, but users and IT staff are human and can make errors, accidental or otherwise. That is another reason for regular backups.
The preceding has highlighted some things to consider when looking at MSPs that offer High Availability Web Hosting Services. You will need to match their offerings with your budget and probably make some compromises.