Cover of masterclass

Masterclass for Web Managers, incl. editable templates ($29.99)

The most important measures in this regard are:

  • Availability. This means the percentage of time that the website is up and running.
  • Reliability. This means the number of unplanned outages that occur on the website.
  • Responsiveness. This means the efficacy with which the website responds to traffic.

It is often the case that a Service Level Agreement (SLA) will stipulate exact metrics for each of these. An SLA is a contract that identifies the level of service that can be expected from a Host provider. Typical measures in this regard include:

  • Website Availability. This is the percentage of time that a website is up and running. Many hosting companies now offer targets of between 99.8% and 100%.
  • Website Reliability. This equates to the number of unplanned outages that occur on the website, i.e. many outages = poor reliability. In many SLAs, this may be set at zero, meaning the hosting company certifies there will be no interruptions to service.
  • Website Responsiveness. This is the speed with which a website responds to traffic. This is often set in milliseconds, though it is dependent on factors such as the specification of the website hardware and the speed of the link to the internet.

While it the responsibility of a Website Maintenance Co-ordinator to ensure the SLA is being adhered to, the creation of a team with the right technical expertise is also important. The skills to be represented on this group include:

  • Server Software Management : This deals with the maintenance of the software used to host a site.
  • Server Hardware Management : This deals with the maintenance of hosting hardware.
  • Information Systems Security : This deals with issues of website security.

The technology of performance monitoring

In order to monitor the actual performance of a hosting infrastructure, a range of technology is now available. Products of this type can be configured to present results about availability, reliability and responsiveness - often in a graphical format. For example, among the analysis they can provide is:

  • Levels of processor loads
  • Percentage hard disk utilisation
  • Percentage of bandwidth utilisation
  • Historical traffic patterns
  • Server response times
  • Server error rates and types
  • Average queue sizes
  • Quantity of timeouts
  • Typical end-user response times

At the low-end of the market some well known vendors include:

  • BMC Software®
  • Mercury®
  • Computer Associates®

Higher-end products provide a much broader suite of services (including diagnostics on infrastructure problems). Some suppliers here are:

  • Coradiant®
  • Keynote Systems®
  • Premitech

Infrastructure Performance Review Meeting

To track performance overtime, it is suggested that every month a tally of site availability, reliability and responsiveness be reviewed at a Technical Review Meeting. At this meeting figures can be compared against the SLA. Where any failure has occurred, the issues can be analysed and a resolution agreed.

This meeting can also serve as a forum for discussing other planned technology changes. For example, a plans to deliver new services from a site could require a review of the SLA, e.g. a shopping cart. The impact of this can then be assessed by interested parties.

The process of Infrastructure Monitoring is explained in more detail in 'The Website Manager's Handbook', now on sale.