How much website downtime is acceptable?

November 1, 2012 by iWeb Technologies in: Web Marketing Articles

There is no single correct answer to this question, but it’s a crucial one for any web-based service to consider. And like most business or service related questions it all comes down to risk and return.

Zero downtime

Over the long term, zero downtime is an impossibility. While it is possible to be close to zero unplanned downtime, sometimes you need to take your site offline in a controlled manner to keep it healthy in the long term. Planning for and scheduling downtime often helps to minimize unplanned downtime.

Even the organisations with the most expertise and resources, and the most to lose, can’t make zero downtime happen over the long term.

In June 2012 Google was down for about 10 minutes. According to some very approximate calculations based on Google’s 2011 revenues spread evenly over the year, this equates to around $750,000 in lost revenue. In October 2010 Yahoo! suffered an even longer downtime of about an hour. And an article by Pingdom reports that over the 2011 holiday season, only 28 of 59 big US ecommerce sites they had monitored managed 100% uptime.

The risks

The risks to your web application come in many shapes and sizes. They can be broadly grouped into these categories, and mitigated in some of the ways described below.

Risk Mitigation
Planned downtime

Upgrades to:

  • Hardware
  • Software
  • Operating systems
  • Devices
Component failures

  • Hardware
  • Software
  • Programming
  • Viruses
  • File corruption
Operator error & malicious users

  • File deletion
  • Unskilled operation
  • Hackers and malicious attacks

The price of downtime

While ten minutes means a lot of money to Google, and a lot of people heading off to competitor Bing (as Google themselves even suggested at the time), ten minutes downtime for a corporate blog is inconvenient, but not the end of the world.

Any website that can attribute a value to lost visits, whether from lost sales or advertising revenue, will fall somewhere on the scale. By understanding what you have to lose, it becomes possible to understand the expected savings (or return) on the money you spend to mitigate the risk of downtime (investment).

Working out lost revenue (or perceived value if your site is not directly commercial) is a difficult task, but should already be addressed to some extent in the justification for its existence. Lost advertising impressions might be easily evaluated, and similarly the average value of a visitor to an ecommerce website. But for a company blog, attribution is more difficult and requires assumptions and analysis of customer touchpoints and their effect on customer purchases and loyalty.

In any case, web applications can be assessed by how critical they are to your operations, and dealt with accordingly in terms of risk mitigation.

How much downtime?

Mitigating the risk of downtime will tend to require some investment in skills and technology. It may involve reducing the complexity of your systems, perhaps with the loss of certain functionality. Or it may involve introducing controls that cost time and resources. It also includes some planning for the inevitable to reduce the impact to a minimum.

By considering individual risks, how they would occur, how they would be dealt with and how long this would take, you can decide if the current situation is acceptable or if you need to take measures to reduce the impact or the risk of the problem occurring.

Often hosting and developing a web-based application involves outsourcing. When looking at service level agreements think about how percentages relate to time:

  • 99.999% uptime = 5 minutes downtime / year
  • 99.99% uptime = 53 minutes downtime / year
  • 99.9% uptime = 8 hours 45 minutes downtime / year
  • 99% uptime = 3.65 days downtime / year

Getting started

Whether or not you are able to exactly measure the cost of downtime, do consider the impact to your organisation. Make sure you are aware of risks and their impacts and explore the options you have to manage or reduce the risk of downtime.

Here are a few tips to get started:

  • Speak to your hosting provider about their network SLAs and how your hosting infrastructure is set up to minimize downtime
    Can you live with the SLAs?
    Do you need to optimise your infrastructure for availability and security? Is the price worth it to hedge against downtime?
    Do you have the necessary security measures in place?
  • Speak to your Systems Administrator and IT department about what aspects of your setup they support
    Where are the gaps in support?
    Who can provide this managed support and expertise in the event of a crisis?
  • Look at which versions of which software you are running, and consult the providers’ websites
    Are they adding value?
    Do they fit with your IT support and human resources coverage?
    Are there more up-to-date versions available?
    What is the cost (in downtime) of an update?
  • Look at your processes and the human resources involved
    Do you need to build in more controls to mitigate against human error?
    Do the users have the required skills?
    Could you roll back to a previous version or use back up data in the event of an error?
  • Consider the typical hourly and daily distribution of traffic using your application
    If you have to have a planned downtime, when will it have the least disruption?
    Will the necessary support and resources be available at this time and shortly after?

To talk about how server clustering, data backup and security planning can help you reduce planned and unplanned downtime, contact one of our specialists for a chat.

iWeb guarantees 100% network and power uptime on all our dedicated servers, VPS and cloud hosting.

Find Sam on

Comments Off tags: , ,  | 

Comments

No comments yet.