Commentary: More hardware won't solve Web site outages
People and process issues cause 80 percent of unplanned downtime, so Web sites can't solve chronic outages simply by installing more hardware and adding redundancy.
By Donna Scott, Gartner Analyst Outages of eBay's Web site attract attention because the vendor is so popular, but the problem occurs industry-wide--and not just with consumer sites.
 | |||
|
| | |
| |||
Achieving continuous availability of Web sites--and the applications that users need--requires a multipronged strategy that mitigates the risks of unplanned failures and planned maintenance or upgrades. Because continuous availability is anchored in complex IT systems, it cannot be bought off the shelf but requires substantial cross-company planning, discipline and control. With most IT projects, however, timeliness takes precedence over planning.
Moreover, achieving high levels of availability and scalability for critical Web sites requires investments in application and infrastructure design, development, testing and proactive operations management processes. Building continuous availability costs three and a half times more than building a standard application.
Consequently, despite the Internet significantly increasing the need for it, Gartner estimates that fewer than 20 percent of businesses' mission-critical Web-based applications will achieve continuous availability through 2005. Around 40 percent will achieve high availability for two and a half times the cost of a standard application, but the balance will likely not meet quality-of-service objectives. (High availability means 99 percent to 99.5 percent uptime, or 43 to 87 hours per year of unplanned downtime and four to 16 hours per month of planned downtime.)
Eliminating downtime is also difficult because outages have different causes. Gartner sorts Web site failures into the following categories:
• Infrastructure--in servers, networks and operating systems, and power failures and other disasters: 20 percent of downtime
• Applications--programming bugs, performance problems and changes to applications that cause problems: 40 percent
• Operations--e.g., not performing a required task or performing one incorrectly: 40 percent
In other words, people and process issues cause 80 percent of unplanned downtime. Therefore, Web sites such as eBay cannot solve chronic outages simply by installing more hardware and adding redundancy.
Best practices
What do the pros do? Gartner recently asked executives of two Internet companies--Digvijay Chauhan, co-founder and vice president of development at AskMe, and Andy Harmon, director of data architecture for Charles Schwab--to list their best practices for ensuring Web site availability. In order of importance, they recommended that e-businesses do the following:
• Clearly articulate business priorities so workers can make decisions with those priorities in mind.
• Do not ignore people and process issues, which cause 80 percent of end-to-end site availability problems.
• Design the entire architecture for horizontal scalability--i.e., for transparently and nondisruptively adding more capacity.
• Design the site for no single point of failure.
• Invest in business continuity planning.
• Design the site to minimize the impact of planned outages and avoid surprises by communicating unavoidable planned outages to users.
• Invest in real-time availability and performance management.
• Invest in proactive capacity planning and change simulation tools.
• Build multiple Web sites to meet varying business requirements.
• If legacy systems work, do not replace them.
(For related commentary on e-business continuity, see TechRepublic.com--free registration required.)
Entire contents, Copyright ? 2001 Gartner Group, Inc. All rights reserved. The information contained herein represents Gartner's initial commentary and analysis and has been obtained from sources believed to be reliable. Positions taken are subject to change as more information becomes available and further analysis is undertaken. Gartner disclaims all warranties as to the accuracy, completeness or adequacy of the information. Gartner shall have no liability for errors, omissions or inadequacies in the information contained herein or for interpretations thereof.