There are many business pressures today that increase the need for 24x7 access to business applications and data. Examples include worldwide operations with inherent time zone differences, mobile computing, competitive pressure related to customer service (e.g., extended online hours for claims or order entry, electronic commerce applications and extended hours for voice-response-unit access to data) and the explosion of the Internet, which has no time boundaries. Downtime of key business applications (including legacy, intranet, Web, extranet and electronic commerce) can result in significant revenue, profit and productivity losses. Historically, enterprises have paid significant premiums for hardware and system software containing a high degree of engineering to achieve high availability. The good news is that those premiums continue to decline. At the same time, however, distributed and Web-based applications are rolling out across the enterprise, making it all the more difficult to achieve application availability given the increased complexity and number of components. This presentation focuses on the technologies that enterprises can implement to achieve high availability, as well as the strategies to achieve it.

Key Issue: How will enterprises invest in technologies, people and business processes that enable high availability and continuous operations?
Each business process is dependent on the availability of critical resources for effective operation. These include skilled employees, phones, electricity, capital assets (e.g., workstations, copiers, machinery), facilities and information technology (e.g., systems, networks, applications). Lack of availability or performance of any critical resource can cripple the business process, impacting revenue, customer and business partner relationships, productivity and reputation. From an IT perspective, the greater the dependence on IT in the business process, the greater the cost of downtime and the greater the need for high levels of application availability. Availability, however, is not something that can be "bought off the shelf." Rather, it must be built into application architectures and planned for.

Implementing high availability is analogous to buying an insurance policy. One pays an increased price to gain high availability and reduce the downtime exposure or impact on the company. Sometimes, the decision to invest in high availability is clearly a "no brainer," but in other cases a company may be willing to assume the risks and costs of downtime.

Not every application requires near-100 percent availability levels. Which applications require investments in ensuring high levels of availability? Those that are critical to the business process, especially where revenues, customer support and relationships are at stake. In the not too recent past, high availability was very expensive and investments were made only for the enterprise's OLTP systems. Nowadays, however, a new breed of distributed applications has become business-critical - including E-mail, ERP, customer care, electronic commerce, and extranet applications. In addition, with the implementation of these new applications into re-engineered business processes comes a significant increase in the user population, as well as increased integration between business processes, and an increased need for high levels of application availability.

Action Item: Identify the critical business processes in your enterprise, and classify the applications within those processes as to their impact to the business process. Develop a policy that guides developers and operations on availability requirements for each application classification.

In most enterprises, IT is increasingly integrated into business processes. Manual work-arounds are no longer practical or available when critical applications become unavailable. Also, today's applications extend outside the enterprise to customers, suppliers and business partners, thus escalating the impact of downtime. For many enterprises, a loss of critical applications is as serious a loss as that of telephones or electricity. Yet, unlike telephones or electricity, critical applications are not a commodity. High availability cannot be acquired out-of-the-box; it must be built into the application architecture and preserved by implementing effective IT processes. To determine appropriate investments, enterprises should first understand the consequences of downtime. This will aid in justifying investments both for day-to-day operational availability and for business continuity (BC). Today, revenue loss, followed by productivity loss, is the primary measure being used to justify high availability investments. By year-end 2000, 80 percent of Type A enterprises and 40 percent of Type B enterprises will include intangible costs, such as those resulting from damage to the enterprise's reputation, as part of high availability and BC investment justification (0.8 probability).

Action Item: Business units and IT should jointly take steps to develop a high availability and BC strategy for critical business processes. A key step in that process is understanding downtime costs for critical business processes.

The complexity of today's distributed applications - with a greatly increased number of application, server, operating system, database and network components - makes it more difficult to achieve high levels of availability. To illustrate the point, if each of 10 components in a client/server application has a 99 percent availability level, the product of each of the availability levels would yield the overall system availability. So 99 percent times 99 percent 10 times gives an availability level of just 90 percent - or an outage, on average, of two-and-a-half hours per day in a 24-hour day. For most enterprises, that would be unacceptable for virtually any application. Achieving high levels of availability for distributed applications is a nontrivial task and requires careful planning, rigorous configuration control and diligent change management procedures.

Action Items: Invest in staffing an organization that is responsible for availability. Because of the complexity of availability management, the team should consist of a cross-functional representation of the IS organization and the business units, including the business user, networking, operations and applications development.

Through 2003, when acquiring complex packaged applications, budget 20 percent of the packaged software acquisition cost in NSM tools.

Key Issue: What technologies and products will be critical to achieve high availability and continuous operations?
There are many technologies available today that enable enterprises to achieve higher levels of application availability. The above chart depicts these technologies; the arrow indicates that implementation up the chain enables higher levels of availability. The goal in availability is predictive, self-healing systems whereby all failures are predicted and corrected before they occur, thus bringing availability levels to 100 percent. Although self-healing systems are an ideal, striving to achieve them will enable dramatic improvements in availability.

Starting at the lowest level, or the single traditional system, enterprises should look for features such as ECC memory, high MTBFs, automatic reconfiguration, remote maintenance and dynamic OS kernel updates. Moving up the line, configuring systems with redundant (mirrored or RAID) disks protects data availability, while additional redundant-hardware components further increase availability. Clusters provide redundancy at the system level. The use of middleware or multitier architectures can mask further effects of application downtime. Emerging technologies in NSM, in the areas of application management, begin to take a predictive approach to availability management to detect and correct failures before they affect application availability. All of these technologies are enablers of application availability. However, none is a substitute for a poorly designed application. High-availability application must be designed as such, and include appropriate instrumentation and built-in recoverability.

GartnerGroup's Conference Presentation SYM824x7App1098DScott.

High Availability — A Road With Many Forks

Achieving 24x7 Application Availability

Achieving High Availability for Critical Distributed Applications

Platform Availability Data: Can You Spare a Minute?

High Availability Q&A: Failures, Standards and Metrics

Feedback Form

Request Additional Copies

In Touch is published by IBM Global Services, Product Support Services U.S. (PSS U.S.). Additional editorial material supplied by Gartner Group, Inc. © 1998. Editorial supplied by IBM Global Services PSS U.S. is independent of GartnerGroup analysis and in no way should this information be construed as a GartnerGroup endorsement of IBM Global Services' products and services. Entire contents © 1998 by Gartner Group, Inc. All rights reserved. Reproduction of this publication in any form without prior written permission is forbidden. The information contained herein has been obtained from sources believed to be reliable. GartnerGroup disclaims all warranties as to the accuracy, completeness or adequacy of such information. GartnerGroup shall have no liability for errors, omissions or inadequacies in the information contained herein or for interpretations thereof. The reader assumes sole responsibility for the selection of these materials to achieve its intended results. The opinions expressed herein are subject to change without notice.