Regardless of scale, early adopters of web services agree on one key issue: if you don't plan your services management strategy at the outset, you're heading for trouble.
| ||express delivery|
| print comment|
|To stay on top of diversity, change and unpredictability in web services deployments, early adopters are prioritizing services management:|
- In many organizations, web services will emerge organically
- Real-time services require proactive management
- Increased flexibility inevitably brings greater complexity
- Constant evolution demands strong change management
- Specialist tools provide much-needed capabilities
Glossary terms: services management, web services, development, lookup tool
"When you hit critical mass [with web services], your organization is really going to change," says Mike Reagin, director of R&D at Providence Health System. "If you don't have the infrastructure to manage it upfront, you'll be in a position where you can't effectively manage the enterprise."
Providence, which prioritized management issues when it launched its very first initiative, is one of a broad selection of organizations interviewed about their web services management strategy for this article. Customers range from the likes of AgentWare, a travel industry solution provider that's built its entire offering on web services, to healthcare provider Providence, to blue-chip organizations British American Tobacco (BAT) and British Telecommunications (BT), who provide an additional global perspective.
In every case, the warnings are very similar, even though there are subtle distinctions in the reasons they cite for prioritizing services management. Whether the motivation is to manage diversity, unpredictability or change, the message is the same: be prepared, or you'll face mounting problems.
BAT, a multinational operating in some 160 countries, committed to an SOA architecture strategy at the beginning of 2002, and is in the early stages of implementing it through a number of proof-of-concept projects. As it set its pilot schemes rolling, it made an early commitment to web services management, and subsequently selected AmberPoint for operational management and Infravio for lifecycle management within those projects. Application technology manager Kevin Poulter, who's in charge of defining global strategy for application infrastructure, says the company has emphasized management so that it can stay on top of the likely diversity and unpredictability of web services adoption.
"The core web services technologies creation, brokering and orchestration are going to become commodities," he explains. "They're going to be things we as a strategy group, in a federated organization using systems from a number of vendors, can't reasonably control." For example, SAP, Siebel and IBM Websphere Portal Server, while conforming to standards, will each adopt different implementations of these core capabilities; and BAT's user divisions will then make their own business-case decisions on when and how to deploy those implementations.
"We see [web services adoption] predominantly emerging in an organic way," says Poulter, "which means we can't predict where these services are going to emerge, and therefore we don't know where we're going to need to deploy the management software."
For such a geographically-diverse company, the deployment issue was a key consideration when it first approached AmberPoint eight to nine months ago. Though not the only factor, it was important that AmberPoint's product can manage distributed web services without first having to deploy code locally to each individual endpoint. "You can if you want to for performance reasons, but for us it isn't really practical to go into some mass roll-out of the software in so many geographies," says Poulter.
For AgentWare, another AmberPoint customer, the drivers for adopting management capabilities have been very different, but the fundamental message remains the same. The Atlanta-based company has built a reservation service that allows travel agents and websites to bypass the four central booking systems used by the majority of the industry, in the process offering significant discounts on reservation costs for booking flights, hotels, car rental and the like. According to CEO Les Ottolenghi, in fourteen months the company has gone from 0.1 percent market penetration to a position where 70 percent of travel agents now use its technology in some capacity.
That rapid rate of growth meant the company required a fast and flexible IT infrastructure. It decided to build that infrastructure entirely from web services, using its own custom tools, and, while management capability was only installed a few months ago, it designed its core architecture from the outset with that requirement in mind. "We considered building our own, but it didn't make sense because there were already people like AmberPoint out there, who've built industrial-strength systems," says Ottolenghi.
Web services management is mission-critical for AgentWare's core business. The company doesn't create or store data; it gathers it in real-time from supplier systems over which, of course, it has no direct control whatsoever. This means it has to manage a range of potential IT problems, from supplier system downtime to spikes in demand generated by high response to special offers, many of which are unpredictable. Its service level agreements are based on a combination of uptime and the number of transactions that can be handled concurrently, so IT efficiency translates directly into business effectiveness. In addition, it has to handle the change impact of business growth: "Each time you put in additional functionality, you have to manage it across several different suppliers."
For its part, BT recognizes these kinds of management challenges from both a supplier and user perspective. Its Global Services division recently launched its own hosted management service, partnering with Flamenco Networks and AmberPoint, based on the company's own experiences of rolling out web services internally in its core UK telecoms business.
As Simon Walker, web services business development manager at BT, points out: "You start off to create flexibility, which is a good thing, but you get to a point where it becomes unmanageable." He adds: "Once you've released a web service to the big wide world, people will start using them in ways you did expect and in ways you didn't expect. Customers have realized management is where all their budget's going."
Diversity is an important factor in management requirements, says Julian Hill, web services CTO for BT Global Services: "It's highly dependent on the organization and the way things are deployed. The more federated your organization, the more this is likely to be a problem."
However, as the number of web services increases, complexity will increase regardless of the type of organization, and that brings significant challenges. In particular, BT points to the difficulty of:
- Implementing common standards especially while they're still evolving
- Providing and revoking services
- Guaranteeing delivery
BT's betting on the fact that rather than tackling these issues in house, users will be tempted to look for a service provider capable of absorbing the strain.
For Reagin at Providence, all these factors were relevant pretty much from day one. But he still rates the effort worth it. The organization, which operates out of Alaska and the West Coast of the US, embarked on a web services project to tie together its different legacy systems and provide a portal for patients, an approach that cut development time by 40 per cent and allowed it to complete in six weeks.
Reagin reports a "dramatic increase in productivity and efficiency with developers", enjoying faster development times and high reuse of services. Providence has committed to web services for its future development efforts. Its next plan is to build secure messaging links to its lab systems, so that physicians can distribute test results more efficiently to patients and reduce telephone interactions.
From a management perspective, Providence opted for Infravio, which Reagin describes as "technically elegant". While security remains a challenge the slow evolution of standards means that the company has had to wrap proprietary security around its critical web services the management system has made a major difference in versioning and build. Change management has been one of the most significant issues: "Since web services are highly customizable, we're changing and adding features all the time it's important to know what application connects to what service."
Providence's first impressions about the importance of management have been fully borne out by its experience, says Reagin. "Early on, when we started to look at deployment of these services, we thought, what will happen when we have fifty, a hundred? We think there will be a point where we hit critical mass it will be too difficult using [traditional] tools. I think it's a key decision before people enter into this space."
Additional reporting by Phil Wainewright.
More on this topic
Managing web services requires tools and techniques ...
HP, IBM and CA have ambitious plans for adding web services capabilities ...
BT is not only a supplier of web services, but also an early adopter ...