Survivor's Guide to 2006: Enterprise Apps and App Infrastructure

All for OneThe good news--or bad, if you're an isolationist IT shop--is that the enterprise application initiatives you'll be undertaking will be orchestrated by the business, so you'll be working even more closely with your line-of-business counterparts to implement application services. And your organization's application infrastructure must consider business needs more seriously than ever before and be able to adjust rapidly to ever-fluctuating market conditions.

Regardless of the application technology, process and business will reign supreme. The enterprise application technologies you'll be investigating will support the former and be driven by the needs of the latter.

Enterprise Applications

The need for applications to control business processes can be found at every layer of the application infrastructure stack. Enterprises are integrating call centers with order-management applications and manufacturing systems and tying them together with BPM (business process management).From integration to routing to management and monitoring, the goal remains the same--optimization and control of business processes in an increasingly dynamic business environment. Process flexibility will remain the business driver for BPM. Before you know it, you'll be designing your enterprise SOA to support managing, optimizing and growing IT-enabled business processes. And you won't be alone. Fifty-one percent of firms will be using SOA by the end of 2005, Forrester predicts.

CRM and ERP

CRM may be dead, but CRM on demand is a booming business. After many large enterprises realized disappointing ROIs (returns on investments) on their huge CRM implementations in the 1990s, they turned to CRM on demand. Oracle's acquisition of Siebel and PeopleSoft may appear to have narrowed the field of products, but salesforce.com and Salesnet and a bevy of open-source-backed CRM on-demand providers are out there and will be vying for your business in 2006.

The wave of acquisitions in the ERP (enterprise resource planning) field also will make 2006 a year to reconsider your ERP implementation. Most organizations with an extensive ERP have never thought about replacing it, but that was before their vendors were snatched up by Oracle and other bigger fish. Bottom line: The ERP market is in flux.

The RepositoryOver the past year, meanwhile, you've upgraded packaged applications, implemented new applications and deployed those developed by your IT staff. Most of those applications, if not all, have a Web services interface or two that facilitate communications with other applications. But these services aren't managed and, in some cases, you might not even know they're ready for access.

You need a repository--one that will help you gain control of rogue services and enable the reuse of services throughout the enterprise. Without a repository for your SOA infrastructure, you cannot catalog or reuse your services. Yes, I'm talking about the UDDI registry. This incarnation of the registry has more bite than a double-shot of hard whiskey. It's geared toward governance, being able to control and manage available services and, ideally, enabling run-time binding that eventually results in business-user-developed, composite applications running on an enterprise SOA.

But reaching this SOA nirvana won't happen in 2006. The major goal of a repository implementation in 2006 will be simply to get a handle on services and manage them from a business viewpoint (what business function does it provide?) and for IT-specific information (what version is it, who can use it?).

The repository also will see wider use of BPM implementations as processes increasingly are defined using service-oriented products from Infravaio, SOA Software, Systinet and others. These vendors will continue to forge relationships with ESP (enterprise solutions platform) vendors like BEA, IBM and Oracle. But Systinet and other service-oriented vendors will get pulled into the enterprise mostly because of their repository products, which, like your SOA, continue to mature and become more useful. More ambitious, composite-application initiatives will emerge in 2007 and beyond. Without the repository, your future SOA won't engender reuse or let services align with business needs.

With all these application services sitting around on the network, SOA management will continue to be an important part of your SOA strategy, especially in 2006 as you begin to seriously construct your enterprise SOA. Both operational and functional management will need attention in 2006, internally as well as externally with partners and customers. The past two years have seen a great deal of consolidation in the SOA management area with the Actional-Westbridge merger, Oracle's purchase of Oblix (Confluent) and IBM's acquisition of DataPower, but there's still plenty of product to go around.

Operational (invocation counts, response times) and functional (routing, transformation, transactional logging) management lines are still blurred, with CA and Hewlett-Packard standing on the operational-only side of the road while Actional, Reactivity and SOA Software hang out on the functional side. AmberPoint and DataPower drive down the middle, supporting WSDM (Web Services Distributed Management) for operational management and WS-Policy for functional management.

You'll need to manage both in the coming year, so start to consider whether a single solution can fill your needs for both functional and operational management, or if you'll need to buy point products to get the job done.

Application InfrastructureRegardless of the technology used, applications and data must communicate, and that communication is growing to include a wide variety of technologies, such as SOAP (Simple Object Access Protocol), JMS (Java Message Service), FTP and custom APIs.

Communication at the application layer has always been accomplished through some sort of integration technology, and that won't change in 2006. SOA is your focus, and it's all about easing integration. Building out your SOA will require a huge change to your application infrastructure. SOA isn't a point product or even a suite of products--it requires multiple components comprising multiple products up and down the infrastructure stack and includes messaging (ESB) technology, application servers (ESP), management (SOA Management and Registries) products, development tools and more.

Middleware-in-a-box will become a viable building block in your SOA initiative. Products from IBM through its acquisition of DataPower and appliance products from Cisco's AON line, for example, will blur the boundaries between conventional middleware implementations and shiny new SOA blueprints. Cisco will offer more integrated middleware solutions in 2006, including specific AON technology for FIX (Financial Information Exchange) in addition to its new RFID (radio frequency ID) AON solution. Now that both IBM and Cisco are on board with middleware-in-a-box, it will be difficult to ignore the advantages of these offerings, such as the performance, consolidation and simplification of the architecture.

Your SOA will change next year, too. You may find hundreds, if not thousands, of existing services within your demesne that could be reused but are not. So your "accidental" SOA will evolve into purpose-built SOAs with the blessing (and possible assistance) of the business.To make your SOA a success, you must control these services. Many of your application projects in the next year will be geared toward governance of your organization's myriad business services--which look suspiciously like Web services (probably because beneath the covers, most are).

Many are watching SOA with suspicion. After all, we've heard that this type of architecture would save the world before--why would SOA succeed where others, such as CORBA (Common Object Request Broker Architecture), have failed? It's all about the "B" for business. The movers and shakers of the SOA world know that it's the bottom of the ninth, there are two outs, and the count is full in this architectural game. Because SOA focuses much attention on business process, with a language that is moving to accommodate the business analyst as well as the IT architect, there's a lot more synergy between the business and IT, and a lot more common ground.

That's something we haven't seen before--a single language that makes sense to tech and non-tech alike. Having learned from the failures of CORBA and RMI, SOA is better positioned to save the game than the technology that came before it.

Conventional integration technologies are evolving and are quickly being replaced with ESB software. ESBs provide integration plus the capabilities of legacy EAI products, and features and functions that integrate both legacy and contemporary services, such as Web services. And ESBs are gaining momentum in the enterprise. In fact, more than half of all large organizations will have an ESB running by year-end 2006, according to Gartner.ESB is heavily focused on message-oriented processing, but adds the complexity of events to the mix. Unlike legacy EAI solutions, ESBs generally transport messages on the bus in a neutral XML format and then convert messages to the format (JMS, EDI, MQ) used by the endpoints receiving the message. Events are nothing new, but they become more business-oriented when combined with business processes and will require even more input from line of business counterparts than in the past.

ESB is considered the underpinning of a well-designed SOA, so if you haven't yet implemented it, then you're likely to in 2006. Gartner says that ESBs will "supersede several types of traditional middleware during 2005 through 2007 because ESBs are better-suited to modern application styles, such as service-oriented architecture (SOA) and event-driven architecture (EDA)."

To support an enterprise SOA, your ESB must support myriad data sources and points of entry, so make sure you coordinate across business units to determine what features it should support. Carefully consider the architectures of different ESB products such as those from Cape Clear Software, Fiorano Software, IBM, Iona Technologies, Oracle, PolarLake and Sonic Software, to determine which is best for your performance and deployment needs.

Bye-Bye, EII

You'll either replace your database or at least massively upgrade it. 2006 promises to be a year full of interesting database activity as IBM, Microsoft, Oracle and Sybase all release new versions of their respective database platforms.Databases aren't just for storage anymore, either. The latest versions of Oracle's Oracle10g and Microsoft's SQL Server 2005 come with built-in features that support the business intelligence and data warehousing needs of business. Microsoft has finally added support for its CLR (Common Language Runtime) languages inside the database to compete with the in-database Java language support of its rivals. And now that these databases come standard with direct XML access/generation to data, it's a fact that databases aren't standalone containers of data anymore, but full-blown platforms that handle data storage, manipulation and application-level logic.

Most databases are subsuming EII (enterprise information integration) capabilities to support business-intelligence and data-warehousing initiatives. Although many BI vendors have snatched up EII vendors so they can provide this same functionality, Microsoft and others are building these capabilities into their databases, which removes the extra layer of integration needed to deploy BI and data-warehousing products.

Lori MacVittie is a Network Computing senior technology editor working in our Green Bay, Wis., labs. She has been a software developer, a network administrator, and a member of the technical architecture team for a global transportation and logistics organization. Write to her at [email protected].

Exam

An ESB is: 1. An integration technology founded on SOA and Web services
2. An integration technology that will fail like all the others
3. The bus that takes us all to the bar after work for 'team building'
4. A song from some 60's band

(correct answer is #1)

SOA stands for: 1. Service-Oriented Architecture
2. Service-Oriented Applications
3. Splendidly Obvious Avarice
4. I can't keep up with all these TLAs

(correct answer is #1)UDDI is the standard supporting implementations of: 1. Repositories
2. Suppositories
3. Depositories
4. Could you stop with the acronyms already??

(correct answer is #1)

Governance is the act of: 1. Cataloging and managing Web services and their meta-data
2. Running a government
3. Being in charge
4. Sounds painful, whatever it is!

(correct answer is #1)

4 correct answers:
You are obviously more capable than you appear. A promotion to a C-level position can't be far off now.3 correct answers:
You've kept up on the industry and you know some stuff, but you should read more Network Computing in the future to ensure that next promotion.

2 correct answers:
You're woefully unprepared for 2006. Have you considered another line of work?

1 correct answer:
Are you sure you even work in IT???

0 correct answers:
Dude, whoever read this quiz to you was obviously no friend of yours. Get some new friends, okay?

>We predicted BPM would be big in 2005 and we were right on. BPM continues to see growth in the enterprise, and it will keep making inroads with the adoption of SOAs (service-oriented architectures) and ESBs (Enterprise Service Busses) in 2006.

On the other hand, BAM (Business Activity Monitoring) was hyped big by vendors but didn't materialize as strongly in 2005 as we had expected. Although BAM is an integral component to BPM, it isn't coming to fruition as a standalone option.

Governance has been a fluid term over the past year, and it's finally settling down in a steady relationship with Web services and SOA. Governance as an initiative didn't make it as big as we thought it would in 2005, but it did manage to morph toward being an integral component of SOA.

Based on our recent reader poll on SCM (supply chain management), 34 percent of you think this is a huge deal, but SCM is too all-encompassing to become a huge IT concern. The components of SCM continued to gain mindshare in 2005, but as an overall strategic IT issue, well, we just didn't get all the way there.

Compliance was, as expected, a huge driver for a host of technologies in 2005--e-mail, instant messaging and process documentation continued to be big on the list of technologies the enterprise focused on in 2005. Logging and auditing, especially in the e-mail and database-transaction arenas, were huge, too, and we expect that trend to continue through 2006.We told you SOA was going to infiltrate the enterprise this year, and it continued to gain support and implementations in many organizations. SOA left the realm of bleeding edge technology and entered the mainstream in 2005, and it will continues to gain acceptance in 2006.