Enterprise mash-ups

Empowering the business’ is a carrot that is frequently dangled in front of IT buyers by overzealous vendors.

IT departments are told that, unlike all technologies before it, the current great white hope will help them become the business’s friend, rather than be viewed as the department whose entrenchment, complexity and inflexibility forces it to say ‘no’ to most requests for new applications.

Enterprise ‘mash-up’ technology has joined the list of classic examples of that. The emergence on the consumer web of simple, lightweight applications or data services, made by combining or mashing up two or more existing web services at the level of the user interface quickly lead to some bold claims from the industry about their potential impact on IT.

“End users who used to be wholly dependent on IT will start relying on enterprise mash-up platforms in a bid to be more self-sufficient in their application development needs,” Vishwanath Venugopalan, an enterprise software analyst for The 451 Group wrote in March 2008, expressing the consensus of the time.

This promise was appealing to both IT and the business. For the latter, it suggested autonomy from the IT department, and for the former it pointed to a day when the tortuous data integration involved in most application development projects would become a thing of the past.

But just nine months down the line that promise is already looking overly optimistic. Most vendors that sell technologies related to enterprise mash-ups agree that, but for the simplest conceivable examples, the phenomenon of business users creating their own mash-ups is a distant reality, if even a reality at all.

“Business users creating mash-up applications? That’s marketing,” says Eric Guilloteau, CEO of user interface integration provider Corizon.

And that’s not the only bad news for enterprise mash-up advocates. While ‘Enterprise Mash-Ups and Composite Applications’ took the number 6 slot in Gartner’s top ten strategic technologies for 2008, as that year approaches its end, real-world examples are nigh on invisible.

Dan Matthews is CTO of manufacturing applications vendor IFS, and he is enthusiastic about the potential of mash-ups in his field. So far, however, “we haven’t seen any mash-ups in the enterprise setting,” he says.

To borrow another of Gartner’s analytical devices, here is a technology hurtling towards the trough of disillusionment. So what went wrong, and how will the business use of mash-ups really pan out?

Mish mash

One possible reason for the lack of take up may be the air of confusion surrounding both the definition of mash-ups, and around how they should be created.

For starters, ‘mash-up’ is a vague term, covering a range of quite different prospects. John Musser is the blogger behind the website ProgrammableWeb, the bible of web-based mash-ups. He distinguishes three main types of mash-up.

“First, there are data mash-ups that combine one source of data with another. Then there are presentation mash-ups, which create new ways of looking at data.”

“Thirdly, there are business processes or logic mash-ups, which have some functionality and allow

you to execute an action,” Musser explains. “These are the most complex, and today, by far the rarest.”

Adding to the confusion, there are no established mash-up creation tools or platforms in which to execute mash-ups. The standards of data transfer may be open and accessible, but they are also numerous – although most mash-ups utilise either the simple object access protocol (SOAP) standard, or the simpler, more popular Representational State Transfer (REST) style of transferring information across web services).

“There is no standard way to build mash-ups, and there probably never will be. It can be daunting and confusing,” admits Musser. “But that’s the downside to choice and flexibility”.

Another factor that has limited the proliferation of enterprise mash-ups is the fact that to date, they have been closely linked to another technology that was supposed to liberate the business from the constraints of traditional application development: service-oriented architecture (SOA).

The adoption of SOA was, by now, meant to be triggering a plethora of web services, ready to be mashed-up at will. One problem: SOA adoption has been not nearly as rapid as expected.

“The momentum behind SOA is not as strong as it was,” says Corizon’s Guilloteau. “There was a lot of expectation that SOA would be more popular than it is. There was a lot of talk, but adoption is about a year behind.”

Indeed recent research from analyst company Gartner, showed that the number of businesses planning to implement service-oriented architecture (SOA) projects has halved in the past 12 months. Although just over half of the organisations Gartner quizzed in its fifth annual SOA survey are continuing to deploy SOA (53%), the number that were planning to deploy SOA for the first time dropped from 54% in 2007 to 24% in 2008. The number of organisations with no plans to deploy SOA grew from 6% in 2007 to 16% in 2008.

The most common reasons given for not pursuing SOA were the poor availability of staff with the required technical skills and the difficulty of building a viable business case for SOA.

“IT organisations are having trouble selling infrastructure projects,” says Rourke McNamara, product marketing director at middleware vendor Tibco. Mash-ups may help to remedy that, he adds. “If they talk about SOA projects not in terms of infrastructure but enabling rapid development of cool new applications, it’s a much easier sell.”

That may be the case, but to this point the wholesale SOA revolution has just not happened. That means the chain of events that was supposed to lead to quick and easy mash-ups has come off the rails.

“I’d say that only between 10% and 15% of large organisations in the UK are in a position to mash-up their information services,” explains Brendan Tutt, portal and social networking business leader at IBM. No wonder mash-ups are so few and far between, then.

Freedom of integration

To date, enterprise mash-ups have been closely linked to the narrative of SOA. But in truth, the failure of most enterprise IT departments to successfully implement an SOA may prove to be the making of mash-ups.

Having a couple of web services, analysts and vendors will often say, does not constitute a proper SOA – many of the advantages of reusability and efficiency come from the supplementary technologies such as the repository and the enterprise service bus.

But by the same token, an organisation does not need to have successfully deployed a mature SOA – and most haven’t – to be able to create a mash-up. Applications and data can be exposed as web services with relative ease, allowing the IT department to create lightweight mash-ups.

Indeed, what examples of enterprise mash-ups that do exist suggest that they are of greatest use not as the cherry on top of a long and expensive infrastructure cake of a project, but as a quick and cheap alternative to large-scale integration work.

“We’ve been talking to one customer in the financial services industry that is undergoing a significant merger,” says IBM’s Tutt. “It is going to use mash-ups [with IBM’s Lotus Mash-Up Centre product] to integrate the two companies’ customer data; if a customer asks for a loan, [agents within the combined operations] can look at all the data they have on that individual in one place.”

“They are certainly looking to integrate the databases in a more traditional fashion eventually,” he adds, “but this allows them to maintain service while taking their time over the integration.”

Bungee Labs is an innovative company based in the US that offers its customers a web-based platform for mash-ups, although Ted Heager, director of the company’s developer network, doesn’t use that term: “It suggests a low-value application,” he says.

One of its customers is a printing company that uses the Bungee Connect platform to create a single point of integration between its various business applications. These are exposed as web services using Microsoft’s SQL Server product, and integrate with one another using REST application programming interfaces (APIs).

The end result is an integration of all its applications in such a way that data and logic is passed between applications, but which requires no supplementary infrastructure investment on the part of the company itself.

Being a middleware vendor, Tibco sees composite applications (the term it favours) as a business case for infrastructure. Even so, many of the 600-odd customers that use its composite application bundle, that includes an enterprise service bus and other pieces of SOA infrastructure, are doing so not as part of an SOA strategy per se, but as an alternative to ERP consolidation.

“Many of our larger customers don’t want to standardise on one single ERP, they prefer to do things their own unique way and not the way prescribed by SAP or Oracle,” explains McNamara. “So what they are doing is keeping those different applications separate but mashing them up at the front end.”

These examples demonstrate just some of the many ways in which web services-based integration is being used by businesses. Mash-ups can be seen as just one tool at the IT department’s disposal to help it integrate web services. They are useful when practical or logical constraints demand that the integration occurs at the level of the user interface, rather than deep inside the IT infrastructure.

Musser, the mash-up guru, believes that the term itself has a limited life span. “What will happen is the word mash-up will be subsumed into broader concepts,” he predicts. “This will just be how software development is done.”

Cloud services

Organisations that do not use web services internally might wonder, in light of the above, what use the skills and techniques required to build mash-ups will be to them.

However, web services – true to their name – need not originate from within the organisation.

To date, most business use of hosted software has concerned discrete applications, as in the case of Salesforce.com or Google’s GMail service. However, the idea that small pieces of functionality – analogous to a software service in an SOA – can be sourced from the web is one that is growing in viability.

Offering ‘cloud computing’ in this sense was pioneered by online book retailer Amazon.com, whose Amazon Web Services allow developers to call storage, computation and database services – hosted on its servers – and include them in their own applications.

In October 2008, Amazon.com announced that it was preparing to add support for Microsoft’s development platforms, such as SQL Server and .NET, to its services, meaning that developers can build Windows applications out of its web services.

And in November 2008, Microsoft itself unveiled Windows Azure, a forthcoming cloud computing service that will allow businesses to deploy applications on a cloud computing platform maintained by Microsoft itself.

“Azure is fundamentally built around the web services model,” explains Mark Quirk, Microsoft UK’s head of technology for development and platforms. “The mechanism for integrating with non-Windows software is based on XML, and the RSS and ATOM web standards.”

These developments again reveal that the use – and therefore need for integration – of web services in business is not tied to the popularity of SOA.

Of course, if an organisation is consuming web services from outside the firewall, then why not integrate them there too?

Certainly, companies such as Bungee Connect which provide an integration platform-as-a-service are likely to become more prominent. And as in the case of Salesforce.com and Google Apps, cloud computing providers will offer their mutual customers pre-built integrations at the click of a tick-box.

But these solutions do not on their own tackle the task of integrating cloud services with internal applications and on-premise data. And despite the headaches it causes them, some IT departments may not want to say goodbye to their integration work.

“There are going to be a large segment of developers who say [cloud-based integration] is a lock-in strategy,” explains Bungee Labs’ Heager. “They may think ‘In order for me to adopt this platform, I am getting myself into a situation that I got locked into’.”

Perhaps then, in the web services-enabled age when computing resources can be either sourced internally or from third-party suppliers, integrating at the user interface level will emerge as a way to fend off vendor control over the terms of that integration.

This is speculation, however. What can be said with certainty is that gaining familiarity with REST APIs, with SOAP and WSDL and with the concepts surrounding mash-ups will put enterprise IT departments in good stead to accommodate emerging software trends, especially cloud computing.

Whether the term ‘mash-up’ persists in the vocabulary of software developers is another matter.

Pete Swabey

Pete Swabey

Pete was Editor of Information Age and head of technology research for Vitesse Media plc from 2005 to 2013, before moving on to be Senior Editor and then Editorial Director at The Economist Intelligence...

Related Topics

Enterprise IT