When it comes to obtaining data, the traditional methods for business intelligence (BI) and application integration relied on IT-centric integration engines. Integration engines are necessarily built near the center of the IT architecture and often require existing systems to be altered or wrapped in some way. The value of an integration engine is that once an application or database fits in with the integration engine, it is also integrated with all the other systems that are plugged in to the integration engine.
But what if you could pull the data you need without altering the existing system at all?
Now that SOA has become a way of life for IT organizations, the integration engine may be due for an overhaul. Because so many systems are Web enabled with either REST or SOAP, the data is fairly easy to access. Enterprise application mashups offer a way to put all interoperable and reusable Web services and REST APIs to the most use. They allow developers to plug into various data sources -- Web services, databases, SAP, NetWeaver, Excel data, data warehouses -- and deliver that data to users in readily usable and customizable ways.
The enterprise application mashup sits in the enterprise architecture stack between the user (above it) and the application layer (below it). Thus, it looks like a user interface. But mashups send feelers down into the other layers -- into databases, enterprise applications, mainframes, to whatever they need to connect. The mashup connects to the existing system and adjusts to what's already there, not the other way around. When it comes to data acquisition, mashups can start up with simple screen scraping techniques that can, if necessary, be replaced by a more permanent data link. Similarly, mashup applications can be built to input data to existing systems using the same methods as their existing user interface.
According to enterprise application mashup evangelist John Crupi of JackBe, the lightweight nature of mashups makes them a better fit for real-time analytics than traditional integration engines. The challenges with real-time analytics revolve around developing apps quickly and meeting rapidly changing business requirements. An integration engine, with its high overhead and strict requirements, is ill fitted to these challenges. The way to attack this problem, according to Crupi, is to use application mashups to build business dashboards that can be customized to fit quickly changing needs.
Crupi says there are five characteristics of real-time analytics that make this field a good fit for enterprise applications mashups. Real-time analytical BI situations tend to be:
- Operationally centric, situational and actionable. They involve frontline people looking at events as they happen and with an immediate need to react. For example, as intelligence agencies track suspected terrorist groups they have to know what's going on and they have to take some action based on what they find. And the situation in which they're working changes as they move from group to group.
- Information is always changing so you might not know what the information will look like from one week to the next. This is a sharp contrast to the type of reporting that IT is accustomed to where traditional information systems run the same report with the same data points on a routine basis.
- Generally, users interact with the data visually via charts, map and drill downs. An Excel sheet with rows and columns full of numbers isn't as useful to the end user as a pie chart that shows what those numbers mean. Also, if the numbers show a sudden drop, users want to be able to take a closer look at the data and find the reason for the change.
- The data is derived from many sources. At one time data was fairly centrally located in a mainframe or a data warehouse. Now, it is common for systems to merge and include multiple data warehouses as well as integrating data from sources like SAP, Salesforce.com and even Excel spreadsheets saved on the desktop. The data ends up too scattered for traditional information-gathering techniques.
- Workers want the information on location. The boom in mobile devices is spearheading demand for enterprise applications to be available on smartphones, netbooks and tablets. As we've already heard from resident enterprise mashups expert Mike Ogrinz, applications mashups and mobile applications are a match that makes sense.