Enterprise mashups: Tools build data integrations Organizations have long labored under an outstanding backlog...
of data integration, business intelligence and customized reporting jobs. Over the years a variety of technologies have arisen to address these or similar problems - most recently, so-called enterprise mashups.
Enterprise mashups have arisen in recent years as a tool said to let end-users - with some help from developers - build their own data integrations. Services are central to most of these offerings - often too are AJAX front-ends. Still, the enterprise mashup space is notable for diversity. Data staging, publish-and-subscribe messaging, XML data mining and other traits may be included at different times.
Mashups present data from a number of internal and/or external disparate data sources in one unified dashboard. Among a host of enterprise mashup and related software specialists are Active Endpoints, Actuate, Composite Software, Cordys, Denodo Technologies, IBM, JackBe, Kapow Technologies and others.
As always, when evaluating options, culture counts; that is to say: different organizations will find favor in different approaches. How close users are to data, how much analytical ease any mashup brings to the job of working with data – this is important.
Vendors variously focus on BPM-oriented mashups, data-oriented mashups and vivid front-end data visualization visualization. Of wide interest at the moment for the mashup crew are business intelligence apps, where diverse data sources are tolled up to provide a useful analytical view of the corporation.
Forrester Research's James Kobielus, analyst, laid out a range of enterprise mashup types for BI:
- Lightweight presentation mashup that pulls data out of applications so users can generate reports. No data warehouses are used here.
- Deep presentation mashup backed by a data warehouse and a large pool of analytic data.
- Full BI mashup in a federated environment. The user sees a logical unified semantic virtualization and can pull data from various systems/databases in a transparent way.
- Full collaborative mashup with IT governance policies around what a user can mash up. Data here is stored in a central repository and accessed through sets of permissions.
In this grouping, level 1 mashup maturity implies no prior BI experience whereas level 4 is practically nonexistent in today's enterprises.
Forrester's Kobielus said there are about as many ways to do mashups as there are enterprises. Integration style all depends on the data infrastructure and the company culture. Of the major vendors, he said those interested in enterprise BI mashups should watch Microsoft Project Gemini and IBM Cognos. These two products are advanced enough to help attain level 4 mashup maturity.
But before any enterprise can start dealing with mashups, it needs to get its approach down.
"Mashups won't succeed in a BI context unless it's fun - unless its highly interactive and the user says 'wow that's totally easy to use let me use that so I can build my reports,'" said Kobielus. "If it's approached where IT is forcing self-service on reluctant users, I think it will fail. It's got to be user enthusiasm that drives this."
"The key is to break up the work a bit," said Robert Eve, EVP of marketing at data virtualization vendor Composite Software. "We work on the data plumbing side. You have your data-oriented people - your data architects. Then you have people more focused on the business consumer and the application usage to work on the visualization side."
Composite Software helps its customers develop business intelligence (BI) mashups by providing a server to virtualize and federate data from multiple sources. In effect, it is a layer that mirrors the data and sends it directly to the system providing the mashups.
Enterprise mashup case study: 2. Knowing you data
Pharmaceutical giant Pfizer has been using SOA and mashups to help bring products to market since about 2005. Research Fellow Michael Linhares said the company configures its own mashups and wouldn't have it any other way.
"We do it ourselves because it's more about knowing the information and the relationships of one piece of information to another," said Linhares. "Knowing the data is what's important. The technology isn't such a big hurdle anymore with all the tools out there."
Pfizer uses fairly straightforward technology for its infrastructure. They use Composite Software Information Server for data integration, BusinessObjects WEBi reports, Spotfire DecisionSite analytics, SharePoint Designer and simple ASP .net pages for presentation.
Linhares said the challenge in developing mashups was the same as in moving to a service oriented architecture: culture. In an industry so rooted in scientific research, employees can get very attached to the data they generate. A cultural shift toward open data sharing had to take place.
In the IT department, the culture shift was one of empowering the business users. With an advanced SOA sitting under fairly easy-to-use utilities, business users were better able to configure some of their own tools. But the IT department needed first to be persuaded that not all infrastructure tools are IT tools.
Once the culture was more open, Pfizer was able to bring about an attitude of experimentation in software development. Whenever they begin developing new tools, end users get to play with them well before they go into production.
"I think about doing software development in a very research-oriented way," said Linhares. "We've now gone to a model where we're working in at most three weeks to three months turnaround from starting a project to delivering it to the business."
In a lot of ways, working with enterprise mashups is like working with SOA, only on a much smaller scale. There is no out-of-the box solution for everybody and it is an approach that matures through trial and tribulation.
Enterprise mashups: 3. In search of mashup standards
The diversity found in enterprise mashup approaches has pros and cons, spurring innovation – and proliferating multiple system types. Even as analysts recognize mashups as one of the top technologies of 2009, they admit the variety plays against standardization.
Not surprisingly, some vendors have banded together under a common standard, Enterprise Mashup Markup Language (EMML). Last month JackBe, the mashup specialist that created EMML, announced the Creative Commons release of the language and the formation of an
Open Mashup Alliance (OMA)
. Already the OMA has recruited top names like Adobe, HP, Capgemini, Bank of America, and Intel.
"This is an up and coming trend," said John Crupi, CTO of JackBe. "The whole domain-specific language idea has been talked about in the past. It's tuned to be more aligned with patterns of enterprise mashups, such as calling the data from the services and doing it in real time."
Mashups are growing in popularity as a way of organizing and displaying business intelligence (BI) data, but what is a company to do when the data it needs to analyze lives out on the Web? This has been a special quest of Kapow Technologies which last month partnered with StrikeIron to provide a visual tool for aggregating and processing any data searchable on the Web.
With Web Data Services 7.0.0, users can create custom "robots" that can log in to accounts, collect information, organize and manipulate data and serve up custom-tailored figures. This configuration is done through a visual process map that does not require any coding.
This would be useful for someone who spends a good amount of time, say, researching product pricing across many sources to present at weekly meetings.
"In almost every company, there is somebody pasting data from Web applications into Excel to produce a report that they have to deliver," Stefan Andreasen, founder and CTO at Kapow Technologies. "Here you can create a robot that goes and grabs the latest numbers and popular excel spreadsheets. So basically all you have to do is click refresh." To be clear, Kapow's new offering does not create mashups. Rather, it can provide the data needed for a mashup. Having recently joined up with the Open Mashup Alliance (OMA), the company's product can interface with the mashup builders of any other member, such as JackBe. The product can also be used to enable a Web mashup for mobile use.
Managing the mashups
"A mashup is a form of integration. Web services and more dynamic interface technology have created deeper interest in utilizing such tactics," said Sandy Rogers, independent industry consultant. Still, she adds, many enterprises will continue to create their own mashups using their own in-house technology.
"As firms look at mashup enabling alternatives, research has indicated that they often opt to use their existing tool sets rather than add yet another development environment into the mix," said Rogers.
Meanwhile, managing how mashups are done across the enterprise is an emerging issue.
"Vendors have been adding capabilities to address 'mashups', including federated information integration, business process management, portal, lifecycle management, and more, along with what has been marketed as "specialized" mashup development environments," Rogers noted.
A key element to consider, according to Rogers, is what is needed to help govern mashup-oriented activity on an ongoing basis. Coordination with other Web services and IT management programs is key, she added.