 |
The rapid pace of business requires new approaches to business intelligence (BI).
The traditional approach — cleansing data, gathering it in a data warehouse, then having the IT department create reports against it for business users — is much too slow for enterprises to remain competitive.
“Big companies like airlines and consulting organizations are looking at faster cycle times to provide applications to business needs,” Mark Smith, CEO and executive vice president of research at Ventana Research, told InternetNews.com.
While most enterprises bought “quite a lot” of different business intelligence
One such vendor is Actuate (NASDAQ: ACTU), which has “large-scale applications at many banks that go to hundreds of thousands, or millions, of customers,” Nobby Akiha, the company’s senior vice president of marketing, told InternetNews.com.
“The Internet has become a key interface and people are seeing the kinds of things you can do on Google (NASDAQ: GOOG), Yahoo (NASDAQ: YHOO), Amazon (NASDAQ: AMZN) and eBay (NASDAQ: EBAY), and are expecting to do similar kinds of things on their banking, investment and insurance sites,” Akiha said. “That will eventually go to applications delivering information inside the corporations.”
Right now, Actuate is “focused on evolving our technology to stay ahead of innovations in Web 2.0 rich Internet application (RIA) capabilities like AJAX and Slash and making the information we present more interactive and dynamic,” Akiha said.
Over the past three years or so, this involved working on the Business Intelligence Reporting Tool (BIRT) project, an initiative of the open source Eclipse Foundation. BIRT is an open source reporting system for Web applications.
In the second half of the year, Actuate will combine its Adobe Flash-based performance management application with more-interactive analytic and reporting capabilities based on BIRT, and integrate it better with other RIAs.
“Many of our customers “are looking to upgrade existing applications with more interactive capabilities,” Akiha said.
Other open source initiatives embedding BI inside applications include JasperSoft and Pentaho, Ventana’s Smith said.
Another vendor leveraging new technologies is LogiXML, which is “a broad BI platform” competing with Business Intelligence, IBI and Cognos, now part of IBM, director of marketing Bill Kotraba told InternetNews.com.
Users “can build all sorts of complex charts and BI reports and place them anywhere with a snippet of Java code,” Kotraba said.
These JavaScript code snippets, or widgets
LogiXML has a full-featured Web reporting system, Logi Info, that offers interactive analysis, portals and dashboards. It includes Adobe Flash dates and maps that go down to the county and state level in the United States.
In addition, Logi Info has a pop-up display feature, so when users want more information from their charts it pops up instead of their having to drill down. The firm recently unveiled Widgenie, an SaaS offering.
Next page: SaaS catches on
Page 2 of 2
SaaS catches on
BI offered in software as a service (SaaS)
A company called 1010data offers BI SaaS and has been building large-scale databases on Wall Street “for decades,” according to David Frankel, the company’s vice president of business development.
It uses standardized hardware, but the software is its own “secret sauce.” In addition, 1010data built its database from the ground up to handle large-scale analysis problems, Frankel told InternetNews.com.
1010data’s product is data agnostic, and can handle “any structured, tabular data” in the consumer products, retail, pharmaceutical, equities and fixed-income securities markets.
The firm currently offers an XML
BI offered as a SaaS could be valuable for organizations that want speed, but it “won’t be a good model” for those that prefer data locked up in their infrastructure and want to add to it, Smith warned.
Changing directions
The advent of new technologies led longtime BI applications vendor Cognos, now owned by IBM (NYSE: IBM), to change directions.
Cognos rearchitected its Cognos 8 platform around Web and Internet-based technologies three years ago to enable SOA. In addition to sitting on an SOA architecture, Cognos 8 is available on mobile devices, including the BlackBerry, and wireless devices running Windows Mobile and Symbian S60.
“We’re seeing a lot of new technologies our customers are adopting as part of their IT agenda — composite applications, mashups
Cognos also provides hardware and software BI appliances in its CognosNOW product offering.
BI appliances can either be software products or a combination of hardware and software. Both approaches are designed to speed up creating business intelligence.
Next page: SQL on disk
Page 3 of 3
 |
Data and more data. |
SQL on disk
Netezza (NYSE: NZ) offers an appliance that consists of a SQL database stored on the disk rather than on the server, and up to 800 snippet processing units, each consisting of a field programmable gate array (FPGA)
All 800 snippet processing units can work concurrently on a problem.
Traditional databases keep data in storage and bring small amounts up into the database on the server when a query is run. This is too slow for business intelligence, where “typically you want to analyze vast amounts of data to find some sort of pattern,” Tim Young, Netezza’s vice president for corporate marketing, told InternetNews.com.
Netezza’s customers analyze “tens, if not hundreds,” of databases. Most are telecommunications companies, which use its solution to analyze the large amount of details in call detail records to detect “core patterns for network provisioning, marketing purposes like special offers and for other purposes,” Young said.
Another new approach to BI is focused not on speed but on minimizing the waste of time and maximizing ease of use by building frameworks of reports users can modify as needed.
Managing resource consumption
“BI is getting bigger, growing from a departmental solution or application to an enterprise resource, so it needs an infrastructure that can support thousands or tens of thousands of users, not hundreds,” Microstrategy vice president of products Mark LaRow told InternetNews.com.
The traditional method of building reports on request consumes a lot of resources, and an IT department would have to build more than a million designs to cover a modest data warehouse, according to LaRow.
Now, IT designs “a small number of reports — maybe 1,000,” and users can take one report, click on different rows and columns to drill down, LaRow said.
He said they can surf the data they want, adding the data they need dynamically and either saving their version of the report or using it online.
For example, one of Microstrategy’s customers, the Loews (NASDAQ: LTR) hardware retail store chain, has 94,000 reports anyone can use. Only 3,000 were “explicitly designed by IT,” and the rest are variations of those with additional data added by users, LaRow said, adding that some of the derivative reports are “much larger than the originals.”
None of these approaches is fast enough for Terry Cunningham, the founder of Crystal Decisions, which sold the Crystal Reports BI tool. He is often considered the father of business intelligence.
BI is “evolving into what we call continuous BI; it never rests,” Cunningham told InternetNews.com.
“Things need to be understood, and decisions need to be made in real time,” he explained. “There’s no such thing as ‘I’m going to run the report tomorrow and see how we did today.'”
Next page: Complex event processing
Page 4 of 4
Complex event processing
Cunningham is CEO of Coral8, which offers complex event processing (CEP), where “you have a continuous pulse line that’s monitoring everything in real time.”
CEP is an event-processing concept in which multiple events in an event cloud are processed to find the meaningful events. It lets enterprises extend BI down to previously ignored areas such as the shop floor, where “there’s a high noise to signal ratio, meaning only a relatively few of the millions of data events are meaningful,” Cunningham said.
“The typical database can handle 6,000 inserts a second; we’re talking 200,000 to 300,000 events a second in CEP; more than 100,000 is typical,” Cunningham said.
CEP analyzes data in real time, and is complementary to existing data infrastructure. “We feed databases, we don’t replace them,” Cunningham explained.
One of Coral8’s customers is Sallie Mae, the nation’s leading provider of student loans, which “has this huge command center looking at its Web site traffic in real time,” Cunningham said. This is in addition to Sallie Mae’s existing BI systems.
Up until now, all approaches to BI, whether traditional or not, have relied on data cleansing — ensuring that data is consistent and accurate. But data cleansing is a laborious process, and Stef Damianakis, CEO of Netrics, believes it is unnecessary.
“The standard is, first you do cleansing, then you do matching, and then you have this huge effort of getting the data perfect, which is not an achievable goal in practice,” Damianakis told InternetNews.com.
“Even if you get your data perfect today, what’s going to happen to it over the next week, two weeks, two months? And can you be sure it was entered perfectly into the system?”
Netrics’ approach is to use mathematical modeling to get results in spite of imperfect data so that, “instead of focusing on the data standardization and data perfection phases, you can use the data as it is and the model can overcome discrepancies in the data.”
Standard methods of querying data with BI tools are clunky because “there is a translation from the question being asked to a set of rules either the business analyst or IT has to write to query the data,” Damianakis said.
He suggests that giving the business user the ability to build a mathematical model, or widget, that answers his questions, would be more productive.
Netrics has two products, The Matching Engine, which models human similarity, and The Decision Engine, which mathematically models human decisions. The two can work together or independently.
The company’s solutions can “mathematically model human questions,” and its technology is embedded in solutions from vendors. Partners include Northrop Grumman (NYSE: NOC), Lockheed Martin (NYSE: LMT) and health care IT system provider Cerner (NASDAQ: CERN).
“Data can never and will never be perfect, and it’s time for software to stop requiring that and to build systems that will work even if the data’s not perfect,” Damianakis said.
