In-Depth
Status update: The real-time enterprise
- By Jack Vaughan, Lana Gates
- October 31, 2002
How real is the real-time enterprise? When ADT first devoted a cover story to this topic in June
1999 (The real-time
enterprise), we proposed that smarter middleware technologies were driving a fundamental
change in the way corporations did business. Then, as now, it was easier to
point to examples of real-time directions than to actually define the idea.
The general theme conveyed by that story, ''The real-time enterprise,'' was
that highly integrated, non-batch-oriented apps might be the wave of the future.
These apps could be connected to balanced scorecard systems that apprised
corporate managers of the status of their organization at any point in time --
so the story went. One event would initiate another, and a virtuous chain of
processes would kick off.
Our 1999 message was that the days of standalone apps were numbered, that the
Straight-Through Processing (STP) ways of Wall Street would become more
widespread, and that the increasingly sophisticated methods used by major
airlines to match demand and supply would be adapted by mainstream
corporations.
EAI vendor companies in the forefront of the movement at that time included
Active Software, Crossworlds, Neon, Talarian, Tibco and Vitria.
Recently, we went back to see what of the real-time enterprise came to
fruition. Well, a funny thing happened on the way to fruition. For one thing,
the Internet bubble happened. Flash-intensive corporate Web sites grabbed IT's
attention and back-room integration work often took less precedence. B2B
exchanges became the rage. Meanwhile, EAI efforts became bogged down in a
laborious effort to create adapters that connected existing systems to
newfangled ones. Then the Internet effort collapsed. Neither Wall Street nor the
airlines had the flair they did before. Strapped IT directors looked for
smaller, near-horizon projects and damn the 'vision thing.'
Moreover, ''Web services'' gained attention. Loosely coupled (some would say
''loosely defined''), Web services could be seen as a thinly veiled alternative to
the tightly coupled middleware adapters that marked the real-time
enterprise.
Of vendor companies we pegged as in the forefront in 1999, only Tibco and
Vitria still remain as standalone vendors. Active is part of WebMethods; Neon is
part of Sybase; Talarian is part of Tibco; and Crossworlds is part of IBM.
Yet the vision of the real-time enterprise has not quite vanished.
As in the past, it has many names. As they do on Wall Street, where new
requirements continue to drive faster settlement methods, some may call it
''straight-through processing.'' It could also be called ''the zero-latency
enterprise'' or the ''enterprise nervous system.''
While the balanced corporate scorecard is seldom discussed, talk of real-time
analytics is active. The corporate cockpit, a concept that goes back to the days
of the Executive Information System (EIS), still resonates. The call of the
event-driven organization is sometimes heard. Sometimes, what is described as
''orchestrations of business flows'' has the tenor of a real-time song. Research
group AMR has taken lately to describing an entity known as the
performance-driven enterprise. It does not sound unfamiliar.
Meanwhile, a host of semi-related niche concepts -- Business Activity
Monitoring (BAM), Business Process Modeling (BPM), Business Technology
Optimization (BTO) -- point toward something like the real-time enterprise. For
its part, IBM, in discussing its rationale for buying Crossworlds, as well as
other, smaller companies, talks about 'process integration' as a primary
corporate task at hand.
But the climate is challenging. Promoters of real-time enterprise solutions
will be asked just how big a job they are proposing, how much speed is really
needed, what type of team is needed, along with the classic question: ''Can't
this wait?''
Perhaps surprisingly, given the spotlight placed on middleware makers, ERP
vendors have been active in promoting real-time enterprise concepts. Some argue
that such apps still show a batch computing heritage, but for users, better
links between enterprise application suite elements are a positive
development.
Too real, too soon
If the move to a real-time enterprise has slowed, it
could well be due to a general reluctance to pursue large, bet-your-business
projects. ''People don't want big-time projects right now,'' said Tony Baer,
principal with New York consulting firm onStrategies. ''That means middleware is
tactical today. It's an uphill battle to justify anything strategically,'' he
said. Many factions must buy in to the project, and there are a lot of
subjective positions to grapple with, he warned. (Not too long ago, Baer took a
look at the phenomenon of real-time analytics in these pages. See Analyzing data in real
time
.)
''What I see is that people are skeptical of the idea of the real-time
enterprise,'' said Chuck Chase, vice president of customer solutions at Vizional
Technologies, a Santa Monica, Calif., provider of distributed operational
inventory solutions.
''People have a lot of software they bought that is also expensive to
implement,'' Chase said. ''Now people are looking ahead in a six- to nine-month
time frame. People will try things if they see value, but no one is trying to
integrate processes down to the minutest detail. Yet, they feel taking out cycle
time is good.'' If development managers try to integrate down to the smallest
detail, they probably cannot price-justify the effort, he said.
''The real-time enterprise is about apps talking to apps without humans
involved. But the real-time enterprise is coming slowly,'' said Mike Hoskins,
who, as president of data integration specialist Data Junction, Austin, Texas,
has seen the fortunes of competitors positioned as makers of real-time
middleware adapters ebb and flow. In the late 1990s, he said, EAI was seen as a
big market where custom software had unnecessarily flourished. Better software
tools would be the answer.
''But, unfortunately, the [EAI vendors] missed the boat on what makes a
successful tool,'' said Hoskins. ''It has to be powerful and quick, easy to learn
and nimble to use.'' Parts of the middleware industry broke down, he said,
because the software was too big and too hard to use. Some of the tools were
barely advantageous vs. integrator troops doing custom work.
''There's no such thing as zero latency because there is friction. Friction
happens at the integration points. What stopped the real-time enterprise was
latency, the integration points,'' said Hoskins.
''I think 'the real-time enterprise' is an obsolete concept,'' said Denis
Mathias, a partner in the high-technology practice of PricewaterhouseCoopers
Consulting. Mathias, who works, for example, with semiconductor firms who must
carefully align Web-enabled inventory systems for optimal collaboration with
suppliers, reiterated a point worth repeating: Real-time is relative. Wall
Street-caliber transactions are not a good example of the way most organizations
must go, he said.
''What kind of response do you need, for example, for a collaborative
inventory system?'' he asked. ''Not real-time. Most companies have monthly
planning cycles. Some have weekly planning cycles. For many, just to go to a
weekly planning cycle is an improvement.''
The problem is that moving to real time is not just a technical problem.
''Real time puts a lot of stress on organizations. To connect, you need a lot of
change in organizations, especially changes in data,'' said Mathias.
There is an inverse relationship between flexibility and integration, Mathias
insists. If the connection is more tight, then it is probably less flexible, he
said. ''That was the problem with ERP systems,'' he concluded.
For Mathias, the real-time enterprise is a suspect concept because it is
''high cost, high risk and low flexibility.
''We need practical solutions -- loose connections,'' said Mathias, noting
that certain high-transaction environments need tighter connections. (That is,
connections relying less on the XML and HTTP solutions that mark Web
services.)
Real-time hangs tough
But the real-time enterprise is not
without its stalwarts. And the idea persists that the types of fast application
connections familiar in financial systems can benefit a wider business
community.
''Wall Street [houses] are early adopters in the broadest sense,'' said Don
O'Toole, director of business integration marketing at IBM. ''They have the
greatest need. But we see use of the real-time enterprise across the board.
''We see it in manufacturing, in retail, in health. It's fundamentally an
outgrowth of the pervasiveness of networking these days,'' added O'Toole, who
admitted that a mix of loosely coupled and tightly coupled protocols may be
required for 'real-time' to become real.
''Web services is part of it. When you look at the broad category of EAI
software, you see a category being transformed. But the model is still very
young,'' explained O'Toole, who noted that emerging Web services workflow
standards are worth watching here.
''We think [the real-time enterprise] is a reality, but it is still a
formative concept,'' he said. ''We're still trying to figure out how to do the
real-time enterprise in the most efficient manner.''
IBM clearly jumped deeper into this fray at the start of this year with its
purchase of Crossworlds. O'Toole said the prime purpose was to gain Crossworlds'
capabilities in business process integration. There is a thread then that
connects this buy to the more recent purchases of Holosofx, for business process
modeling and monitoring tools, and Metamerge, for its meta directory support,
said O'Toole.
Increments happen
Roy Schulte, vice president and research
fellow at Gartner Inc., Stamford, Conn., is among those who anticipated greater
progress toward the real-time enterprise in recent years. But, said Schulte,
slow uptake does not mean diminished impact. ''The idea is more widely practiced
now than four years ago, and will be more so four years from now,'' he said.
Web services and the real-time enterprise are not contrary notions, he
maintains. ''Web services will help some types of real-time enterprises,'' Schulte
said. ''If you have Web services, it is a little faster to build sometimes,'' he
said, adding, ''hand-coding adapters can take a lot of time.''
The hard part is looking into the application to be integrated, said Schulte.
It may not be built to effectively participate in real-time communication
exchange. And some might suggest that ''wrappering'' legacy apps, an alternative
to rewriting, may have limits here.
If application development managers can build an app from scratch with a Web
services interface, it will be easier to move to real time in the future, said
Schulte.
Thus, AD can incrementally create real-time enterprise integration. Schulte
said development managers should proceed in increments. Gartner Inc.'s
recommendations: Look at the business processes; identify which should be
real-time and which should not; and look at which app has the biggest payback
before establishing a priority of targets. Schulte notes, as do others, that
there are a lot of processes where ''overnight,'' or even ''a week,'' is pretty good
performance.
ERP at bat
In manufacturing companies where multi-year efforts
have created mature, tightly knit ERP systems, the move to real-time
cross-departmental systems is incremental. Yet large-scale, inside-out efforts
to create the event-driven organization are also afoot. According to Kevin
Prouty, research director, automotive strategies at Boston-based AMR Research,
''The most important part of real time is being just in time with information.''
That involves knowing where the right information is at the right time --
getting the right information soon enough to solve the problem.
Real-time enterprise performance management, as noted by Roddy Martin, AMR's
service director for consumer goods and life sciences manufacturing, should
focus on the time-sensitive critical path of business and marketing processes.
''Content standards and clean data are critical,'' he said. And it should go
without saying that the key to the real-time enterprise is first defining ''real
time'' for your organization. (Martin, Prouty and others spoke about real-time
issues at the AMR Spring Executive Conference in Scottsdale, Ariz. earlier this
year.)
Albert Stroucken, chairman of the board, president and CEO of H.B. Fuller
Co., St. Paul, Minn., acknowledged, ''Opportunities wrought by e-business are
exploding.'' In fact, those opportunities have changed the way H.B. Fuller
conducts business. ''The key to incorporate e-business,'' Stroucken said, ''is
re-architecting the inside and outside of your business, not simply 'Webifying'
the company.''
For H.B. Fuller that meant becoming an event-driven organization. This
involved becoming customer-centric and beginning to cater profitability
transparently to its Web customer base.
TaylorMade-adidas Golf Co., Carlsbad, Calif., is also using technology to
drive changes. The company started with a legacy system, a small manufacturing
module and an accounting package. Everything else was manual. The company
launched products in January and delivered in May. ''Sixty percent on-time
delivery was normal,'' noted Mark Leposky, vice president of global
operations.
The company stepped back and took a new look at its goals, objectives and
past performance. It selected a new technology platform and became a
process-driven rather than a functionally oriented organization. TaylorMade
views technology as the key enabler. ''Technology is not a strategy,'' Leposky
noted. ''It's an enabler to your strategy.''
TaylorMade also added supply chain as a key component of its business
strategy. With all these changes, it now takes TaylorMade 25% of the time it
used to take to administer an order for golf clubs.
Verizon on horizon
In its simplest terms, real time, according
to Joe Serra, director of ERP systems at Verizon Wireless, Orangeburg, N.Y.,
means ''not batch. I want my customers -- internal employees -- to be able to get
at information that's up-to-date, update information on themselves, and have
that information reflected immediately in whatever internal system that data's
used in,'' he said.
To do that, Verizon Wireless has implemented PeopleSoft's ERP system. With
the functionality in PeopleSoft 8, which Serra said is ''leaps and bounds'' beyond
the previous version, meeting those objectives is becoming more of a reality.
Although Verizon still uses batch processing for payroll jobs, the updating and
availability of that information to employees and managers is made in real
time.
Prior to adopting PeopleSoft, Verizon had much more limited self-service
capabilities and ''information was always on lag, maybe a day or so behind,''
Serra noted. Verizon had its own homegrown kind of product that required
specialized skillsets within a team. The company felt at risk because if a team
member left, no one else knew that particular job.
Verizon always wanted to migrate toward real time, but did not have the
capability to do so. Yet, Serra said, there is a high demand for information
that is up-to-date and to-the-second. That demand helped drive Verizon closer to
real time through PeopleSoft. As a result, ''we've been able to get rid of our
legacy operation and are now using out-of-the-box self-service functionality,''
Serra explained.
For example, if an employee is moved from department A to department B,
online organizational charts immediately reflect that. Before, if an employee
changed their home address, they would have to update that change in three or
four different places; now, a single entry automatically propagates through the
system.
Before adopting PeopleSoft, Verizon saw a trend of people not using certain
functions in the homegrown product because there was uncertainty as to what was
truly accurate. Now, however, ''people are more apt to use the functionality
that's out there because people know it's up-to-date and what they're seeing is
up-to-the-minute,'' Serra noted.
Verizon Wireless is completely moving toward a model of real-time updates,
although the company does still use batch processing for some things. ''There
will never be a need to do everything in real time,'' Serra explained. ''Even in
the financial world, it makes sense to run some things in batch mode. There is
always room for that type of thing.
''When accessing information,'' he continued, ''that's where real time makes
sense.'' And Verizon Wireless is well on its way to extending that to its entire
organization or enterprise. ''We're going to expand that to wherever we can,
minus the typical payroll jobs that are just built to run in batch,'' said
Serra.
Part of that expansion at Verizon will include rolling out an employee portal
in 2003. ''One of the big drivers,'' Serra noted, ''is the ability to access a lot
of ERP functions without a user having to log into those applications
separately.'' In addition, the user would not have to be aware they were
accessing different applications.
Although what Verizon currently provides in the way of self-service is small
in scope and limited to employees, managers and recruiters, the company has a
desire to customize information to the individual accessing it, and thus move
even closer to real time.
See the related story Aligning IT and business goals
by Colleen
Frye.