Real Time Enterprises A Continuous Migration Approach
Vinod Khosla, Murugan Pal
Vinod Khosla, Murugan Pal Real Time Enterprises – A Continuous Migration Approach Current market trends, global competition, and technological innovations are driving
enterprises to adopt the practices of Real Time Enterprises. Real Time Enterprises are
organizations that enable automation of processes spanning different systems, media, and
enterprise boundaries. Real Time Enterprises provide real time information to employees,
customers, suppliers, and partners and implement processes to ensure that all information
is current and consistent across all systems, minimizing batch and manual processes related
to information. To achieve this, systems for a Real Time Enterprise must be “adaptable” to
change and accept “change as the process”.
Any business process within the enterprise, including relevant processes in use by its trading
partners (the extended enterprise), must be instantaneously reflected in all enterprise systems.
In other words, all INFORMATION is ―real time‖ within a ―real time enterprise‖. All manual or
batch processes related to information in an enterprise are inefficiencies in the delivery of
products and services – unless the manual and batch mode processes (as in process industries) are required as part of the business nature. For example,
? In a Real Time Enterprise all the systems everywhere could recognize the new product
entered in a catalog system so that billing and customer service can be done right from
the moment that product becomes available.
? A wireless carrier could activate a wireless phone as soon as the credit card payment is
processed with out any time loss or manual intervention.
? A credit card company could improve customer loyalty by automating the dispute
notifications starting with the customer, and extending to the credit card company, the
merchant‘s bank, and all the way back to the merchant.
? Last year, Lexmark had $1 million worth of nonconforming material returned to one of
its plants in a single lot. Investigation revealed that providing engineers with adequate 3information online and in real time could avoid this situation and future inefficiencies.
? Citibank, to avoid heavy call volume in Poland around paydays, introduced cellular text
messaging (SMS) to inform all customers of any changes in their bank balances
instantaneously on their cell phone.
Today‘s business practices and models demand an operational environment acting as a virtual
enterprise, with insight into the status of customers, partners, and suppliers on a real time
basis while lowering SG&A costs. Cisco and Dell, with better service and higher revenue per
employee than their direct competitors, are great examples of leading technology adopters 3who have leveraged some of these capabilities. At the same time, companies like Lexmark and 7Cutler-Hammer have realized similar benefits through automation. Still, automation of an end-
to-end value chain has not been widely adopted or fully achieved. Even though technologically
this has been possible for some time, only now has it become realistic with the advent of
Internet-driven standardization. This has led to orders-of-magnitude cost reductions plus the
elimination of debate regarding the technical infrastructure to be used. With the advent of
Internet technologies such as HTTP, HTML, standardization initiatives around XML, Web
services, UDDI, and SOAP, it is now possible. Lexmark leveraged Internet and thin client
technology to enable their engineers to monitor production processes at their suppliers in real-
time, from anywhere in the world and put defective product on hold at the source. Dylan
Tweney in his column states, ―Web services will enable companies to link up their enterprise
systems with the production processes, bringing executives ever closer to the ideal of ‗real-
time enterprise computing,‘ and in turn will make companies better able to respond rapidly to 7changing market conditions‖. We agree.
Page 2 of 30 Real Time Enterprises – A continuous migration approach Vinod Khosla, Murugan Pal March 2002
Cisco‘s ambition to close their books on a daily basis is the litmus test for a Real Time
Enterprise. Consider the benefits of having all information current in all systems such that
books can be nominally closed within hours of the close of a quarter (or day!). Cisco does that
today, and after adjustments including managerial and auditor input can announce financial
results within three days of the end of the quarter. Think of the cost savings in finance alone!
Cisco Systems' much-vaunted electronic order-entry system has decreased the rate of errors for 5the company from 20% to 0.2%. If a majority of the orders come in untouched by humans,
think of the sales force efficiency and yield improvements. If most employee information (such
as vacation days and 401K‘s) is ―self service‖ on the intranet, think of the savings in HR. If order status, product configuration, and ―available to promise‖ dates are self-service for
customers on the Web, think of the improvements in customer service that are possible, while
reducing costs. These improvements are not just about a Website but a structural change to
Web enabled IT. The benefit of ―self service‖ is enormous as data will be cleaner when the
owner enters it and the process will be efficient because it is outsourced to the end customers
themselves. This is how the Internet is being used for information transport rather than a
browsing medium. It represents a change in the way finance, operations, HR, logistics, and the
whole corporation works. According to Roland Berger's Geissbauer, manufacturers already using
the Internet see annual cost savings of 6 percent across the value chain. From procurement to
Web-based supply chain management and after-sales service, it may be possible to cut costs by 3as much as 8 percent to 10 percent. We think the savings are potentially larger. On the other hand, Cisco failed to automate its supply chain deep enough into it‘s partners, resulting in
hundred‘s of millions of excess inventory beyond what a normal demand forecasting error
would have caused in a full real time, full visibility environment.
The reality is that the vast promise of IT, by and large, has been a mirage for most corporations.
But does this have to be true? Does the promise only work for some organizations? Are missed
opportunities for productivity gains just that, or can they be realized? Are budget overruns,
process delays, and ―additional costs‖ to recover the investments already made an unavoidable fact of life? We have gone from MIS departments with large in-house software development
efforts, to inexpensive desktop enablement, to large packaged software applications, to
productive application development tools, client server applications, portable environments
like JAVA, and system integration tools. We have gone through IT consultants, outsourcing
vendors, ASPs, the Big 5 and their system integration expertise, but the problems of IT remain
largely the same. Even more tantalizing are the stories of ―benefits‖ of good IT strategies.
Cisco has substantially higher revenues per employee than its direct competitors, Nortel and
Lucent. The cost of doing business is lower at Cisco, and their responsiveness and customer
service levels are higher. Geissbauer stresses that many of the top challenges for
manufacturers relate to competitive pressure and manufacturers need to respond faster to
customers in order to achieve their top-line sales goals. Dell can offer ―mass customization‖
and still maintain much higher inventory turns than its competitors, generating greater
profitability in its PC business. Wal-Mart and Amazon have used IT technology as strategic
weapons to increase their competitiveness. FedEx could not economically provide the level of
customer service that it does without IT. The cost to FedEx of a ―package pick up call‖ or a
―where is my package‖ inquiry have declined substantially (greater than 10X) because of the
use of appropriate technology. We guesstimate that each 1% (of sales) of increased IT spending
or spending redirected from rigid and outmoded forms of systems integration in a corporation
should reduce SG&A spending by 1.5% to 2% (of sales) beyond the improvements in IT
productivity. It is important to note that the bigger role of IT is not managing IT functions and
expenses, but rather to manage expenses and service levels for the ―rest of the corporation‖.
IT can be a strategic weapon and help reduce SG&A costs relative to competitors while
improving customer experience. The rate of savings depends on industry sector; those sectors
with high SG&A can realize the most significant benefits. This applies especially to business
processes, collaboration environments, and personalized applications where the bulk of
enterprise activity can be automated. It is clear that every business process, every manual or
Page 3 of 30 Real Time Enterprises – A continuous migration approach Vinod Khosla, Murugan Pal March 2002
batch process, and every human touch point that is properly automated and eliminated will
result in an economic saving as well as an improvement in quality by reducing the risk of human
error and improving the availability of information.
Any astute CEO or CFO will choose to automate and eliminate provided that the risks can be
managed and results can be demonstrated in small projects with 90-180 day implementation
and payback cycles. IT can become a competitive weapon, with the CIO becoming as critical in
reducing costs or improving ―product‖ as the VP R&D, in defining products as the VP of
Marketing, in improving customer service as the VP of Customer Support, or in improving
operations and reducing operations and administrative costs as the VP of Operation or the CFO.
In summary, the CIO becomes a strategic leader.
The goal of this paper is to define the ―ground rules‖ of an IT transformation from the stark
reality of legacy applications forward to the promise of the future. In fact, the goal should not
be a radical transformation but rather continuous migration of systems, transforming the
organization into an adaptive enterprise. It is easy to define the ideal ―new‖ world, particularly with the promise of Web services, but harder to achieve a practical reorientation
towards such a goal. No magic bullets or ideal solutions exist. However, biases in certain
directions and small probing steps do help. Differences in approach can result in a significant
impact over a three to five year time frame. Ground rules and technology choices can make
environments more flexible, adaptable and pliable over time. It is this paper‘s goal to highlight
these, as well as the ―current best targets‖ towards an ideal environment. Keep in mind we are
dealing with first generation attempts at this new model and will run into many ―gottchas‖,
issues and practical problems. But iteration towards the above goal is necessary because it is
unlikely that we will ever have an instantaneous or ideal tool for the transition. What we will
need to do is dedicate an increasing part of our maintenance budget, the largest component of
most IT budgets, to approaches that will enable the legacy environment to move towards the
The inflexible structure of conventional systems has long been the subject of loud complaints
by top management. Today‘s customer expectations, evolving business models and technology
trends demand the need for adaptability. It is important to accept CHANGE AS A PROCESS,
rather than as an EVENT. IT spending must be evaluated against the total expenditures of an
organization, and the potential savings from the effective usage of IT. These evolving business
? ―Inter-enterprise integration‖ to shift from Enterprise Resource Planning (ERP) to Inter-
enterprise Resource Planning (IRP) and to resolve issues arising from the rigidity and
cost of linking systems.
? ―Intra-business functional integration‖ to unify business process beyond what packaged
? ―Self sufficiency‖ from a systems perspective, allowing adaptive, low cost iterations,
customizations and change isolations, rather than a requirement to get it right the first
? Understanding that optimization for adaptability is more important than optimizing for
cost, performance or features.
The ability to encapsulate existing systems, automating them as business processes, and letting
users collaborate via appropriate interfaces are the keys to Real Time Enterprises as illustrated.
Page 4 of 30 Real Time Enterprises – A continuous migration approach Vinod Khosla, Murugan Pal March 2002
The value of real time enterprises is in capturing the greatest value obtainable from the
systems people have created so far, and operating with the same data set that previously
There are a number of philosophical approaches we will discuss later. System integration has
been a problem. Customization has made systems static and unchangeable. We will recommend
federation, not integration of applications, configuration, not customization, and a bias
towards a more dynamic architecture. The big advantage of Web services is that it is inherently
open, perhaps even “micro-open and multi-vendor”. Web services are inherently multi-
vendor, but preserving this may take some conscious decisions. Security, authorization, and
entitlement are major issues and hence the very basics of a Web services vision have to include
a comprehensive entitlement system in multiple granularities.
Today‘s enterprise IT problems are tied to the ―islands of information‖ caused by many legacy
architectures distributed across geographies, business units and M&A subsidiaries. The
technology evolution has forced many enterprises to buy new software and hardware resources.
This has resulted in ―best in class‖, sometimes ―most in class‖, and many times ―Try, Buy,
Throw‖ environments. The challenge is to leverage existing operational systems, evaluate
―most in class‖ systems, and reuse the most meaningful. Many times the real problems are
legacy processes rather than legacy systems. In such cases, integration is a wrong approach to
achieve the necessary plasticity and adaptability. As per Gartner, 70% of all infrastructure
efforts fail or substantially miss their objectives. Large integration projects like Boeing's I-Man
portal fail to achieve the desired results 60 percent of the time, according to Giga Information 6Group analyst Julie Giera. She says companies put too much faith in technology's ability to cut costs and fail to adapt old processes to make use of the new technology. There are many
reasons that such projects fail.
Our recommendation is to achieve change in small steps. Projects should last from 90 to 180
days, whenever possible. The key to a successful migration is identifying internal champions
who realize the benefits of Real Time Enterprises and are willing to evangelize the required
efforts. The other dimension of this strategy is to offer end-user configurable processes so that
“iteration” becomes a specification methodology: getting things approximately right and
iterated by the end-user.
Page 5 of 30 Real Time Enterprises – A continuous migration approach Vinod Khosla, Murugan Pal March 2002
Optimization of systems can be achieved in four dimensions: flexibility, cost, performance, and
reliability. Our recommendation is to prioritize flexibility ahead of cost, reliability and
performance. The evolution from MIS built applications to packaged applications to today‘s
need for more adaptability is a tough challenge. The right answer is found neither in in-house
applications nor standardized third party packaged applications; it is somewhere in the middle.
A little bit of both allows business to achieve customization via configuration. These are the
―composite applications‖ using processes (pieces) from different packaged applications to
assemble (configure) a business process to match how things are really done in the enterprise.
But more than any other single factor, flexibility and adaptability are the most important
selection criteria for new technology given the rate of change in both technology and
enterprise requirements. Unfortunately, this tradeoff is seldom made, with features,
performance, and price often winning out.
Automation of processes configurable to end user needs is the key (where the end user is
defined interchangeably as a human being or a machine capable of understanding specific
semantics). This migration can be a continuum and need not be a one-step change. Evolving
architecture changes are not only demanded by technology and/or business changes, but also
by the rate of adoption. The processes themselves can be assembled from sequence of other
processes or can span into a supplier organization using different technology stacks. All these
processes operate in a world of common data, semantics, protocols, and translators that we 6call an ―information base” for a Real Time Enterprise. Jeff Jensen of Boeing identifies pulling data together into a single repository as key for their I-MAN project‘s success. This single repository must be represented as a ―virtual‖ repository — an abstraction for all coexisting data sources based on a flexible schema. Current data repositories, such as relational databases, do
not support such a model. We address some requirements under the section ―Information Base‖
(iBase), but as a first step, documents are extracted from the company's various legacy systems
and converted to XML as queries are invoked. Boeing plans eventually to move data currently 6residing in its multiple systems into this repository for easier administration.
The challenge is to retrieve data from multiple business units, sort and analyze them. The Web
services vision has much of what is needed and we recommend it. It builds on existing Web
technologies and accommodates legacy systems to a reasonable degree through a process of
node enablement we will define later. It works very well for green-field implementations, but
is also workable for real legacy environments. The goal for every enterprise betting on this
vision should be to create an “application assembly environment” where end-users can create “composite applications” or “composite Web services” suitable to their personal or corporate work style. This environment operates within the constraints of corporate business
processes, often coded into the end user programming environments by business analysts (often
as rules or objects, not as computer programs) and to limited extent by programmers to create
components and services in languages like Java and C#. Think of an Excel or Visio as the front
end for programming Web services to suit the business process modeling needs. Most non-
technical end-users can ―program‖ their application into an Excel spreadsheet within the
constraints of ―macros‖ that might be defined by sophisticated end-users or business analysts. This creates a “mass customization environment for business processes, workflow and collaboration” with sufficient support tools for administration, personalization, versioning,
upgrading, and more — allowing end-users to keep their own store of templates and a template
exchange to make them productive. The recommended changes should articulate benefits both
at macro and micro levels (i.e. each individual project must justify that their value adds
individually and as an encapsulated service within a wider scope). This could be a big challenge
given the scale of many corporations but planning this migration is a priori. For example,
changes in an ―Order Status‖ process may benefit all users within a group (micro level), but an
organization also can be impacted at a macro level by offering the process as an automated
Web service incorporating entitlement and personalization to key customers. Although this
migration can be painful in the short term, savings in maintenance costs and the ability to
Page 6 of 30 Real Time Enterprises – A continuous migration approach Vinod Khosla, Murugan Pal March 2002
enable end-users to serve themselves will be a huge benefit in a long run. The lack of an ideal
environment or too many immediate requirements should not be an excuse to keep doing
yesterday‘s thing. Every organization, to the maximum extent possible, should encapsulate the
old world through a process of ―enabling legacy nodes‖. These nodes would work in a Web services world, making iterations in functionality, the maintenance spend, the Web services,
Java/C#, UDDI, SOAP, XML, XML and schema world of Internet technologies. Many companies
who gained their knowledge from Y2K exercises and large-scale system conversions can
leverage that experience to abstract legacy functionality as encapsulated modules and expose
those modules as Web services — making the migration process easier.
This automation requires all the ―point‖ applications — CRM, ERP, supply chain systems — to
provide information and interfaces to business processes that have touch points in other
systems or by humans within the extended enterprise. EAI is a first generation solution catering
to data synchronization needs, leveraging or extending into the Web services environment by
node enablement of much of the legacy world. Many older world tools, often developed for Y2K
transformations, will be useful to this node enablement process, as will more modern tools like
portals and ―Web service publishing tools‖. Tools like Citrix can encapsulate many legacy
environments from the PC world with Web front ends of packaged client server applications.
Node enablement is only the start of the solution. Replacing the many interfaces to older
layers underneath the Web services is the key for migration. Imagine a company with 10 billing
systems, each dealing with a different customer segment or region, but each offering basically
the same services. These systems will have many thousands of interfaces that can be presented
as Web services. All of the business services offered by those systems can be migrated (possibly as common abstractions) as services that route appropriately to the underlying
systems, with every new system talking to the Web service layer. This first step makes it
possible to start building new, workflow-driven business processes without reference to
systems, but referring to services. Over time, the cost/benefit balance will gradually tilt
towards forced migration of all older interfaces. The resulting enormous benefit is the ability
to restructure, rationalize, and consolidate those 10 billing systems as appropriate. When this
type of migration (occurring over several years) happens in every area, then you are able to
completely realize the even bigger benefits of integrating new businesses and outsourced
services (such as billing, order management, and data management). The goal is composite
applications that leverage legacy applications to model the business processes, collaboration
and workflow needs of the enterprise, done in an incremental way.
The grand vision of federated Web services will happen in stages. Security and other practical
issues (like billing, SLA‘s, etc) will constrain most Web services to work initially within the
firewall as applications inside the corporation are updated to work together. The level of
granularity will test our wits and increase only slowly. Semantic and ontology differences will
limit the ideal world of possibilities into a world of very useful but far from ideal possibilities.
Think of Web services as the next leap from the Web request architecture (typical HTTP, CGI,
and Application Server requests) to a Web services request architecture. This goal is as
achievable as the transformation from client/server to Web request architectures in the late
Web Request Web Service
? Ad-hoc ? Pre-defined, well negotiated,
deterministic and reliable ? Stateless
? Support for stateful transactions ? Simple and light-weight to define
semantics ? Semantics can be defined and
embedded ? No Encapsulation for object transport
? Encapsulation supported for XML ? Undefined behavior resulting from
Schema objects Exceptions
? Infrastructure available to support
exception handling behaviors
Page 7 of 30 Real Time Enterprises – A continuous migration approach Vinod Khosla, Murugan Pal March 2002
With more than 61,000 orders processed electronically last year, Cutler-Hammer realized their
design-to-delivery vision by using Bid Manager on complex assembled products manufactured
across 26 satellite plants across the U.S. and Mexico. Bid Manager handles small but
significant details. For example, it can automatically compose labels that specify capabilities,
such as speed and power, of each motor and its components. Then it can direct the nameplate
engraver to print the label. In the past, a technician would type the nameplate information,
increasing the likelihood of error and slowing the process. Cutler-Hammer's customers, field
reps, and distributors are able to electronically design and place 95% of their orders remotely,
bypassing plant engineers. One such customer, Grove Madsen Industries, a supplier of
equipment to Las Vegas casinos, feeds their design directly to the Cutler-Hammer assembly
line 2,050 miles away.
Over time as this subset of corporate processes migrates or infiltrates into the whole
corporation and forces migration of legacy systems, the fusion of new technologies and legacy
systems with rigid processes may pose a paradigm mismatch, thus complicating the migration
process. In such instances, we recommend encapsulating those areas and abstracting them
via well-defined interfaces and behaviors.
Many will ask why an enterprise should transform into a Real Time Enterprise now, and how IT
can play a role in this process. Changing business models, evolving market, competitive
pressure, and cost benefits of leveraging existing systems drive the need for this transformation.
The balance between Return On Investment (ROI) and Risk Of Not Investing (RONI) is the key
factor in answering the ―why now‖ question.
Return On Investment (ROI) Risk Of Not Investing (RONI)
Infrastructure renovation in an evolutionary Legacy systems not matching current 6way dynamics as in Boeing example
Cost savings + real time adaptability Expenses, both cost and competitive losses
Improved responsiveness, customer loyalty Lack of real time insight, process overheads
Leveraging technological improvements Dwelling on past accomplishments
Five years ago, if someone quoted a scenario such as our Cutler-Hammer example, it would
have raised many eyebrows regarding infrastructure for communication, consensus on
semantics, and agreements on exception handling, encapsulation and implementation
mechanisms. Today, one can model a solution using HTTP (S), SOAP, XML, Rosettanet,
XMLschema, Web services, and language-independent abstracted interfaces for this seemingly
There are four golden rules and corollary implications vital to the IT transformation process:
? Plan on being wrong.
? Adopt thy partners.
? Design thy architecture.
? Implement thy solutions in steps.
Page 8 of 30 Real Time Enterprises – A continuous migration approach Vinod Khosla, Murugan Pal March 2002
Our proposal is based on our experience with the IT industry and technology evolution.
Interestingly, similar recommendations and guidelines can be found from The Forrester 13 14Report’s ―Start Using Web services Now‖and ―The Web services Payoff‖, and the HBR 12article ―Your Next IT Strategy‖.
Plan on being wrong
As previously discussed, it is important to identify the IT customers, their expectations,
behaviors, usage trends and future needs. Cutler-Hammer enlisted not only software writers,
but also experts at the plants, sales engineers, and many others to compile the requirements.
It is also important to consider various types of customers such as:
? Interacting customers via browsers, bandwidth constrained devices, and voice access
? Machine-based connectivity via XML2XML, EDI, Partner Interface Processes (PIPs), and
other client environments.
Beyond the obvious attention to users (customers) is the far more important and often
ignored fact that most requirements cannot be fully specified prior to actual use. Often 30-
50% of the cost of a new implementation goes into specification of screens in a hypothetical
environment. It is much more reasonable and feasible in the Web services world to only
attempt an approximate solution to the business process. Through actual use, the end user can
modify the ―application Web service‖, personalize it, and compose it with informal practices
and processes. These modifications can be based on the idiosyncrasies of each user or practice
group and subject to the continually evolving constraints placed on the user by the business
decisions (through a business analyst). This process leads directly to the need for a mass
customization language for business processes utilizing the Web services environment. Enabling
real use of IT because of increased relevance to each user may be the single biggest benefit of
the Web services architecture. It is important to understand the variations and evolutions (e.g.
multiple and customized versions with different granularity) that the customer community will
demand. For example, Applied Materials (AMAT) could have a generic selling process that has
to be customized specific to Intel and Motorola‘s needs and derived from the generic process.
Change and adoption are more critical in dynamic environments than getting today‘s
environment exactly right. For example, end user modifications/programming at the level of
Excel programming is very desirable in a system where end-users fine-tune the environment to
fit their needs.
? Assemble a virtual team of people (not exceeding 3 or 4) to identify and compile the
requirements. The team should represent a cross section of the customer community
along with developer and business analysts, and have a senior techno-functional person
as the moderator, as well as clearly defined deliverables, and specific timelines. Qwest
Communications gets these teams together for intense 3-4 day ―sessions‖ before a 90-
day implementation cycle commences.
? Understand existing current solutions, their gaps, and what end-users actually need.
For example, a trading broker using Excel would not want to use a browser interface
even though it gives real time capability. The right solution is to provide an Excel
interface with real time capability.
? Implement a prototype (with stubbed interfaces) that can be quickly customized for
end user needs to understand real caveats of the proposed functionality.
Page 9 of 30 Real Time Enterprises – A continuous migration approach Vinod Khosla, Murugan Pal March 2002
Adopt thy partners
The Real Time Enterprise must span the physical boundaries of an enterprise and should extend
as a virtual enterprise through the entire end-to-end value chain. For example, Colgate-
Palmolive trimmed its inventory 13% and saved $150 million by attaching its order planning 9systems to thousands of Wal-Mart and Kmart cash registers. This extension includes the
collaborative partners, suppliers, sub contractors and vendors (all referred as ―partners‖
hereafter). Each one of these partners may vary in size and have their own infrastructure for
implementing their IT systems. Certain partners may exceed the expected enterprise guidelines
and some vendors may not have any infrastructure at all. Your enterprise and your IT
organization should adopt all these partners and adapt infrastructure support to their level.
This eases the paradigm mismatch in information flow across partnering organizations and
helps transform your enterprise into a Real Time Enterprise. The goal is to create a partner
eco-system around your organization — helping them help you become successful. A successful
eco-system allows the partner set to be easily and rapidly changed based on evolving business
needs and level of sophistication. Partner collaboration is another important aspect from
design to implementation, as well as during execution. The usage model, user management,
rich security model, and transparent seamless processes across enterprise boundaries are key
factors of a collaborative application environment. Web services are a great way to implement
partner collaboration systems: they abstract the interfaces across partner systems, do not
depend on the implementation language or object types, and provide loosely coupled
connectivity (unlike RMI, CORBA or RPCs).
? Identify short-term projects; with incremental value added and well defined loosely
coupled interfaces (initial iteration).
? Identify candidate value chain companies (customer, tier 1 supplier, sub vendor, 13contractor), going 3 or 4 levels deep ―into your close partners‖, as suggested by Frank
15? Pick the right type of projects and partners for your domain needs.
? Define semantic interfaces, exceptions, and data types for information flow across the
chain (Rosettanet PIPs); provide pre-implemented plugins that can run within a
standard runtime (servlets or Web server plugins), if necessary.
? Agree on periodic iteration updates and be willing to spare your own resources to help
implement partner solutions.
? Include partners when talking to solution providers and create usage scenarios and
proof-of-concept environments based on real partner needs.
Design thy architecture
Once you identify your customers‘ needs and connectivity requirements to your partners, the
right architecture must be designed. Your architecture‘s mindset, belief, and vision should
cater to the short and long term goals of your enterprise. The architecture should be based on:
? Open standards and multiple vendors.
? Flexible, loosely decoupled interfaces considering rapid changes and evolutions of
? A production system that can manage, monitor, and support versioning of the Web
services life cycle.
? Ability to leverage existing systems and reduce cost.
? Flexibility as more critical than optimization for cost, performance or features.
Page 10 of 30 Real Time Enterprises – A continuous migration approach Vinod Khosla, Murugan Pal March 2002