Quality Management: Make Interoperability Happen

November 1, 2005
/ Print / Reprints /
/ Text Size+

Interoperability has become a buzzword. Everyone in manufacturing has heard it. But just what is it?

Interoperability is the ability of two system components to communicate correctly and completely with components from any vendor worldwide with minimal cost to either component user or component vendor.

Each part of this definition is important:

"...two system components..." Even if one component can broadcast to multiple recipients, such as on a network, interoperability can always be narrowed down to two components.

"...system components..." These components can be of any size, as long as they are distinct-or separable-components.

"...communicate correctly..." There must not be any encoding or decoding errors.

"...communicate completely..." The interface language must allow each component to express anything and everything that both vendors and users agree needs to be said over the interface.

"...with components from any vendor worldwide..." The system is not really interoperable if the Europeans have interoperability among themselves but not with the North Americans. A critical mass of vendors worldwide supporting the standard is essential, particularly in this age of global corporations.

"...with minimal cost to either user or component vendor..." The system is not interoperable if it takes a team of five software developers two weeks to get the two components communicating correctly and completely. The system also is not interoperable if the component vendor has to spend lots of money needlessly maintaining the various interface language versions as well as having to spend money for training and support for the multiple versions.

A precise definition of interoperability makes it less likely that someone can falsely make a claim of interoperability. For example, say an OEM requires that all suppliers write and read in a single computer-aided design (CAD) vendor's native CAD format. The OEM may claim that this solves the interoperability problem, however, interoperability has not been achieved, because additional burden has been placed on suppliers who have to support the sometimes different CAD formats required by all the other OEMs they work with.

By definition, interoperability has not been achieved because one of the users-the supplier-is not expending "minimal effort." Of course, the OEM will end up paying for it because its suppliers need to add that cost to the cost of their products and services.

Interoperability Problems

To determine if a company has an interoperability problem, take a look at a few scenarios.

• An automobile manufacturer needs to move model operations to another location. The new location has a different software application for the same function, so all software programs must be rewritten, potentially resulting in millions of dollars lost because of a product launch delay, additional labor costs and loss of product quality.

• An automobile supplier wants to purchase an inspection probe from another vendor and integrate it into his system. He cannot because his current coordinate measuring machine (CMM) will not interface with the other vendor's probe. His current CMM vendor offers to modify the system interface, but for the quoted price of integration, the supplier can almost buy a new CMM.

• An aerospace corporation has spent lots of money training CMM operators on a particular inspection software package. They now want to use a new CMM. They cannot unless they get new inspection software from the new CMM vendor. Heavy retraining and reprogramming costs inhibit freedom of choice.

• An inspection software vendor spends a large percentage of his or her programming costs maintaining version compatibility with proprietary interface languages.

These scenarios are, of course, biased towards dimensional metrology. None-theless, the interoperability problem is essentially the same in the automotive, aerospace, electronics or medical sectors, or in the machining, design or quality control functions.

The question to ask is, are there any costs-hidden or obvious-in the operation that are related to the kinds of problems illustrated in the scenarios?

Who suffers from INTEROPERABILITY?

The cost of the interoperability problem is currently known most through anecdotes and it seems to be high and persistent. Even so, there have been several formal studies on the cost of the interoperability problem. They have concluded that interoperability costs are roughly:

• $1 billion for the automotive sector for engineering and business data

• $5 billion for the discrete manufacturing supply chain

• $22 to $59 billion for inadequate software testing infrastructure

• $15 billion for the capital equipment sector

It should come as no surprise that everyone suffers from the interoperability problem: OEMs, suppliers, vendors and customers. There are some interesting twists to the distribution of the suffering, however. The problem occurs broadly among users, but large users seem to suffer the most. This is because large users-through acquisitions, preferences of employees, preponderance of legacy systems and simply the size of their corporations-typically have multiple products from multiple vendors that perform the same function. Furthermore, these products are typically not interoperable.

Smaller users can sometimes require a single vendor's products throughout the corporation, which may provide a temporary solution. System users cannot easily and cheaply build systems from components. Users cannot achieve best-in-class because the lack of interoperability makes it more difficult to exploit technology changes. The manual reinterpretation and re-entry of data is costly and error-prone. Excessive operator/

programmer training also is costly. New technologies either cannot be introduced, are delayed or cost too much.

The interoperability problem also occurs broadly among vendors, but small vendors suffer the most. Large vendors can sometimes dominate the market so that their proprietary interface languages become de facto standards. If large vendors define the interface language, small vendors are always playing catch-up, increasing their time-to-market. Furthermore, they may find it hard to differentiate their products by the introduction of new technology because the larger vendors have a lock on the interface language definition.

Finally, the dominance of single vendors reduces competition, which invariably increases costs to the users. Component vendors have to support many different proprietary interface languages, sometimes costing them up to 60% of their software development budget.

In the end, no matter whether users or vendors suffer, higher costs are passed on to the consumers in the form of higher prices for products and services. Solving the interoperability problem can save time, money and resources. What is more, there will be more time and energy to devote to other problems.

Response Options

Everyone has experienced the interoperability problem, even if it is as simple as being unable to paste a graphic into a document. Most of the so-called solutions to the interoperability problem are not really solutions at all, but either get the system working, at substantial cost, or push the problem onto someone else in the supply chain. Responses to the interoperability problem fall into three categories: point-to-point, single vendor and common language.

The point-to-point response is not really a solution because it cannot be done with minimal cost to either user or component vendor, and it is short-

sighted, costly, error prone and stagnant.

The single-vendor response can provide a tentative relief from incompatible systems, but it has problems. It can encourage lack of competition, which can raise costs-cost savings depend on the stability of the single vendor-and one is restricted in choosing best-in-class solutions. The single-vendor response is costly, particularly because it generally pushes the problem on to someone else. On the other hand, the common language response actually solves the interoperability problem.

The common language response reduces measurement errors caused by translation mistakes. It requires less time and effort of both user and vendor and provides greater freedom for successful acquisitions and best-in-class preferences. There seems to be little concrete data that the cost of the time and effort required to enable a common interface language actually is much less than the cost of the point-to-point or single vendor responses. In spite of this, consider the following analogy.

Does having a single common interface solution for plug-in cards, printers, network connections and wireless links enable productive, inexpensive use of the computer? Were things better prior to the emergence of these various standards?

Of course, in several cases there are multiple competing standards, for example, wireless systems and plug-in printed circuit board standards, but most will agree that that situation is much better than having a slew of proprietary standards for each vendor's PC board or wireless networking system.

The common interface language response is the most satisfying solution to the interoperability problem.


Reasons why the interoperability problem should be solved and why the common interface language solution is the only true solution that has been discussed, but what does it cost to realize such a solution? How can the industry expect to get competing users and vendors worldwide to agree to a single common language? Timeliness and concurrency are essential ingredients.

The interoperability solution requires worldwide support and concurrent development of the following elements:

• Interfaces. Identify appropriate interfaces, identify existing interface standards, and identify gaps and overlaps.

• Interface languages. Develop timely, unambiguous, sufficiently functional and consensus-based languages.

• Implementations. Create implementations that are timely, compliant, fully functional, interoperable and performed by a critical mass of vendors worldwide.

• Tests. Product must pass conformance and interoperability tests before purchase.

The first step is to identify the key interfaces. All the key stakeholders worldwide need to participate in this identification, or at least be in agreement with the conclusion. Then the industry needs to identify what languages currently exist worldwide for those interfaces, considering both proprietary and non-proprietary, closed and open languages.

If additional or new interface languages must be developed, develop them. Vendors, typically, are the best ones to do this development under user supervision. Make sure that the interface languages are correct and complete as they are being developed, that they have accurate syntax and well-defined words and grammar, and that they define all the functionality required by both users and vendors. There also must be a process in place to make changes, additions and improvement to the standard as technologies change. Get agreement from users worldwide to support a single interface language, by specifying the language in their purchase requirements. Only this step will discipline the vendors to commit resources to the development of the standard, which only they can do. These resource commitments can be somewhat costly.

At the same time, vendors worldwide need to be participating in the implementation of the interface language in prototype versions of their products. This participation must be mandated by a critical mass of users worldwide, otherwise vendors will typically not participate. There must be a process whereby the knowledge gained about the standard by doing implementations is fed back to the standards writing committee. It also is good if the language writing committee is a small group, either a single vendor or small group of vendors, so that things move quickly.

Concurrently, tests must be developed and run against all implementations. At least two types of tests are required, conformance and interoperability tests. Conformance tests are tests of one implementation, on either side of the interface, against test utility software using carefully defined test cases. The test cases must cover as much of the anticipated functionality as is possible and reasonable.

Interoperability tests are between two implementations, one on each side of the interface. Public demonstration events work well because such events act as a motivator for vendor implementers to work diligently on their implementation.

Organizing, defining and arranging these tests and generally enabling and ensuring successful timely execution of the standards development process outlined above has turned out to be an excellent role for an organization like the National Institute of Standards and Technology (NIST). It has played this role for many different interface language standards, including the I++ DME and DML standards, which have been popular and successful in dimensional metrology systems.

NIST has observed that neither defining interface languages by committee works because it takes too long to develop a working interface language; nor does supporting an open, proprietary interface language because all the other vendors are frustrated and hamstrung by the single-vendor control of the language definition.

Having a single vendor, or small group of vendors, define the interface language while the rest of the community does implementation and makes input to the interface language definition seems to work as long as the functionality requirements in the interface language are not inherently complex.

Interoperability is a goal worth seeking and, if the correct process is employed, there is hope in attaining the goal. The common interface language approach is the only one that really solves the interoperability problem.

But do the benefits outweigh the costs? If considering the analogy of the personal computer, surely the benefits will outweigh the cost.

Finally, is a partial solution OK? Yes, as long as everyone recognizes that there is still more work to do to gain more benefit.

Interoperability is worth the effort, but it is an effort that requires some level of commitment from everyone. Q

John Horst is an electronics engineer for the Intelligent Systems Division, Manufacturing Engineering Laboratory of the National Institute of Standards and Technology (Gaithersburg, MD). He can be reached at (301) 975-3430 or john.horst@nist.gov.

Quality Tech tips

• Solving the interoperability problem can save time, money and resources.

• Responses to the interoperability problem fall into three categories: point-to-point, single vendor and common language.

• The common interface language response is the most satisfying solution to the interoperability problem.

• The interoperability solution requires worldwide support and concurrent development of the following elements: interfaces, interface languages, implementations and tests.

Did you enjoy this article? Click here to subscribe to Quality Magazine. 

You must login or register in order to post a comment.




Charles J. Hellier has been active in the technology of nondestructive testing and related quality and inspection fields since 1957. Here he talks with Quality's managing editor, Michelle Bangert, about the importance of training.
More Podcasts

Quality Magazine


2015 January

Check out the January 2015 edition of Quality Magazine for features!

Table Of Contents Subscribe

The Skills Gap

What is the key to solving the so-called skills gap in the quality industry?
View Results Poll Archive

Clear Seas Research

qcast_ClearSeas_logo.gifWith access to over one million professionals and more than 60 industry-specific publications,Clear Seas Research offers relevant insights from those who know your industry best. Let us customize a market research solution that exceeds your marketing goals.


facebook_40.png twitter_40px.png  youtube_40px.pnglinkedin_40px.png