The World of Integration--Part Three: Innovators and Innovations
on May 18, 2016
Integration is an evolving field, blending data, things and people across systems, mobile devices and the web. Over thirty systems and some important innovations are included in this article.
Full Article Below -
We continue here from Part Two: Market Segments, where we revised and defined the
integration market with an orientation towards supply chain applications.
Innovators and Innovations
This article is excerpted from the report:
“The World of Integration”
available for download here.
The world of integration is about sharing data and processes. So today’s innovations should take in any kind of data and transform it—EDI, spreadsheets, big data, and so on.
With so much connectivity and sharing of processes, moving data may not always be the way to go. Data movement was the standard approach: shipping files between parties. However, in the web world with REST architecture and federated data models that enable sharing and leveraging of each other’s web pages, processes and data can be shared without data movement.
Here are some interesting examples:
Federated data models—here, the master data, metadata, and physical data are distributed. But there is a unifying data schema that allows for integration. The solution maps schemas and yet allows for autonomy of each database and/or organization. Stone Bond, an up and coming integration platform, has an innovation they call Enterprise Master Service (EMS) that provides a federated data model.1
Unifying B2B and A2A—as we mentioned earlier, many tech companies provide both of these. But only a few have fully harmonized them into one engine. EXTOL’s EBI or Informatica, as examples, have tools for data transformation. These are engines which pull almost any data format, translates (one protocol to another a la EDI/B2B), and transforms it (as in data cleaning). Using a rules engine, it then moves the data to its destination—export service and mailbox for B2B or, through an enterprise service bus, to an application.
Multi-party processing—another challenge in integration is how to harmonize and handle evolution and changes in data and processes, yet at the same time maintain autonomy. The requirement here is that each entity still needs to operate within the rules of their own data model, yet for certain processes they need to share data and perform workflows and transactions.2
An example of how this is done is One Network’s approach: it builds on open Java standards and employs an approach that is called mixins. Rather than taking the reader through a long obtuse technical PhD discussion here, the point is that this approach allows entities to work independently, but to also inherit data models when working on a joint task. So beyond translation, a process may want to adopt the partner’s rules or model for a process or task.
For example, think of the contract manufacturer (CM) who stores components for several customers. The CM has a parts master in a certain format. Each customer has a different parts master, so there could be a dozen parts masters floating around. The CM can’t change their model for each customer. And the customers surely are not going to change their model. Now add to that specific accounting rules or inventory liability, status, stocking rules, and allocation. When and how inventory would be considered available to promise is very important, since supply chain owners/dominators expect highly integrated digital supply chains and rapid response.
As we stated, it is not only data models that can be inherited, but processes as well. One Network accomplishes this by leveraging a new kind of Extensible Master Data Management (E-MDM) and Common Transaction Management (E-CTM). One other fascinating innovation is their process inheritance architecture, which we understand was the raison d’être of One Network from their inception.
Applications as a Service—visualize the world of things. As we know, each thing may have many applications that govern its use, special digital features or upgrades, apps for communicating service issues, apps for connecting partners, and so on—in other words, a thing will have many apps associated with it. But rather than having users find all these apps, imagine that the thing knows and can access (or download) the appropriate app. PTC’s ThingWorx/ThingBrowser, though in beta, has been demonstrated in public and has several big companies in the program. The concept is that each thing has its identity, of course, on the internet: its own URL. By scanning or clicking on the URL, all the associated data and apps appropriate to that thing can be called. For an integrator, this allows re-use as well as more rapid development.
Purpose-built—another useful approach is by platform providers who do focus in on communities of needs, use cases and/or industries. Rather than a buyer of tools having to build or find their own libraries, these providers have and are constantly adding to a library of relevant analytics, use cases, end-point connections and so on. An example here is Savi. Their platform has design and integration tools for establishing a sensor network, integration to applications, as well as creating analytics that extend logistics applications.
In general, with open source, purpose-built libraries, or BPEs, users are in a discovery and adapt mode, rather than development. It does provide a real leg up (which translates into speed to completion) that can assist today’s organizations (who already have so much on their plate).
Today the market looks a bit different than it did a few years back. Focusing on applications vs. trading partners, and tools vs. solutions, the emerging unified platform is becoming the choice in the market (See Figure 1).
In Figure 1 we position some of the players. This chart is not meant to show superior placement, but rather how players are positioned and their main focus.
Figure 1: Position of Players
Firms continue to invest in deepening their solutions, adding more end points and making the solutions more open and easy to adopt. Expect to see a lot more mergers and acquisitions as companies consolidate their approach to selling to the CIO. Over the next few years the ability to differentiate one solution from another will shrink as solutions platforms grow.
Next we show a short list of options available in the market. Integration solutions, de facto, place the user in the situation where they have to think about the overall architecture and data models.
Another issue is that some of these companies are tools providers only. Though integration tools for A2A are mostly the domain of IT, tools vendors like Red Hat and WSO2, as well as Talend and Jitterbit are totally the domain of developers.
The table here, Integration Market, is only a short list of players in the market. As you can clearly see, the players are expansive. That is, they are trying to fulfill the many functional areas required for integration today. Also note the platform/cloud versions vs. on premise.
Table 1: Example Players
IoT and RFID Middleware
IoT is yet another data type that is truly data in motion. Most integration firms treat IoT as a distinct capability and module. They do this because IoT data may not come in a digital format. A segment of the integration/middleware market has been dedicated to RFID, sensor, and device-level data for decades now.3 However, we see many integration firms and application solutions finding ways to absorb IoT into their approaches. Since we have written a great deal on this topic already, we direct the reader to the IoT library and the report: RFID or IoT.
Integration is inescapable. The question is how to lessen the burden now as well as find a solution that is modern and will keep up with all the new end-points. IoT, mobile, and so on represent huge challenges for IT departments.
And for the supply chain professionals whose world of trading partners keeps increasing, we are moving from a world of transaction messaging to process sharing. This means that B2B integration solutions can’t be arm’s-length, EDI alone, in order to meet today’s needs. Hence, solutions that reflect process, data, and applications in a harmonious way should be of vital interest.