One of the earliest blogs I posted was about service-oriented architecture (SOA) [Great Article on Service-Oriented Architecture]. I published it after having met with Accenture CTO, Donald Rippert, who, the day before, had published an op-ed piece in the Financial Times ["The building blocks of a simpler future are in place," by Donald J. Rippert, 10 May 2006] In that article, Rippert wrote:
"Imagine a future where IT systems are not created by computer analysts speaking the languages of Java and C but instead by business managers speaking the languages of supply chain, customer service or product development. It is a future made possible by Service Oriented Architecture (SOA) -- an evolution in the way enterprise IT systems can be built . ... Today, businesses either build and maintain custom applications (such as invoice processing or personnel systems) at great cost, or they conform their business processes and organisation to pre-packaged software applications. Both of these approaches have helped automate business processes but have become operationally and intellectually threadbare in recent years."
Remember, Rippert was calling "intellectually threadbare" many of the systems we still see in place today -- five years later. He continued:
"The usually pejorative phrase 'legacy system' was coined to describe old, custom applications that have grown hard to maintain and almost impossible to replace, whereas packaged software brings its own set of challenges since modifying it to meet the needs of a particular business is expensive and risky. ... With an SOA, business applications are constructed of independent, reusable, interoperable services that can be reconfigured without vast amounts of technical labour. ... The fundamental building blocks of an SOA are web services. ... An SOA is a collection of web services brought together to accomplish business tasks (checking a customer's credit, for example, or generating an invoice). ... Because these services are accessed through a standard, they provide unprecedented flexibility: business processes can be added or altered quickly; software applications can be integrated easily."
Rippert wrote this before the term "cloud computing" became widely accepted in the IT world. Even though cloud computing is better understood than it was back in 2006, some companies have been reluctant to embrace it. That, however, is another story. Rippert concluded his piece this way:
"Because the services can interact with systems outside a single organisation, they provide the ability for companies to collaborate with customers and suppliers. And because services are simpler than hard-wired applications, they lower maintenance costs. ... Heard it all before? You probably have. People have been talking about taking the programmer out of programming for decades. So what is different now? In a word, standards are what set SOA apart from previous generations of integration technologies, which were largely proprietary to each vendor. The standards behind SOA have been around in some form for a few years, but they are just reaching maturity. ... Perhaps the most compelling impact of SOA is how it stands to rewrite the rules on IT governance and organisational structure. In most organisations, IT managers tend to be linked with the specific applications they support. Because SOA delivers the promise of solutions that transcend lines of businesses -- and the organisations themselves -- IT managers, newly decoupled from applications they manage, will have a broader view of the potential they can deliver. Once IT speaks the same language as business, it will be primed to design services that help companies bring distinctive capabilities, products and services to market quickly."
Five years on, IT is now speaking the same language as business. It's an exciting time to be involved in supply chain optimization. The editorial staff at Supply Chain Digest also believes that service-oriented architecture is a game changer. It writes, "Service oriented architecture can enable a Warehouse Management System to act more like a proxy DC supervisor." ["The Emerging Event-Driven WMS Model," 8 December 2010]. The SCD staffers continue:
"Service Oriented Architecture, or SOA, is changing the way a Warehouse Management System (WMS) will run, offering users new levels of efficiency through increased system intelligence. Though the topic of SOA often puts operations managers to sleep the reality is an SOA-based WMS, if properly developed, can result in a system that is based on thousands of discrete pieces of functionality or services, rather than different specific applications (receiving, putaway, etc.) that may be linked but that must behave in limited and usually inflexible ways. With SOA, these capabilities can be linked together to create a more flexible and intelligent WMS, one that can react to events in the DC to further optimize distribution performance."
SCD staffers write about "increased system intelligence." I like to think about such frameworks as "sense, think/learn, and act" systems. Warehouse management is an ideal fit for such a system. The article explains why:
"In distribution and broader logistics operations, processes involve nothing more than a sequence of events, decisions and actions. These events can be normal, expected, predictable -- but at times they can be unusual, unexpected, variable, or random. Using the event-driven WMS model, it is the occurrence of specific events, rather than application code, that triggers functionality and subsequent action. For example, consider a cross dock application, as illustrated in the figures below. Let's suppose that an inbound shipment was planned for cross docking, but at unload time the outbound door or trailer is not yet available. In a traditional system, this would likely require several manual decisions and processing –- a supervisor observing the lack of the available door, [directs] dock workers to put the inbound goods in a temporary staging location, additional directions [are required] when the door becomes available to start loading the outbound truck, all these activities primarily conducted outside the logistics system itself."
A distribution center equipped with an SOA-based Warehouse Management System can let the system make many of the routine, event-driven decisions that must be made -- freeing the supervisor to deal with other pressing challenges. The article continues:
"The event-driven system approaches this situation in an entirely different way (see graphic below). The WMS scans the dock environment when the inbound truck arrives and recognizes the outbound trailer is not yet available, triggering system direction (via handheld terminal or workstation) to move the goods to an available staging area. In the middle of this process, when the event of the outbound door becoming available occurs, the system recognizes this changing condition and both directs goods from the truck into the outbound trailer, as well as creates tasks to move the staged goods. This is a fundamentally different way of handing WMS system development, and has enormous potential benefits for improving efficiencies and optimizing logistics flows."
Enterra Solutions' Supply Chain Optimization Platform uses Complex Event Processing in its sense, think/learn, and act framework; so you can understand why I'm such a proponent of these kinds of systems. The article admits that "there are many existing functional parallels to this cross docking example" and it also admits that results can be achieved other ways. But, SCD staffers insist, none of the other methods are as robust. They explain:
"For example, in many systems when an order picker is unable to pick an item due to a discrepancy between what the WMS system thinks is the inventory quantity in a location and what the operator finds, that 'event' will trigger a task to do a cycle count of the location. However, to achieve this effect, the WMS provider must hard code this logic into the picking application. If there are multiple scenarios that could trigger a cycle count, this functionality is achieved by replicating the logic in different application areas, and/or writing a long series of 'if/then' statements into a specific piece of code. This significantly limits flexibility to make future changes, and is an approach that can be used for a relatively small number of event-action sequences. In the true event-driven approach, the cycle count function is not tied to a specific piece of application code, but is linked to or triggered by any number of different events. It is independent of a specific hard-coded application, and therefore can be easily augmented or changed based on new processing requirements. This flexibility makes it easy to craft event-driven processing in an unlimited number of logistics scenarios."
In a number of past posts, I've discussed how supply chain analysts believe that system flexibility is one of the characteristics that is going to differentiate great supply chains from good ones. As the above example demonstrates, an SOA-based WMS is inherently more flexible than other types of systems. To further make their point, SCD staffers provide "other examples of the way an event driven-system might intelligently handle specific exceptions or conditions in the logistics flow." They write:
"The system could look forward into downstream packing and consolidation areas for congestion, and reroute or delay releasing work until the 'event' of a work area freeing up occurs. Perhaps the DC is using an automated material handling system, and a photo eye on the conveyor becomes blocked or disabled. The WMS recognizes this unexpected event, triggering the use of alternate paths along the system, or a more manual mode of order consolidation. At a broader logistics system level, the 'event' of an exception notification from a truckload or LTL carrier that a shipment is going to be late might trigger that a small quantity of the item be picked and shipped via overnight to ensure the customer has enough of the needed item until the scheduled shipment arrives. This dynamic Event-Decision-Action model, conforming so naturally to the way logistics functions really operate, is possible with the event-driven logistics system because it occurs without any specific logistics application program getting involved. The event-driven approach in turn is really dependent on an SOA-based WMS architecture to enable it."
It simply makes good sense to allow an intelligent system to make routine decisions about common problems. The SCD staffers assert, "This event-driven model will also improve the ability of the WMS to work with different sorts of DC automation systems, and increasingly allow the WMS to act like a 'proxy supervisor' that monitors and manages what is happening in the DC at even greater levels than today's vary sophisticated WMS solutions." They point out that the more information a system obtains the smarter the decisions it can make. They explain:
"As adoption of RFID continues, the event-driven approach becomes even more important, as the WMS must react to RFID reads wherever that are encountered, rather than to expected bar code scans within a specific application. SOA may be mostly in the realm of the IT wizards, but it has already changed how WMS and other logistics applications are deployed and used today – and will have an even more profound impact in the coming years. Logistics managers need to get at least a little smart about SOA and what it can do for them in WMS."
Great articles like the ones written by Donald Rippert the Supply Chain Digest editorial staff go a long ways towards making logistics managers smarter about service-oriented architectures and sense-think/learn-act systems.