Is RFID Application Oriented Networking’s Killer App?

RFID is a widely hyped next generation technology for wirelessly “scanning” items.  AON, or Application Oriented Networking, is widely hyped next generation technology for managing middleware messages.  Put the two technologies together and not only do you have a whole lot of hype, but you may also have a match made in technology heaven, one which ends up making both technologies better.

AONs Second Generation
I have written about AON, or what I prefer to call Message Aware Networking, quite a bit in the past.  AON is at the core of a revolution in software architectures that is moving a significant amount of functionality out of applications and pushing it down into the network.  The first generation of AON devices act as simple gatekeepers/routers and are limited to pre-processing messages that they then deliver to applications.  Datapower, which was recently sold to IBM, was the leading player in this space.  While the space is still quite young, already a 2nd generation of AON devices is under development.  These devices will not simply serve as intermediaries, but they will actually become an integral part of an application and an embedded part of a business process.  By embracing the concepts of distribution and virtualization, as well as standards such as BPEL, these 2nd generation devices will turn the network into an integrated part of a given application and in doing so should enable significant new functionalities while reducing costs and improving performance.

RFID: Almost Ready For Prime Time
RFID seems worlds away from AON, but RFID’s future may be more closely tied to AON than one might at first suspect.  Much of the attention around RFID has focused on the core hardware technologies required to make RFID a reality.  Not only are there very difficult RF-related physics problems that must be solved, but the costs of silicon-based RFID “tags” must be reduced in order to make RFID economically viable.  Now however, with companies such as Impinj shipping cost effective RFID 2.0 tags in volume and with retailers like Wal-Mart mandating RFID adoption, it appears as if RFID is finally poised for rapid growth.

Unfortunately for RFID, if anyone ever does install the hardware, the first thing they are likely to do is crash all of their software.  That’s because a fully instrumented RFID installation is bound to kick off thousands, even millions of messages a day and processing all these messages is likely to bring even the most robust computer system to its knees.  To solve these performance issues companies will either have invest a fortune in significantly upgrading their core logistics systems or they will have to figure out another way to get the job done.

That other way may be a new class of AON-based devices that are specifically focused on RFID.  Such devices could not only locally process millions of RFID messages, thus greatly reducing the overall “load” generated by RFID systems, but they could also offer new capabilities, such as the ability to efficiently “multiplex” messages across an entire supply/demand chain.  For example, AON-based RFID devices could use cached lists of expected shipments to identify in real time when specific shipments failed to arrive or arrive incomplete.  These devices could also simultaneously notify all of the different supply chain members about missed shipments or out-of-stock events.  While a retailer’s or distributor’s centralized computer systems could no doubt do all of this, constantly hitting those servers with updates on every single SKU from every single store, warehouse, truck, etc. would not only be wildly inefficient, it may also distract those systems from more important higher level tasks.  In this way, AON-based RFID devices provide a relatively cheap way to “off load” to the network the processing of routine RFID related messages.

Startups To the Rescue
But don’t take my word for it, just look at the start-ups that are springing up to take advantage of this potential opportunity, the most interesting of which I have seen to date is a company called Omnitrol.  Omnitrol has built a very intriguing RFID focused AON-based device.  Unlike many 1st generation AON players which used Intel-based Linux “pizza boxes” as their hardware platform, Omnitrol has engineered a sophisticated hardware platform that uses Gigabit Ethernet to interconnect multiple RISC processors allowing for greatly increased message processing capabilities within a single box and the ability to seamlessly interconnect multiple boxes.  Such an architecture not only guarantees that Omnitrol should be able to process massive message loads, but it also offers a great example of how 2nd generation of AON devices are likely to use much more sophisticated hardware platforms that look more like network devices than enterprise servers.

With Omnitrol’s AON-based RFID devices in place, retailers can keep a large portion of their RFID processing “local”, i.e. confined to the actual store or distribution center where it takes place.  Only critical events and/or summaries of activities need be forwarded to the central system (which in most big retailers remains an IBM mainframe).  The RFID devices have also been designed from the ground up to be compliant with the various EPCGlobal RFID related standards and have the ports needed to accommodate a variety of RFID readers.  In this way, Omnitrol has customized both the hardware and software of its device so that it is RFID centric.

In addition to Omnitrol another start-up focused on this space is Reva Systems.  Reva is already shipping a device that is focused on managing RFID networks and clearly has designs on providing more sophisticated “application aware” capabilities over time.

Despite their promise, the success of these RFID-focused AON start-ups heavily depends on how quickly large companies embrace and roll-out RFID.  While recent signs seem to indicate that adoption should dramatically accelerate in the next year or two, this is still a major risk.

A Sea Change?
Despite this risk, both companies are great examples of how Application Oriented Networking is rapidly evolving.  Perhaps the most interesting trend that these devices underscore is the ability to tailor an AON device, at both a hardware and software level, to a specific application.  This tailoring raises the interesting question as to whether or not we will see an explosion of application specific AON devices, each targeting a different enterprise application.  Such an explosion may not only provide a whole new class of investment opportunities, but it may also significantly challenge the dominance of general purpose computing platforms in enterprise software.  After all, if you move enterprise software into the network, why won’t the devices used to process that software look a lot more like network devices than enterprise severs?   Of course, we are a long way off from such a change, but it is interesting to contemplate and once again underscores that Application Oriented Networking is one of the most important and potentially disruptive trends within the software industry.