Sep 25, 2008A new study from Dulles, Va., RFID systems integrator ODIN Technologies shows that passive EPC Gen 2 RFID tags and interrogators can be effective for tracking IT assets, but that no single RFID tag or interrogator model is appropriate for all use environments. The IT Asset Tracking Benchmark is the 12th installment of ODIN's RFID Benchmark series, which aims to independently evaluate RFID equipment performance.
To carry out its study, ODIN attached the tags to rack and blade servers, desktop and laptop computers, inkjet printers and other types of electronic devices. The study found that the proper passive RFID system could significantly streamline IT asset-management processes. For example, a rack of 40 servers could be inventoried in less than 30 seconds, while 19 electronic devices could be inventoried in an office cubicle five times faster using RFID than manual methods. The study also found that RFID could be employed to identify the memory cards installed in a server.
For some time, RFID has been used to track IT assets, from laptops to network routers. But companies have typically had to utilize active RFID because passive UHF tags did not perform well on IT assets containing metal, which can interfere with RF waves. For instance, Cognizant, a provider of IT, consulting and business process outsourcing services, had deployed a passive RFID system to track laptops, but the company found that the tags were not being reliably read because uncontrollable factors—such as metal objects being carried in bags along with the laptops—were causing RF interference with the tags. Therefore, it opted to switch to active tags from RF Code (see RF Code Announces Platform for Tracking IT Assets).
However, a growing number of metal-mount tags, specifically designed to mitigate metal's interference, have come on the market and are offering companies new options (see HP Launches Service That Applies Tags to Items It Ships and IBM Offering IT Asset-Tracking Solution). In its new study, ODIN evaluated seven metal-mount tags compliant with EPC Gen 2 from four vendors, to determine whether passive UHF is a now a viable technology for tracking IT assets.
"There's been a large increase of interest in IT asset tracking using passive technology. But no one had really done any testing, to explore what's possible and not possible with the technology," says Adam Bennett, a co-author of the benchmark study. "We've noticed, in the last year, a real increase in knowledge of RFID and IT asset tracking, and several different vendors—both IT hardware vendors and producers of RFID tags and integrators—have all been looking into this market." What's stimulated that interest, he says, is the increased selection of metal-mount tags. "People are starting to look at ways to use these tags."
The tags tested include Omni-ID's Prox and Flex tags, Confidex's Steelwave, Steelwave Micro and Halo, TROI's MMT-3004, and Sontec's C0101. ODIN selected the tags based on a number of factors, including physical dimensions, performance and commercial availability.
Multiple interrogators were employed during testing. The tests in the study set out to determine sensitivity (how a tag will respond if reader power is decreased); orientation sensitivity (which tags require a handheld reader to be at a particular angle); distance (the optimal range to read tags with a handheld reader); material dependency (which tags work on both metal and plastic equipment cases); and blade server inventory (which tags perform best when placed on blade servers).
ODIN initiated the study in August, and spent more than a month testing tags in various use-case scenarios. The systems integrator then partnered with a company, which Bennett asked not to be identified, and conducted tests in that company's data center—a fairly large, secure facility housing more than 500 devices, including servers, routers and other IT equipment. ODIN attached the tags to rack servers, blade servers, RAM cards, desktop and laptop computers, inkjet printers, CRT and LCD monitors, IP phones, calculators and spectrum analyzers. It then tested whether RFID could be used to inventory many of these electronic devices spread throughout an office cubicle, as well as how RFID compared with manual inventory, and whether a stack of IT assets could be interrogated while passing through a doorway.
The idea to test the use of RFID for tracking individual sticks of RAM came from the partner company that had offered up its data center to conduct the testing, Bennett says. "They brought the idea to us," he notes, adding that by accurately tracking RAM, companies can know exactly which RAM cards are in which servers, whether those cards are older or newer, whether upgrades or maintenance is necessary, and other vital inventory-management data companies require to ensure servers have sufficient memory to support their business operations. By affixing RFID tags to individual RAM sticks, companies need not manually check them, Bennett says. "If you want to see what RAM is in which server," he explains, "it is kind of a pain because you have to pull each one out of the server, and then you run the risk of damaging them."
RAM cards measure only a few inches in size. ODIN had to use very small tags, Bennett says, and only two of the seven tag models tested fit the bill. One was 33 millimeters by 10 millimeters by 4 millimeters (1.3 inches by 0.4 inch by 0.2 inch), the other 38 millimeters by 13 millimeters by 3 millimeters (1.5 inches by 0.5 inch by 0.1 inch). The tags were affixed to the bottom side of the RAM sticks, in an area that is minimally visible when the server's cover is open.
The RAM sticks used during the test included Hynix's 2-gigabyte RAM card with a heat sink, and Super Talent's 1-gigabyte RAM card, which has no heat sink. (Heat sinks, which are metal, absorb and dissipate heat from a RAM card.) The tested RAM cards also contain a removable, protective plastic covering.
"We needed to check the RFID tags' performance, to see if you could pull a signal off these tags," Bennett says. "And we didn't quite know, going into this." Since some metal-mount tags must be attached to a significant amount of metal in order to operate, not all will function with RAM cards that lack heat sinks. What's more, because metal-mount tags are sensitive to their orientation in relation to interrogators, the tags ideally need to be tested in multiple positions to determine the optimal placement. But because there was only one location where the tags could be affixed on the RAM sticks, multiple placement options were not feasible.
During the five-second test, ODIN documented the number of redundant reads. If a high number were documented, the test would prove that tag able to work consistently. If there were only a few redundant reads, however, chances are the tag would not be a consistent—and, thus, viable—option.
Of the two tags, one consistently outperformed the other (ODIN has asked that RFID Journal not identify individual tags' performance, though that data is available in the report). The poor-performing tag was unable to function when placed upon RAM without a heat sink, because it needs to be attached to a metallic object to consistently provide reads. In contrast, the other was able to perform adequately on bare RAM since it need not be attached to a metallic object to function properly.
According to Bennett, the test showed that sweeping the handheld over the RAM perpendicularly was more effective than passing it over in a parallel manner, and reads were better when the plastic cover was closed over the RAM. As expected, he adds, no reads were possible when the server's metal case was closed.
"A lot of this gets down to, in this case, where the tags are positioned [and sending RF signals], in the gaps between the [sticks] of RAM," Bennett says. "That will certainly cause interesting reflections." When tags give off signals, he adds, those waves often reflect off of nearby objects, particularly with metal. The RF signals can also be absorbed by plastic and silicon. "It is quite possible because of that, at least in this use case, you saw the results we had."
Nonetheless, Bennett says, the test demonstrates that RFID is effective for inventorying RAM sticks if the proper tags are selected, and if the correct methods are employed to interrogate those tags. "Reading tags attached to RAM is not always a straightforward process," he states, "and it very much depends on the tag you are using, whether a cover is on and the type of RAM—but it is possible."
The IT Asset Tracking Benchmark is now available for downloading from ODIN's Web site. The cost is $1,500 for an enterprise-wide business license; educational institutions can contact ODIN for a free copy of the report.
