Slashdot Mirror
Finding Lost IT With RFID
CWmike writes "Vendors are increasingly trying to sell users on the idea that they need to stick RFID tags on IT equipment to keep track of it. Users are interested in this technology because they would much rather automate inventory tracking then go server-to-server with a bar code scanner and clipboard. But the new push for RFID tags in data centers also hints at a larger issue: There may be a significant amount of equipment that can't be located. And while out-of-sight, out-of-mind is not always bad, there's a least one nagging problem: 'Ghost server' systems, which may still be drawing power but perform no work and may be difficult to locate. One vendor at the Afcom data center conference suggests IT shops get some 'GPS for your assets.'"
2) Stick RFID tag to rack...
3) ???
4) Profit!
"National Security is the chief cause of national insecurity." - Celine's First Law
Obligatory bash.org quote:
#5273 +(30069)-
<erno> hm. I've lost a machine.. literally _lost_. it responds to ping, it works completely, I just can't figure out where in my apartment it is.
[alk]
And there's Sheldon, putting RFID tags on all the mice and keyboards in the server room (after he finishes RFIDing his socks).
"With all due respects, Dr. Cooper..."
Free Martian Whores!
Find stuff that migrated to somebody's apartment?
"Common sense will be the death of us all"
A decade ago I heard about an office move where they found a locked closet that nobody knew about.
They opened it up and there was an OS/2 server that hadn't been rebooted in 2 years.
Its a bit like those Japanese soldiers they used to find periodically on Pacific islands, thinking the war was still on. That poor little OS/2 server, not knowing netcraft had long ago declared victory, and that there was nobody left to talk to it.
...then how is knowing that the server you're looking for is (or more likely is not) somewhere within X meters going to help?
There's no failure quite as dissatisfying as a complete and total solution to the wrong problem.
The RFID systems I have seen in the field are poorly implemented. Most were thick, think 9v battery, tags that were either attached via zip ties or velcro. Even if it was securely attached, most were attached to removable face plates, while others were attached to the rear and would actually prevent you from pulling out the server and/or damage the cabling if you did, as it tended to hang down and catch on stuff. (snap off fibers, pull out power cords, etc.) They offered no assurance that that piece of equipment was in the room since they could easily be separated from the tag. Even with this system, you'll still need people to visually verify it anyway.
How often do you actually lose a piece of hardware? This is a solution to a problem that does not exist.
Barcode or your own SN sticker followed up by visual inspections is cheaper, safer, and more reliable compared to the RFID solutions I have seen out there.
http://www.informationweek.com/news/showArticle.jhtml?articleID=6505527
By John Rendleman
InformationWeek
April 9, 2001 06:58 AM
The University of North Carolina has finally found a network server that, although missing for four years, hasn't missed a packet in all that time.
Try as they might, university administrators couldn't find the server. Working with Novell, IT workers tracked it down by meticulously following cable until they literally ran into a wall. The server had been mistakenly sealed behind drywall by maintenance workers.
When information is power, privacy is freedom.
Even if inventory and operations live together in perfect harmony, the tags identify PHYSICAL servers. Thanks to the magic of virtualization, you might have several zombie virtual machines along with [maybe] one that is truly needed -- all in the same physical box.
Even if the tags do their job and you think you have positively identified a defunct box to be shut down and removed, what level of confidence do you have that NONE of the virtual machines are still necessary?
The primary motivation for this technology -- last time I was told about it -- was in hospitals. Expensive equipment is wheeled around a lot, and people sometimes need to know where it is now. An RFID scanner in rooms/doorways and tags on the equipment could tell you this -- so long as the tag was resistant to being bashed against a doorway.
We had this problem in the mid-90s. We had a Sun server in the building which was regularly used by remote logins (I think it was a build machine so just used to build the Sparc version of the software), but one day we had to find for a hardware upgrade and no-one could remember where the heck it was... we eventually had to get it to play music so we could walk around the building and listen for it.
As an electrical engineer, I frequently have to work with IT folks to provide data gathering systems on the equipment we install in our manufacturing facilities. Some of these plant floor networks are huge, and have tentacles that reach into every machine and sub-system processor. I never cease to be amazed at the complete lack of documentation that the IT folks put into physically mapping their network equipment. They will quite literally wave their flashlights and point to where they want the central network switches installed. While we and the mechanical engineers draw plans which show general equipment arrangements, and draw up network diagrams showing how our equipment is to be networked, and we label our equipment and electrical panels, the IT guy typically will typically tell me that yes, he thinks there's a switch around here somewhere I can use, and starts hunting around for it.
In my world, while it is quite possible to build and erect a machine without any prints or plans, any future maintenance or additions to such machines would prove to be doubly expensive since it would require a not inconsiderable amount of detective work to come to understand what exists so that it can be modified or changed. (Indeed, back in the early days of engineering, that's how things were built, and it took many decades before the value of making plans and documenting them was recognized.)
It seems to me that creating and maintaining a complete set of documents which map and explain the equipment and network should be adequate, and would prove to be simpler to keep up to date than any sort of RFID system of tying cowbells to servers. Granted, it requires resources and consistent effort, but this has long been the norm in the field of manufacturing engineering. If it works for machines and manufacturing equipment, why wouldn't it work for IT systems?
Ah, but that would mean managers would have to pay money to actual workers to do actual work. Which is boring, and they don't get to sit in any meetings.
They much rather just pay money to some consultants that just tell them all that is well with the new gimmick they are about to buy, while they look at a nice Powerpoint presentation and drink coffee.