On 03/26/98 13:45:12 you wrote:
Exactly how were they "bad"? Did they not work when you added them, or did they work but you had to troubleshoot crashes that caused you to swap them out, etc.? Netapp does indeed do testing that catches bad SIMMs from the vendor, but it can't catch problems where the SIMM is damaged after testing in handling, or in shipping, or improperly inserted into the board.
Failed to boot w/ the error, "Unsupported NVRAM size 5MB", after placing the six new SIMMs in the NVRAM card. Did the memory shuffle (got very annoyed because it doesn't report this error until close to the end of the boot process) and narrowed the problem down to 2 SIMMs. Then I proceeded to get really pissed (only allowed to do non-emergency hardware maintenance on the filer Friday's after 19:30) because I spent over an hour and ruined a Friday night to add 2MB of NVRAM. (F330's support 2, 4, or 8 MB - not 6.)
When I've seen this in the past, it's been due to improper insertion of the SIMMs into the board. But it could be the SIMMs were bad or even the slots on the board were bad. I assume you've gotten new SIMMs since then and they worked fine?
Sometimes the time at which a SIMM will fail in the future is beyond their horizon of testing. It seems to me getting it straight from the vendor is
How were these tested?
I can only speak in general terms. To say they're just put in a memory tester would be incorrect; they get burned (or at least, they used to) just like the DRAM.
My filer didn't even acknowledge their exictence. Granted, they could have been damaged during shipping... Or, I could have zapped them when I opened the packaging while grounded to our ESD bench... Maybe I don't know how to install SIMMs and just always got lucky before... (Please excuse the sarcasm.)
No need to be sarcastic. All of those are very possible, no offense intended to your consider memory installation skill. :)
But surely they were tested at Netapp, so it seems like you can't "blame" them for a failure later. Is the chance of a failure later still large enough that Netapp's testing isn't worth it? Again, that's an individual choice.
only going to increase your chances of failure, unless maybe there is a particular vendor with which Netapp has rarely encountered failures.
I disagree. I'd prefer to go to a vendor with which _I_ have rarely encountered failures.
Well, that is mostly wordplay. Surely one person's experience is not as predictive as testing thousands of such SIMMs, but most people do develop such individual biases, especially if they get a couple of failures in a certain vendor in a row. The point of the sentence was simply that NVRAMs are tested and they do fail, so if you want to go with a vendor that isn't tested, it would seem your best choice would be one that historically had failed little in said testing.
Does it justify increased price? Hard to say. Depends upon the atmosphere of the company/environment your in. If it's like many bosses I know, the increased price isn't worth it until a component fails, and then they get upset that you didn't get better quality. :)
Increased price isn't a problem if you _do_ get better quality. We paid a 600% mark up on parts that had to be shipped back and replaced as soon as we tried to use them.
That doesn't mean you don't get better quality. Without Netapp's testing, it's more probably that you would have gotten bad parts twice in a row. If you have multiple filers, you would generally see more bad SIMMs, etc.
I'm not saying NetApp is low quality. We've seen great uptime w/ our F330 and keep asking for details on their HA system because we think NetApp is a good bet. I just don't like the way they handle memory and disk sales.
This is more or less what I was getting at, too. Netapp does do quality testing, and I think that's better than no testing. Is it worth, assuming your figures, a 600% mark up? It seems many people on mailing list say it isn't. What about a 200% mark up?
Bruce