I wonder what they had to 'nuke from orbit'? Apparently whatever it was, it was serious or critical enough to cause problems with the NYSE.
Most likely scenario, IMHO:
NYSE calls Netapp support-- Says they have a problem with some Netapp proprietary technology that has a list of quirks a mile long-- For shits and giggles, let's say it is snapvault. One of the kinds of snapvault relationship you can create is intended for government record keeping compliance. It WILL NOT ALLOW YOU TO DELETE ANYTHING until the retention date that was specified at the vault's creation is reached. No tool in the netapp software arsenal will permit this.
So, let's say some earstwhile NYSE admin, who is probably some schmuck in india who can't speak english, because despite the NYSE operators having more money than you can possibly imagine, they just cant be assed to hire competent local admins, because profits. That said-- this guy is not really properly versed in how the storage controller actually works, or any of the quirks of the various proprietary solutions offered with the platform, and so he creates such a protected relationship to test it out. Then cant delete the vault afterwards, because *GASP*, the retention date is not yet reached. Because he is an incompetent boob, he did this with a production filer, and now is fucked beyond all reason, because that production server's resources are (as usual) overcommitted, and under provisioned. Every second that the vault persists on that production filer is lost money for NYSE, and a potential legal liability.
So, he calls ntap support. They tell him there is no way to remove the vault before the retention date has occured. (because there isn't.) This escalates out of control, until finally a NYSE corporate retard DEMANDS satisfaction on the issue, and some level 3 or higher engineer begrudgingly says that special boot option 4A will be able to eliminate the protected vault. This is only after having exhausted the much more sensible option of altering the NTP server to report that the retention date specified by the vault has occurred, and thus faking out the software, allowing normal deletion-- because doing that would fuck up the whole network (part of the network would think it is in the future, and it would fuck up trading EVERYWHERE, because it would appear that live trades were happening in the past.) Instead, they bite their lips, grit teeth, and implement option 4A, hoping that cluster failover will cover their asses while the then nuked filer gets rebuilt from DR backups.
However, because of how mission critical their installation is, THEY NEVER TESTED FAILOVER. As a consequence, nuking the filer takes the whole damn trade floor down, and their cluster panics like a politician who just had their dirty laundry aired.
I dont KNOW that this is what happened, but I can CLEARLY see this happening, as it is a perfectly plausible chain of events.