Free Republic
Browse · Search
General/Chat
Topics · Post Article

To: discostu
If people work locally then copy to a central repository if the repository craps out you should have the latest out there somewhere, much better chance of data recovery then than if everything lives exclusive in the repository.

So rather then properly architect a data storage solution you want to depend on the cheapest disk dell can send you? this would not pass in an enterprise environment.

ummm Raid 10 has no 'parity disk'. Raid10 is either s mirror of stripes or a stipe of mirrors. Raid 4/5 and some of the more exotic ones you might see on a dell array or an e series box from IBM use parity, and usually they allow redundant parity. Other raids which do use parity disk now can have more than one, and you're still ignoring hot swaps.

Quick example: I have two raid fives which are mirrored, each raid 5 has two hot disk, how many disk can I lose?

In addition to all the above the disk used in network storage devices are usually on (1) always on, (2) on UPS, and (3) of higher quality than what dell might send you in a desktop. All of this means the failure of any given raid disk is far less likely than the failure of the disk in your pee-cee.

83 posted on 10/05/2005 11:54:00 AM PDT by N3WBI3 (If SCO wants to go fishing they should buy a permit and find a lake like the rest of us..)
[ Post Reply | Private Reply | To 82 | View Replies ]


To: N3WBI3

No, what I'm saying is that working locally provides another layer of backup over and above your porperly architected data storage solution. Even if you set things up right restoring your shared data area after a crash takes time, time when people can't be productive if that's the only source of data.

There's always a magic combination of lost drives that'll hose your raid array, it's the nature of the beast. I'm not saying raid is a bad thing, in the modern world where data storage is frequently too big for effective backup raid is an absolutely necessary, I'm simply saying it's not bulletproof. If the gremlins are in a bad mood you can always be hosed.

Always on until your network goes down, or that computer pukes all over itself. Centralization of functionality does nothing but centralize your critical point of failure. The more a company lives and dies by a small number of computers the worse shape it's in if one of those computers has troubles.


89 posted on 10/05/2005 12:31:55 PM PDT by discostu (When someone tries to kill you, you try to kill them right back)
[ Post Reply | Private Reply | To 83 | View Replies ]

To: N3WBI3; discostu
Quick example: I have two raid fives which are mirrored, each raid 5 has two hot disk, how many disk can I lose?

Follow-up question: assume one of the RAID arrays leaves Pittsburgh travelling on a train bound westward at 62mph and the other leaves San Francisco on an eastbound train at 56mph, on the same track. How long until the head-on collision makes the RAID arrays irrelevant. Show your work.

100 posted on 10/05/2005 1:01:24 PM PDT by Petronski (I love Cyborg!)
[ Post Reply | Private Reply | To 83 | View Replies ]

Free Republic
Browse · Search
General/Chat
Topics · Post Article


FreeRepublic, LLC, PO BOX 9771, FRESNO, CA 93794
FreeRepublic.com is powered by software copyright 2000-2008 John Robinson