DMX "Enable Deferred Service" option

same came up here, we as organization decided not to sign up for that. I understand EMC wants a CE come onsite and get all failed drives replaced at once, in the mean time i have a box with hundreds of drives and no hot spare. No Thank you.

When you no longer have a permanent spare available for any drive in the array a dial home is initiated to have a C/E come out and do the replacements.  If a drive fails prior to the C/E coming onsite there should still be spares available and standard sparing can take place.  The difference in standard sparing and permanent sparing is that a permanent spare must comply with all the standard rules regarding mirror positions, power zones, etc.

Robert,

What would happen when I have another drive fail and no proper permanent spares are available ?

No you do not need any more spares than before.

Spares are spread throughout the arrray and not every spare is available to be a permanent spare for every drive.  A permenant spare must comply with the rules for mirror positions as it is a permanent swap; bin change and all.

Dyanmic spares do not follow all the same rules that permanent spares must follow.

This is from a post by Michael Lee:

Note that any "hot" spare drive can be used for either traditional "Dynamic Sparing" OR the newer (Enginuity 5x71 and above) "Global" or "Permanent Sparing" feature.

The reason for the increased number and different types of spares is to provide a viable minimum quantity for Permanent Sparing operations.

In Dynamic Sparing operations ANY available physical spare drive can be used as a 'temporary" mirror for a failing drive.

With the newer / better Permanent Sparing feature the spare drive becomes the failing drive (i.e. an automatic configuration change is performed that "swaps" the failing drive and Permanent Spare drive in the Symmetrix System bin file).

Therefore Permanent Sparing operations require that the physical spare drive matches the physical attributes of the failing drive AND that this spare drive is in a suitable "backend" location to fully maintain hardware fault tolerance otherwise the Permanent Sparing operation will NOT be performed (and we use traditional Dynamic Sparing until the failing drive can be replaced).

https://community.emc.com/message/395287

So in the end I end up needing more spares then I did before ?

Thanks

You should not run out of hot spares.  My understanding of  the program is that permanent sparing will be replacing drives as they fail.  When enough drives have failed so that no one drive has a permanent spare available as new case is generated for EMC to replace the failed drive which become spares.  There will still be spares available just that a proper permanent spare is not available.

i guess i asked about that 2 years ago    ..thanks Robert.

Thanks for the responses & clarifications.

Together with them and a little more research I can see that an array should not run out of hot spares. However my understanding is that it could if drives failed in the right (or possibly wrong) sequence, i.e. a sequence that allowed all the spares (for a drive type) to be used up as permanent spares.

If 'deferred service' was enabled, a call would then be logged with EMC, but until the CE arrived & installed the replacement drives, some of the luns would only be raid protected.

I am also not sure that having more hot spares, as suggested in the earlier discussion by both Stefano Del Corno - `You obviously need more drives` & Michael Lee - `The "disadvantage" of Permanent Sparing is the need for additional spares in the Symmetrix bin file:' lowers the risk by much. If `deferred service` is allowed & if a call is only logged at the point when a permanent sparing operation causes any drive in the system to no longer have permanent spare coverage, then the array could still potentially run out of hot spares before triggering a service call.

Thanks Steve I believe that I understand the process.

However I am not comfortable with an activity (replace the failed disks) being triggered only after a resource has been completely depleted. Particularly since the disk replacement could then be delayed by factors completely outside EMC's controll - I am thinking of an 8 hour traffic jam on the main northern freeway out of here earlier this week...