Vinod Subramaniam

Spare Drives - That tired topic !!

Discussion created by Vinod Subramaniam on Aug 20, 2016
Latest reply on Aug 21, 2016 by Swee-Chin Liem

Just came across a article which says that Netapp now has Triple Parity RAID and that Netapp was the originator of RAID5-DP.

See link below.

https://www.google.com/url?sa=t&rct=j&q=&esrc=s&source=web&cd=1&sqi=2&ved=0ahUKEwjS38-ZvtHOAhVFRCYKHYlTByAQFggcMAA&url=h…

 

For those who are in the dark as drive capacities increase so does drive rebuild times.

If you have a second drive fail during a drive rebuild with Traditional RAID5 you will have data loss.

Enter RAID5-DP or RAID6 which allows for a double disk failure.

Now Netapp claims to have introduced RAID5 Triple Parity which allows for a triple disk failure.

 

Personally I thought this was a ridiculous approach in this day and age to extend RAID5-DP algorithm to accomodate another disk failure.

 

Consider this scenario :-

1. You have a storage array where everything is traditional RAID-5 or RAID-10

2. You have L1 volatile cache of around 1TB and lets say L2 non-volatile cache of 10TB.

3. Assume that all drives in this hypothetical storage array are 1TB drives

4. All writes have to go through L1 cache but not through L2 cache

5. Lets say a drive fails during a RAID-5 drive rebuild. A virtual spare drive is immediately created in L2 cache to replace the failed drive.

    All writes to LBA addresses that translate to the failed drive will have to go through L2 cache now.

6. The number of drives that can fail is only limited by the size of the L2 cache.

 

Interestingly enough IBM has a patent just for this concept.

The date on the patent is March 2015.

 

See URL for the patent

https://www.google.com.ar/patents/US8972778

 

Some thoughts

1. What sort of media will or should the L2 cache be ?

 

Post your thoughts ............

Outcomes