Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[sbrbot]

There is ambiguity with description of JBOD volume structure. When I was installing my two NAS drives I was noted in installation procedure that in case of failure of one disk in JBOD stucture data from other disk will not be lost (contrary to RAID1 array). Maybe that's true but it does not automatically mean that one will be able to access data on other disk in NAS in case of one disk fails. This recovery of data on second disk is possible only using data recovery tools and connecting this drive to PC. NAS will NOT give you access to other JBOD disk if one disk in JBOD strucvture fails!!! That was not what I understood from JBOD description in installation procedure. There's one guy made a test by extracting one disk from JBOD stucture and describing this situation from first hand: http://www.tomshardware.co.uk/forum/244547-14-jbod-failure-recovery

[JavaLawyer]

. . . in case of failure of one disk in JBOD stucture data from other disk will not be lost (contrary to RAID1 array).

I think you're confusing RAID-1 and RAID-0. A RAID-1 array will remain accessible in the event of one HDD failure. The data on a RAID-0 array will be irretrievably lost in the event of a single HDD failure.

. . . This recovery of data on second disk is possible only using data recovery tools and connecting this drive to PC. NAS will NOT give you access to other JBOD disk if one disk in JBOD strucvture fails!!! That was not what I understood from JBOD description in installation procedure. There's one guy made a test by extracting one disk from JBOD stucture and describing this situation from first hand: http://www.tomshardware.co.uk/forum/244547-14-jbod-failure-recovery

The test performed by the user on Tom's Hardware was conducted using a DNS-343, which is a much older ShareCenter model. The firmware / web UI has undergone a significant work-over from the DNS-343 to the more recent DNS-320L. You'll need to test the DNS-320L behavior to validate the results.

One thing the user had correct, is that it's always a good idea to play with the different formatting options on your NAS to test it's behavior under different circumstances before committing real data to the unit.

[JavaLawyer]

Your post prompted me to draft the following entry in the ShareCenter FAQ: DNS ShareCenter - Always Test Your ShareCenter Before Storing Live Data

[sbrbot]

I think you're confusing RAID-1 and RAID-0. A RAID-1 array will remain accessible in the event of one HDD failure. The data on a RAID-0 array will be irretrievably lost in the event of a single HDD failure.

You're right, I messed RAID numbers, I thought about mirroring and talked about mirroring but wrote RAID0 instead of RAID1. I'll correct it in OP not to confuse others.

The test performed by the user on Tom's Hardware was conducted using a DNS-343, which is a much older ShareCenter model. The firmware / web UI has undergone a significant work-over from the DNS-343 to the more recent DNS-320L. You'll need to test the DNS-320L behavior to validate the results.

One thing the user had correct, is that it's always a good idea to play with the different formatting options on your NAS to test it's behavior under different circumstances before committing real data to the unit.

I intend to change my volume structure and maybe I try the same thing on my 320L.

[sbrbot]

I made the same test; I formatted two 2TB disks in JBOD structure of 4TB, saved some data on it, extracted one of disks from NAS and ... as expected - I was not able to read any data from NAS until I returned the first disk. Then I returned the first one and extracted the other one, the same thing - I was not able to read any of data. So if you loose one disk from JBOD structure, you'll not be able to read the data on other disk.

[JavaLawyer]

I am curious how data recovery applications would treat the mounting of one of the two JBOD HDDs. With rare exception, the data will never span both HDDs, so all data should be recoverable.

[sbrbot]

Not all data will be recoverable, only data from healthy disk of JBOD structure (in best case 50% of them). Since JBOD is not striping, I don't know how it spreads data between disks. If JBOD saves data by means to first fulfill one disk and then continues on other disk - this is the worst method because the first disk, which is fully utilized, will broke then first and all data from it will be lost.

[JavaLawyer]

I'm referring to your test case. Both JBOD HDDs were healthy, you simply removed one from the ShareCenter to see if the other mounted independently. I'm surmising that both HDDs are independently recoverable using third-party alternatives. Based on the JBOD file handling protocol, individual files should route to only one physical HDD. A single file should only span HDDs if there is not enough space left on one of the two physical HDDs.