Page 1 of 2

Speed or reliability

PostPosted: Sun Jun 04, 2017 7:36 am
by DrTeeth
I knew I had seen reference to metadata before. I ran a search and
found this
https://www.terabyteunlimited.com/ucf/viewtopic.php?f=4&t=2043. So is
the new metadata feature the unreliable metadata mentioned in that
link above?

It IS a lot faster when backing up, but backing up is not compatible
with a lack of reliability so I ain't switchin', especially as most of
my OSs do not use NTFS.
--
Cheers,

DrT

** Stress - the condition brought about by having to
** resist the temptation to beat the living daylights
** out of someone who richly deserves it.

Re: Speed or reliability

PostPosted: Sun Jun 04, 2017 1:21 pm
by Brian K
I'd be interested in knowing if the current metadata method is the same as that described in the above link. Or if it's an improvement. I'm using /usemd for backups but not for restores.

Re: Speed or reliability

PostPosted: Sun Jun 04, 2017 7:43 pm
by TeraByte Support
It's not about metdata methods, it's about metadata. If metadata is
unchanged but data changed, using metadata won't detect the change.


"Brian K" wrote in message news:13878@public.image...

I'd be interested in knowing if the current metadata method is the same as
that described in the above link. Or if it's an improvement. I'm using
/usemd for backups but not for restores.


Re: Speed or reliability

PostPosted: Mon Jun 05, 2017 7:17 am
by tas3086
TeraByte Support wrote:
> It's not about metdata methods, it's about metadata.

It's not about metadata, it's about RELIABILITY. It's that simple.

Re: Speed or reliability

PostPosted: Mon Jun 05, 2017 4:23 pm
by Brian K
In the link above is...

"While the write tracking method recently added is not totally reliable, it's
a step above metadata."

Regarding Track Writes, the IFW manual states...

"In addition, it can only track writes that have been sent to the drive and would not detect changes that are cached. Therefore it's not recommended to use this option,"

So should we stick with restoring from the standard non metadata/ non tracking choices? I assume metadata backups are OK as the generated .TBI is the same whether /usemd is selected or not.

Re: Speed or reliability

PostPosted: Mon Jun 05, 2017 6:00 pm
by TeraByte Support
tracking isn't for restores, it's for backups only, and only until
reboot/power down. It is a step above because even if metadata isn't
updated, the write will be known.

However, It's as simple as if you want to use metadata options, just realize
that if a file is updated where no metadata is updated, it won't find a
change. If you want to do metadata restores, ensure you use vss for your
backups. Or in general, it's recommended to use VSS if using the metadata
option.



"Brian K" wrote in message news:13881@public.image...

In the link above is...

"While the write tracking method recently added is not totally reliable,
it's
a step above metadata."

Regarding Track Writes, the IFW manual states...

"In addition, it can only track writes that have been sent to the drive and
would not detect changes that are cached. Therefore it's not recommended to
use this option,"

So should we stick with restoring from the standard non metadata/ non
tracking choices? I assume metadata backups are OK as the generated .TBI is
the same whether /usemd is selected or not.


Re: Speed or reliability

PostPosted: Mon Jun 05, 2017 8:38 pm
by Brian K
TeraByte Support wrote:
>
> However, It's as simple as if you want to use metadata options, just
> realize
> that if a file is updated where no metadata is updated, it won't find a
> change.

I'd love to do metadata restores and I don't think I'm the only one here who doesn't understand what you are saying. Can you give an example of "if a file is updated where no metadata is updated, it won't find a change".

Are metadata restores of data partitions less likely to be affected by the above issue than OS partition restores? Especially as files in the data partition might not have been updated for days.

Re: Speed or reliability

PostPosted: Mon Jun 05, 2017 11:25 pm
by TeraByte Support
It's all answered in the very first answer/explanation in the other thread.

It comes down to simply, it's recommended to use vss if you want to /usemd
and you should only /usemd if you understand that using it is only as good
as the metadata provides.

personally, the servers will continue full backups to mirrored nas over gb
network each night, other systems will continue using simple operations to
usb devices, however I'm considering using metadata on the main computer I
use, probably set that up when the 10TiB drive is connected.


"Brian K" wrote in message news:13883@public.image...

TeraByte Support wrote:
>
> However, It's as simple as if you want to use metadata options, just
> realize
> that if a file is updated where no metadata is updated, it won't find a
> change.

I'd love to do metadata restores and I don't think I'm the only one here who
doesn't understand what you are saying. Can you give an example of "if a
file is updated where no metadata is updated, it won't find a change".

Are metadata restores of data partitions less likely to be affected by the
above issue than OS partition restores? Especially as files in the data
partition might not have been updated for days.


Re: Speed or reliability

PostPosted: Tue Jun 06, 2017 12:33 am
by Brian K
I'm using VSS and I'm using /usemd for backups. I assume the .TBI files are the same as if I hadn't been using /usemd.

My confusion is with restores. What can go wrong with a metadata restore if the metadata doesn't match the data? Can files be missing from the restored partition?

Re: Speed or reliability

PostPosted: Tue Jun 06, 2017 12:11 pm
by Bob Coleman
I just don't get it. Why would something known to be less reliable be added as a feature?