Ultrium Recording Format vs. HDD LBA

[Shiera]

Hello everyone,

I'm only just a neophyte on storage... and I desperately need your
help on these:

1. How are data stored in Ultrium 1 and Ultrium 2 format?
I mean, is there any block addressing method?

2. Do all tape formats have the same block addressing methods?

3. If the tape has logical block addresses, is there any correspondence
between Tape LBA and HDD LBA?


I've been searching for a sufficient resource material on Tape but
couldn't find one. I was kind of hoping you could recommend one.

I would really appreciate your help. Thanks a lot!

- Shiera -

 

[Arno Wagner]

Hello everyone,

I'm only just a neophyte on storage... and I desperately need your
help on these:

1. How are data stored in Ultrium 1 and Ultrium 2 format?
I mean, is there any block addressing method?

I would be surprised if not. It will not really be random access, though.

2. Do all tape formats have the same block addressing methods?

Physically: no. However SCSI-Tapes have all the same
command set, since it is standardized. If I remember correctly
from some experiments with DAT, you can tell it to seek to a
position and then read the next n sectors in a second command.

3. If the tape has logical block addresses, is there any correspondence
between Tape LBA and HDD LBA?

No. HDD LBA is beacuse HDD addressing is 3D: head, cylinder, sector.
All SCSI drives use linear addresses anyway. So there is no LBA.
(Unless they started making IDE Ultrium drives?)

I've been searching for a sufficient resource material on Tape but
couldn't find one. I was kind of hoping you could recommend one.

Look into the SCSI command set, takpe storage section. Info e.g.
here: http://www.danbbs.dk/~dino/SCSI/SCSI2.html
I think tape drives are usually "Sequential-access devices".

BRW, there are some tools to dispatch SCSI-commands manually,
at least under Linux. I used them some years ago to play around
with a DAT drive. You can find the Linux SCSI docs here:

http://www.tldp.org/docs.html#howto

I hope this helps.
Arno

 

[Folkert Rienstra]

I would be surprised if not. It will not really be random access, though.


Physically: no. However SCSI-Tapes have all the same
command set, since it is standardized.

And so should ATAPI.

If I remember correctly
from some experiments with DAT, you can tell it to seek to a
position and then read the next n sectors in a second command.


No. HDD LBA is beacuse HDD addressing is 3D: head, cylinder, sector.
All SCSI drives use linear addresses anyway. So there is no LBA.

Huh? That is what LBA is all about.

 

[Shiera]

Hello everyone...

Just some follow-up questions...

Since data are stored sequentially, so the block positioning are also
sequential, right? How can we identify an end of the block, especially
if a single cartridge contains blocks of different sizes? Is this
where filemarks come in?

Once again, thank you so much for your help. ^-^

By the way, thanks for the references you gave me, Arno. They helped a lot!

- shiera -

 

[Arno Wagner]

Hello everyone,

[...]
Physically: no. However SCSI-Tapes have all the same
command set, since it is standardized.

And so should ATAPI.

Since ATAPI uses the SCSI command set, that is definitely the case.

Huh? That is what LBA is all about.

There is no LBA in addition to the native addressing is what I
meant. There is nothing else except the original linear block
addressing, which you could call "LBA". But the capitals are
not justified for SCSI.

Arno

 

[Arno Wagner]

Hello everyone...

Just some follow-up questions...

Since data are stored sequentially, so the block positioning are also
sequential, right?

Yes, but that is more a question of efficiency, not causality.
Any SCSI device, also HDDs, will give you a linear addressing
of the bytes/blocks on the medium. That is just an SCSI
convention. The idea is that the user should not need to know
about geometry. Old concept in professional computing, relatively
new in the low cost computing market.

How can we identify an end of the block, especially
if a single cartridge contains blocks of different sizes? Is this
where filemarks come in?

Physically the drive writes some codes at the end of the block.
Logically I am not sure you can actually determine where a block
ends with variable block length. Then the filemarks could
be used as separators between different byte-streams.
However I am not really sure how they work.

I also have no idea whether the LOCATE command works
with variable size blocks.

Once again, thank you so much for your help. ^-^

By the way, thanks for the references you gave me, Arno.
They helped a lot!

You are welcome. I also found them quite helpful in the past.

Arno

 

[Folkert Rienstra]

Hello everyone,
[...]
Physically: no. However SCSI-Tapes have all the same
command set, since it is standardized.

And so should ATAPI.

Since ATAPI uses the SCSI command set, that is definitely the case.

Huh? That is what LBA is all about.

There is no LBA in addition to the native addressing is what I meant.

The native addressing for SCSI *IS* LBA.

There is nothing else except the original linear block addressing,

There IS no such 'linear block addressing' as opposed to 'Logical Block
Addressing'. They are one and the same.

which you could call "LBA". But the capitals are not justified for SCSI.

Yes, they are.