Stateless tape model

Mon Aug 28 10:43:40 PDT 2000

* From the T10 Reflector (t10 at t10.org), posted by:
* JoeBre at Exabyte.COM
*
This message is in MIME format. Since your mail reader does not understand
this format, some or all of this message may not be legible.

------_=_NextPart_001_01C01117.8084CA00
Content-Type: text/plain; charset="iso-8859-1"

Rob (et al) - 

I can see that I was not appreciating the distinction between
'stateless' and 'explicit state'. Given the distinction between the two,
I believe 'explicit state' is the correct model. Trying to make a tape a
random-access device is problematic at best. Indeed, given variable
length blocks, compression, etc., truly stateless tape (random access)
may be an unsolvable problem for writes.

Joe Breher 
Exabyte Corp 

> -----Original Message----- 
> From: Rob Basham/Tucson/IBM [ mailto:robbyb at us.ibm.com
 ] 
> Sent: Monday, August 28, 2000 8:09 AM 
> To: t10 at t10.org 
> Cc: JoeBre at exabyte.com; Rob Basham/Tucson/IBM; 
> MikeTa at boulderpost.exabyte.com; LarryF at boulderpost.exabyte.com; 
> ChuckK at boulderpost.exabyte.com; Jimsc at boulderpost.exabyte.com; 
> paul_a_suhler at notes.seagate.com 
> Subject: Re: Stateless tape model 
> 
> 
> Ralph, 
> These are good comments.  Your second comment really gets to 
> the heart of 
> the matter and the main question I have been asking myself: 
> Are we going stateless, or is this just a way to make the 
> state transitions 
> explicit?  If the answer is truly stateless, a new model is in order. 
> Otherwise, it seems to me this fits inside the streaming command set. 
> 
> At this point, I'm open to going either way with this.  My 
> initial leaning 
> is to not make this truly stateless.  Tape drives are 
> streaming devices. 
> Particularly when they are writing it is difficult to 
> envision a stateless 
> model.  On reads a stateless model makes more sense. 
> 
> Along with introducing the proposal, this is the only issue I 
> want to bring 
> up at the meeting in September.  Once we decide this 
> question, we can go 
> off and work on the details. 
> 
> Regards, 
> Rob Basham 
> 
> 
> (Ralph's comments below) 
> 
> 
> Comments embedded. 
> 
> JoeBre at exabyte.com wrote: 
> 
> > 
> > (4 or 6?) byte absolute LBAs should be 8 byte: 
> > 
> 
> 8 byte LBAs, absolutely.  If larger CDBs are needed, let's find the 
> operation codes and do it. 
> 
> > 
> > 3.5 - SPACE(10)... 
> 
> > 
> 
> If the base model for stateless tapes is disks, then the SEEK 
> is the right 
> verb. 
> 
> > 
> > Perhaps we should define a new Device Type identifier for Stateless 
> Stream Device. 
> > 
> 
> By all means a new Device Type seems in order. 
> 
> Thanks. 
> 
> Ralph... 
> 
> 
> 
> 
> 
> 

------_=_NextPart_001_01C01117.8084CA00
Content-Type: text/html; charset="iso-8859-1"
Content-Transfer-Encoding: Quoted-Printable

<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 3.2//EN">

RE: Stateless tape model Rob (et al) - I can see that I was not appreciating the distinction = between 'stateless' and 'explicit state'. Given the distinction between = the two, I believe 'explicit state' is the correct model. Trying to = make a tape a random-access device is problematic at best. Indeed, = given variable length blocks, compression, etc., truly stateless tape = (random access) may be an unsolvable problem for writes. Joe Breher 
Exabyte Corp > -----Original Message----- 
> From: Rob Basham/Tucson/IBM [mailto:robbyb at us.ibm.com] 
> Sent: Monday, August 28, 2000 8:09 AM 
> To: t10 at t10.org 
> Cc: JoeBre at exabyte.com; Rob = Basham/Tucson/IBM; 
> MikeTa at boulderpost.exabyte.com; = LarryF at boulderpost.exabyte.com; 
> ChuckK at boulderpost.exabyte.com; = Jimsc at boulderpost.exabyte.com; 
> paul_a_suhler at notes.seagate.com 
> Subject: Re: Stateless tape model 
> 
> 
> Ralph, 
> These are good comments.  Your second = comment really gets to 
> the heart of 
> the matter and the main question I have been = asking myself: 
> Are we going stateless, or is this just a way = to make the 
> state transitions 
> explicit?  If the answer is truly = stateless, a new model is in order. 
> Otherwise, it seems to me this fits inside the = streaming command set. 
> 
> At this point, I'm open to going either way = with this.  My 
> initial leaning 
> is to not make this truly stateless.  Tape = drives are 
> streaming devices. 
> Particularly when they are writing it is = difficult to 
> envision a stateless 
> model.  On reads a stateless model makes = more sense. 
> 
> Along with introducing the proposal, this is = the only issue I 
> want to bring 
> up at the meeting in September.  Once we = decide this 
> question, we can go 
> off and work on the details. 
> 
> Regards, 
> Rob Basham 
> 
> 
> (Ralph's comments below) 
> 
> 
> Comments embedded. 
> 
> JoeBre at exabyte.com wrote: 
> 
> > 
> > (4 or 6?) byte absolute LBAs should be 8 = byte: 
> > 
> 
> 8 byte LBAs, absolutely.  If larger CDBs = are needed, let's find the 
> operation codes and do it. 
> 
> > 
> > 3.5 - SPACE(10)... 
> 
> > 
> 
> If the base model for stateless tapes is disks, = then the SEEK 
> is the right 
> verb. 
> 
> > 
> > Perhaps we should define a new Device Type = identifier for Stateless 
> Stream Device. 
> > 
> 
> By all means a new Device Type seems in = order. 
> 
> Thanks. 
> 
> Ralph... 
> 
> 
> 
> 
> 
> 
------_=_NextPart_001_01C01117.8084CA00--