Conference cookie::hsm

re:                   <<< Note 333.0 by CSC32::V_HEINICKE >>>
    
>    Have a customer who only wants to use magneto-optical devices for 
>    storing data - he wants no tape drive associated with HSM.  He thinks
>    the volume requires specification of a tape drive - BASIC MODE. How 
>    would he set up this up?
    
These statements seem to be contradictory. HSM supports use of Magneto-Optical 
in any one of the following modes: 

    1.  As temporary cache devices, the MO's accumulate data which is 
	eventually copied to archive tapes. 

    1a. Selecting /BACKUP will cause the archive copies to be written during 
	shelve operations. This is the most secure approach. 

    1b.	Selecting /NOBACKUP will cause the archive copies to be written during 
	the next cache flush operation. This is the highest performance 
	approach. 
	
    1c. The frequency of flush operations is controlled by either the /INTERVAL 
	or /HIGHWATER qualifiers, or both. Data is deleted from the cache, 
	after the archive copies are made (according to 1a, 1b).

    2.  As permanent cache devices, the MO's accumulate data which remains 
	in the cache. This mode is specified by setting both /NOINTERVAL and 
	/HIGH=100 qualifiers. 

    2a. Selecting /BACKUP will cause the archive copies to be written during 
	shelve operations. This is the most secure approach, and HSM will 
	retain both near-line and offline copies of the data. 

    2b.	Selecting /NOBACKUP will prevent archive copies from ever being made. 
	This approach relies upon either the inherent reliability of MO, or an 
	external backup vehicle for data protection. 

	
>    He also wants to use expiration date of 180 days and along with that he
>    would like to set retention on the volume.  It is my understanding
>    that he should not set volume retention to anything but the default
>    (mandatory).
>    Is that a correct assessment and if so what is the reasoning
>    behind this.  
      
VMS volume retention is used by HSM to approximate a file's last access time. 
The min/max values suggested in the HSM ops guide were selected to provide the 
most accurate approximation, while imposing the minimum overhead. These values 
are certainly *not mandatory*. 


Hope this helps.
/Paul

    So if option 2b is used a customer could configure a system with
    optical and HSM but no tape devices?
    
    Would this work with a jukebox also?  If so, can optical media be
    removed but still be known about by HSM?
    
    Rob (OSxS/VMS guy)

>    So if option 2b is used a customer could configure a system with
>    optical and HSM but no tape devices?
 
Yes.  This is what we call the "permanent cache option".  However, as
Paul said, we highly recommend that these permanent cache devices be
backed up on a regular basis.  You don't have to have optical to do this.
Any old disks setup as permanent caches will do.
   
>    Would this work with a jukebox also?  If so, can optical media be
>    removed but still be known about by HSM?

Yes.  I'm assuming you are talking about an optical jukebox such as an RW500.
HSM supports these kind of jukeboxes if a product like OSMS is installed.
However, HSM does not keep track of the platters.  To HSM, they are just 
another disk device (thanks to OSMS).  If HSM needs to access one of the 
optical "devices" (platters) but it is now gone or a new platter has 
replaced an old one (files no longer present as expected), HSM will probably 
fail the operation.  Does this answer your question?

Barb

    Barb,
    
    Your response answered those questions but leads me to these questions;
    
    Does HSM contain the concept of "offline" data sets where "offline"
    means the data is not currently available and requires human action
    to make it available.
    
    If so, can this concept be applied to data sets contained on optical
    media that is not in a jukebox?
    
    If not, is this idea, or similar, being considered for HSM's future?
    
    Rob

>    Does HSM contain the concept of "offline" data sets where "offline"
>    means the data is not currently available and requires human action
>    to make it available.

HSM doesn't within the product itself.  This is left to MDMS/SLS as the
media manager and applies to tapes only.
    
>    If not, is this idea, or similar, being considered for HSM's future?

We've been thinking about a concept where MO caches can be used as archive
classes, but I don't think the current design has HSM keeping track of the
location of individual platters.  To me, this would fall under the realm 
of a media manager which HSM is not.

Barb

    I like the layering of products and the distinction of functionality.
    From the names, sounds like the media manager should be responsible
    for keeping up with the data sets.
    
    I don't know if any optical based products can provide the core piece
    of functionality needed to develop a robust media manager.  That is
    a unique media id that can be made available in a timely manner.  The
    only optical media I have seen with this is the Kodak 14 inch media
    and it must be read from the disk.  Enough dreaming....
    
    Thanks for the information.
    Rob

>    of functionality needed to develop a robust media manager.  That is
>    a unique media id that can be made available in a timely manner.  The

This is exactly the problem!

Thanks for your interest.

Barb