RE: LTANS-ERS: Verification and data groups

["santosh chokhani" <chokhani@xxxxxxxxxxxx>]

That will be fine.  I am not as conversant in it, but will try to slog
through it.

-----Original Message-----
From: owner-ietf-ltans@xxxxxxxxxxxx [mailto:owner-ietf-ltans@xxxxxxxxxxxx]
On Behalf Of A. Jerman Blazic
Sent: Tuesday, February 07, 2006 8:09 AM
To: ietf-ltans@xxxxxxx
Subject: RE: LTANS-ERS: Verification and data groups


Basic idea was presented, I think at paris meeting by Peter (Sylvester). We
are working with XML currently (it is a version 0.9 albeit in production
already...). Will that do?

> -----Original Message-----
> From: owner-ietf-ltans@xxxxxxxxxxxx 
> [mailto:owner-ietf-ltans@xxxxxxxxxxxx] On Behalf Of Santosh Chokhani
> Sent: 7. februar 2006 13:55
> To: A. Jerman Blazic; ietf-ltans@xxxxxxx
> Subject: RE: LTANS-ERS: Verification and data groups
> 
> 
> Aleksej,
> 
> Can you point me to or send me the ASN.1 for your proposal?
> 
> Thanks.
> 
> -----Original Message-----
> From: owner-ietf-ltans@xxxxxxxxxxxx
> [mailto:owner-ietf-ltans@xxxxxxxxxxxx] On Behalf Of A. Jerman Blazic
> Sent: Tuesday, February 07, 2006 4:34 AM
> To: ietf-ltans@xxxxxxx
> Subject: RE: LTANS-ERS: Verification and data groups
> 
> 
> Santosh
> 
> Preserving of archive data is indeed performed as maintenance 
> of archive objects. Such objects are composed (or collected) 
> as they enter the archive and they in general consist of:
> 
> - archive data itself
> - metadata or any associated data (may come in flavor of 
> digital signatures, managerial data, etc...)
> - preservation/conservation data
> 
> The last element of the logical structure is what we try to 
> achieve with the ERS (but not only ERS!). Now, if 
> preservation is actually handling with archive objects the 
> preservation data must be an individual part of the object. 
> It really does not matter how the evidence information is 
> generated or collected, by grouping or not, the matter is 
> only that the evidence information as a part of the 
> preservation data and must always be decoupled for any common 
> evidence structure as the hash trees are. The ERS 
> specification as it is, lacks or redirects the focus on hash 
> trees and their generation. It solves the problem of 
> "degrouping" but statements such as retimestamping of the 
> whole hash tree are IMO wrong. There is no need to 
> retimestamp the whole hash structure but perform the same 
> type of operation on existing archive objects regardless of 
> the fact that they are in the group or not (you may always 
> build a new group form the elements of two or more groups). 
> Grouping is just another mechanism which may be used or not.
> The focus is really on the archive objects as they carry 
> evidence information individually.
> 
> Another problem of the current ERS specification are digest 
> algorithms.
> If I
> remember correctly the specification proposes the use of the 
> same algorithm through the hash structure, which is improper 
> approach. It is true that the chain is as strong as its 
> weakest element, but as we already have TAS requirement such 
> as transfer of archive data and its evidence from one archive 
> to another, sustaining the requirement of consistent hash 
> structure is just impossible.
> 
> What I propose is a better definition of the archival process 
> and the data structures to be defined by the LTANS. I tried 
> to summarize such ideas in the LTAP specification but am 
> afraid this is not the right document.
> 
> Aleksej
> 
> > -----Original Message-----
> > From: Santosh Chokhani [mailto:chokhani@xxxxxxxxxxxx]
> > Sent: 6. februar 2006 23:39
> > To: A. Jerman Blazic; ietf-ltans@xxxxxxx
> > Subject: RE: LTANS-ERS: Verification and data groups
> > 
> > Aleksej,
> > 
> > If you proposing to replace the data groups and hash trees with the 
> > archive objects as simplification, I am in full agreement.
> > 
> > I am not sure this complexity is that useful or needed.
> > 
> > -----Original Message-----
> > From: owner-ietf-ltans@xxxxxxxxxxxx
> > [mailto:owner-ietf-ltans@xxxxxxxxxxxx] On Behalf Of A. Jerman Blazic
> > Sent: Monday, February 06, 2006 11:15 AM
> > To: ietf-ltans@xxxxxxx
> > Subject: RE: LTANS-ERS: Verification and data groups
> > 
> > 
> > > Of course it is up to the client to decide which data
> > object should be
> > > pooled into a group and which not. But a verification
> > component should
> > > be able to unambiguously decide if a document group is 
> complete or 
> > > not.
> > > With the structure I sketched, verification of a single
> > data object is
> > > still possible, but additionally it can be verified if 
> the group is 
> > > complete.
> > 
> > Hmm, I could hardly agree with this, as the ERS is meant 
> for providing 
> > evidence information of an object and not group of objects. The 
> > logical (or any other reason for) grouping is done by the 
> supportive 
> > infrastructure, e.g. data management (or DMS if you want). IMO the 
> > grouping is proposed purely as the answer to scalability 
> issues that 
> > may arise when handling significant amount of data. Basically the 
> > evidence creation process should perform necessary actions to group 
> > data objects in a way that a single evidence is enough. 
> Instantly the 
> > objects archived must be degrouped and no particular information is 
> > needed to identify the group members. This is out of the scope and 
> > that is exactly what the ERS does.
> > 
> > Aleksej
> > 
>