enveloping, batching, signing

[Rob Seaman]

On Feb 4, 2008, at 11:46 AM, Steve Allen wrote:

> If there be thousands of asteroid candidates daily, do we expect that
> they will send out individual VOEvent packets for every single one?
> (Maybe so, at least for the exceptional cases of imminent impactors,
> but most cases will be main belt asteroids of no urgent interest.)
> Are they going to prefer to process the night's worth of data
> until they have a digested list of candidates and then send that
> whole list out once daily as a huge batch?

I believe planning for LSST remains to run to pipeline within seconds  
or minutes of data-taking.  With that volume of data you certainly  
don't want to sit on the data until daylight.  Note the key use case  
of permitting follow-up during the same observing session as the  
discovery.  Batching guarantees that follow-up cannot possibly occur  
until the following night.

> So would it make more sense to define some sort of VOEvent envelope
> into which arbitrary numbers of fully-formed VOEvent documents could
> be transmitted?

We have to support both.  Can the envelope also be made lightweight  
for the single packet case?

> If the latter, then the use case question for digital signing of the
> VOEvents might be better if the envelope contains the signature which,
> under most cases, would externally sign all of the VOEvent documents
> in the envelope.

This seems reasonable.  Note that this sort of envelope need not  
actually encapsulate the VOEvent packet.  In fact, we should embed the  
signature in an entirely separate follow-up packet since a document  
and its signature can travel separately.

> This would alleviate the need to include the
> complexity of the W3C Signature element in the VOEvent proper, and it
> would greatly simplify and speed the authentication process.

Sound like good goals.

> But in closing I note that if there are not two parties already
> exchanging VOEvent packets who find utility in digital signing,
> then there may not be enough justification to bother including
> it at all.

I assert that two such parties exist (e.g., NOAO-South and NOAO- 
North).  Reliable publication will become increasingly important.  We  
can't afford to become casualties of data barbarians spamming our  
alert networks.

Rob