TAP, skynode, or boutique?
szalay at jhu.edu
Fri Sep 5 08:51:40 PDT 2008
Exactly. And this is how the SkyNode protocol should be redesigned, but in
order to do this we had to wait until ADQL and TAP have the required
features in place. For Roy's needs I would use directly the 3.3.6, and this
should also be the way for the SkyNodes to transmit these requests. This
will need some extra work, though, related to a multi step crossmatch
process, etc. I would prefer to discuss these issues if people feel that the
time is right in a short special meeting during the IVOA.
From: Guy Rixon [mailto:gtr at ast.cam.ac.uk]
Sent: Friday, September 05, 2008 11:42 AM
To: Roy Williams
Cc: dal at ivoa.net
Subject: Re: TAP, skynode, or boutique?
quoting from the TAP/PARAM draft that was current before the Trieste
3.3.6 Multi-Position Queries
A multi-position query generalizes POS, SIZE to a table of positions,
arbitrarily large number of spatial position-based queries to be executed
simultaneously. In a typical scenario the user uploads a list of the
their favorite objects, and executes a spatial cross match against some data
table. The multi-position query provides this simple spatial cross match
(full text at http://www.ivoa.net/internal/IVOA/TableAccess/tap-v0.2.pdf)
Unless the NVO people working on this have found problems, this is still to
be part of the initial TAP spec (we'll find next week when we start to edit
the proposals into a unified draft).
Therefore, I would expect that you might be able to use TAP installations
for cross-match sometime this year with prototypes and next year with
On 5 Sep 2008, at 16:14, Roy Williams wrote:
Dear DAL WG
I am writing to ask if the Skynode protocol will continue to be supported in
the broader IVOA. Most of the current implementations seem to be at Johns
Hopkins University -- is that true? Are there other skynodes out there that
are and will be properly maintained?
A group of us in US-VO are developing an application (Vim) that handles
multiple sky positions by running proximity (cone) searches against
published catalogs. Thus a single click of the mouse can cause execution of
a thousand separate cone searches against a single server. This can be very
slow, and we would like to scale things up. We want a crossmatch service --
by which I mean no more than multiple cone searches. We want to do a
thousand or a million cone searches in one "crossmatch" request.
Which of these paths should the Vim team follow:
(1) After years of discussion in this WG, I am still not seeing the
emergence of anything like simple crossmatch in the TAP specification. So I
guess we are still more years from widespread, solid implementation of TAP
that can do crossmatch. Is that true?
(2) The Skynode protocol can do crossmatch already, and I wonder if this is
the direction the Vim team should follow instead of waiting for TAP.
However, before doing this, we would really like an assurance that there
will continue to be multiple organizations supporting Skynode!
(3) The third way to get fast crossmatch would be to look through all the
data center web pages looking for their own crossmatch services (eg
), and then build a custom interface from Vim to each data center.
Thank you for your help
California Institute of Technology
626 395 3670
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the dal