sample queries for NVO logs (fwd)

[Maria A. Nieto-Santisteban]

What is going on about logging in NVO ...

Maria

---------- Forwarded message ----------
Date: Thu, 11 May 2006 11:58:15 -0400 (EDT)
From: Ani Thakar <thakar at skysrv.pha.jhu.edu>
To: NVO Technical Working Group <techwg at us-vo.org>
Subject: [nvo-techwg] Re: sample queries for NVO logs (fwd)


john asked me to forward bob's sample logging queries to the WG ... i 
also included my initial response.  pl. send me additional questions if 
you can think of any. 

	ani

---------- Forwarded message ----------
Date: Thu, 6 Apr 2006 18:34:57 -0400 (EDT)
From: Ani Thakar <thakar at skysrv.pha.jhu.edu>
To: Robert Hanisch <hanisch at stsci.edu>
Cc: Ani Thakar <thakar at pha.jhu.edu>
Subject: Re: sample queries for NVO logs


bob,

i dont see any problems with the majority of these in principle, 
although extracting some of this information from just the http request 
string could be challenging.  a couple that would be problematic under 
the current model are 2) and 11).  to get the fraction of mast activity 
means we have to harvest even non-VO mast logs, right?  right now i 
think karen is separating the VO activity.  i dont think we have enough 
information at least in the weblog data model to get trustworthy 
estimates of bandwidth and other computational capacity usage, or even 
to separate the sql usage out.  we wd need service logs for that kind of 
information, i believe.

	ani

On Thu, 6 Apr 2006, Robert Hanisch wrote:

> Hi Ani.  Here are some possible queries that I would be interested in
> running against the NVO logs.
> 
> 1) How many requests to MAST/HST originate as a result of DataScope (or some
> other NVO application or interface)?
> 
> 2) What fraction of MAST data deliveries results from requests coming from
> NVO services?
> 
> 3) How many hits were there on all NVO interfaces/services last month?
> 
> 4) Show me a time-based plot of hits with granularity of
> hourly-daily-weekly-monthly.
> 
> 5) Which are the most frequently used NVO interfaces and services?  Show me
> a histogram.
> 
> 6) How much data is being retrieved as a result of NVO interfaces and
> services?
> 
> 7) Which NVO data provider services provide the best service?  Where best is
> defined as 
>     a) Fastest initial response
>     b) Fastest completed request
>     c) Fastest completed request normalized by data volume
> 
> 8) Which NVO data provider services are the most/least reliable?
> 
> 9) What kind of information are users looking for in the registry?  What
> terms were used in queries, and with what frequency?
> 
> 10) What are the "hot objects" that people are looking for data about?  That
> is, what object names are most frequently queried?
> 
> 11) What fraction of capacity are we using in
>     a) Bandwidth to various sites?
>     b) Computational services (sites like WESIX or the WCS Fixers or the
> mosaic services)?
>     c) Database queries?
> 
> 12) How many different users are there of NVO interfaces and services?
>     a) Where are they? (university, national center, US-based, other
> countries)
>     b) How many are repeat users vs. one-time explorers?
> 
> I'll keep thinking.  These are all pretty obvious, no?
> 
> Bob
> 
> 

-- 
Aniruddha R. Thakar, Research Scientist
Center for Astrophysical Sciences, JHU, Bloomberg 375
3701 San Martin Drive, Baltimore MD 21218-2695
410-516-4850, Fax: 410-516-5096  
thakar at jhu.edu, http://www.sdss.jhu.edu/~thakar
-----------------------------------------------------------------------
Never doubt that a small group of thoughtful citizens can change the 
world. Indeed, it is the only thing that ever has. [Margaret Mead]