Mast Kalandar

bandar's colander of random jamun aur aam

Sat, 09 Feb 2008

< Some significant changes in the IMSc CC since 1997 | · | What is Langlands' programme? >

Documenting past mistakes in CC purchases


Tags: , [link] [comments (0)] [raw]
purchase

I am not trying to assign any blame in the following list. (I was a signatory to the minutes based on which many of these purchases were made. I was also among the people who expressed interest which led to the purchase of some these items.) I am merely trying to document what I perceive as past mistakes of the CC.

On Thursday I wrote:

If so required I can provide examples where the CCM has justified the purchase of equipment which our users subsequently did not use (or used insufficiently to justify the expense).

  1. The Power Challenge (darth). This heavily under-utilised computational machine was even off for a while without anyone noticing. The difficulty of installing useful software on this machine was one of the “power challenges” to making long term use of it.

  2. We recently paid more than double the basic desktop price for higher-end desktop machines based on the idea that we will use them for distributed computing. That has not happened. This is not the first time that IMSc has paid in excess of 50% above the cheapest desktop price for desktop machines. In most cases this has also meant additional lost man-hours spent on installing useful software on these machines. Bleeding-edge hardware means that the user of such hardware bleeds time debugging it.

  3. We bought gigabit network infrastructure (switches and cabling). This was never really justified except that it is “the wave of the future”. Since we do not do any internal video/audio conferencing, these speeds have never been really used. In fact, I am not sure we know how to make use of all the software features of these switches.

  4. We have bought a number of SUN/Solaris machines based on the idea that some people need to use them for computation. As far as I know the number of people who actually do this is less than the number of SUN/Solaris machines. The machines have also contributed to the eating up of man-hours spent installing useful software on it. Equivalent commodity hardware would have cost about 1/2 of what we paid.

  5. We originally bought proprietary hardware for the LANL mirror. Other than the negative value caused delays in disk expansion due to unavailability of the (proprietary) disk, no value was added to our use by this expensive proprietary hardware. (After we moved to commodity hardware we have added disk entirely seemlessly.)

  6. We bought extra hardware like a laptop with a docking station, an Iomega ZIP drive, a graphics tablet, a MacMini. Most of these have not been used by even a handful of users.

  7. We spent a lot on special sofware like compiler kits for SGI, SUN, extra licenses for Matlab, Mathematica. There are very few users of most of these licenses (perhaps Mathematica is an exception). On the other hand we have not paid any money to the TeX User’s Group, the Free Software Foundation, the Apache Foundation or the Software in Public Interest—all of whose software we use on a regular basis.

  8. The lack of utilisation of the Apple PowerMacs has recently been discussed in the CCM. Part of the reason is that the machines were not properly installed. Another part of the reason is that none of our computational users had enough of a need of this additional hardware to ensure its proper installation.

  9. About 4 years ago, we spent a lot on (what was) high-end graphics hardware for a number of workstations saying that 3-D/GL was required to do interesting work with simulations. This (graphics hardware) was not used. If only we had waited until we were ready to make use of this hardware! Nowadays similar quality 3-D/GL is supported on commodity hardware—to the extent that we use it.


Name
E-mail (will not be displayed)
OpenID (required)
Simple HTML and wiki markup are allowed.

Archives

< February 2008 >
SuMoTuWeThFrSa
      1 2
3 4 5 6 7 8 9
10111213141516
17181920212223
242526272829 

2014, 2013, 2012, 2011, 2010, 2009, 2008, 2007, 2006, 2005, 2004, 2003, 2002, 2001, 2000, 1999, 1997, 1995,