During the past few decades the easiest solution to a computer resource constraint was to wait 9-18 months and buy new hardware. Now we are approaching a time when the ecological and economic cost of computational resource consumption is again rising above our threshold of attention. Knowing which resources are constrained and which applications are consuming these resources could make the difference between a successful GNOME deployment and a lost opportunity.
Typical GNOME resource consumption is higher than on some alternative open source desktops. We may suspect that much of this is necessary because of GNOME’s superior functionality. However, by making this assumption without accurate measurements we might overlook areas where GNOME really could use improvement.
Tools now exist in OpenSolaris, BSD and GNU/Linux which help us more accurately measure the footprint of a GNOME desktop and point out any “hot” areas which could use our attention. This talk explains the pitfalls and misunderstanding of tools such as free, top and pmap. It also briefly discusses the use of tools such as dtrace, exemap and xrestop to answer questions such as “How much memory does a GNOME desktop really consume?” “How often is this applet on the CPU?” “How many happy GNOME users can I deploy on my hardware?” “Which resources are constrained and which applications use these resources?”
I intend to share lessons learned from these tools and from data collected during large thin client GNOME deployments. I also intend to offer suggestions and open discussion on the future and discuss whether there anything we can do to shrink the GNOME footprint.
