>One thing I would like to see much more of is graphical display of what's >going on. You can have all the numbers in the world, but good graphical >visualization tools can make it so obvious what's going on! >
I completely agree. Quite a while ago Bryan and Mike did a presentation which included a rather nice looking diagram of system wide sleep/wakeup activity. It was generated with a few lines of D and AT&T's graphviz package. I was blown away! I went away and used this same package to generate a graph of all the work done by a single ufs_write() and the printout was over 7 feet wide! In the end a colleague and I put together a call graphing application which lets you dynamically generate and manipulate call graphs for any kernel function (supports various graph manipulation features and source browing for any node in the graph). Being able to actually "see" what the code flow looks like helps understanding of the code tremendously (at least for dullards like myself!). I think the marriage of people who actually understand visualisation with a good source of data will be powerful indeed. I recently visited a University in the North of England that researches application visualisation. The techniques that they have for visualising application behavioural characteristic are amazing and now that we have an almost unlimited source of data I think the possibilities are extremely exciting. Cheers. Jon.
