qatsi (qatsi) wrote,


Book Review: Measuring Computer Performance - A Practitioner's Guide, by David J Lilja
A rather dry book, which was to be expected. Something worthy that sits unread on the shelves for years, because you never have the incentive to read it. And when there is a problem, probably not the sort of thing, despite its title, that you can just pick up and dive straight to the pertinent section.

Despite the dryness, there were some interesting sections, particularly on design of experiments and queueing analysis. A lot of the early material is a reiteration of statistical concepts that were in A-Level maths. Some techniques, such as analysis of variations (ANOVA), were new to me. Like many statistical things, it's not clear how the technique can achieve what it purports; I think some perspectives need to be emphasized rather more to really see what's going on. Of course, being a textbook for practitioners rather than intrinsic students of statistical techniques, a lot of results are quoted without proof or clear reasoning. Many of the examples are rather low-level and hardware-oriented, which is perhaps indicative of scenarios where performance is easier to measure objectively and with the minimum of systemic interference, and therefore more suited to a student audience. It can now go back on the shelf, but I have some idea of what tools it can provide when the need arises.
Tags: books
  • Post a new comment


    default userpic

    Your reply will be screened

    When you submit the form an invisible reCAPTCHA check will be performed.
    You must follow the Privacy Policy and Google Terms of use.