Talk:BukManiac

CS290F Fall 2006 - UCSB Computer Science - Thorsten von Eicken

Jump to: navigation, search

Project Comments

This all seems to be quite good. I think you may want to have search be an advanced feature. I know that sounds a bit strange since it seems like such a fundamental feature. However, it's not essential to what we're trying to acheive in the class, and implementing searching is harder than it might seem.

Also, the controller names could use some work. Having so many of them start with book_ is a little odd. Think about the CRUD model: your controllers should in general correspond to the major data models and provide Create, Read, Update, and Delete functionality for the corresponding model, interacting with any related or ancillary data models at the same time.

Good work!

Stefan 14:24, 26 October 2006 (PDT)

Made changes to the names of the controllers according to Stefan's suggestions.
Tushar 13:44, 28 October 2006 (PDT)

  • I can't get to your site, I only see the Fedora Core Test Page. Please fix and email me when ready.
  • I see the SQL statement list but not the analysis... You will need to figure out for which queries you need indices... Also, you should probably change the %blah% string search to a blah% prefix search so you can create an index for that.

TvE 00:39, 18 November 2006 (PST)

  • Great detail for the SQL statements. What do you conclude? What do you need to optimize? What indices will you need?
  • The back link on a book page leads to a redirection loop.
  • Is there no rating summary on the book listing page?
  • Good progress, keep on goin'.... :-)

TvE 17:04, 27 November 2006 (PST)

  • Nice performance analysis & good work with the MyISAM index!
  • Time to reorganize your information into multiple pages...
  • The response time graphs don't seem to be too meaningful, they just show queueing delays when the offered load vastly exceeds the delivered service.
  • The response rate graph shows very clearly at what point the system gets saturated, nice!
  • What is the rate of real Rails page requests per second (ignoring static & image requests)?
  • What is the bottleneck at that ~22 req/sec rate? Is it the DB? Rails? CPU? Disk?
  • Assuming you understand where the bottleneck is, your next step should probably be to break the server into two: one for mysql and the other for apache+rails. Now the fun really starts!

TvE 00:48, 29 November 2006 (PST)

  • Reorganized the Wiki and added some analysis of graphs
  • Could you please provide some pointers to identify the bottlenecks

--Tushar 00:54, 29 November 2006 (PST)

  • Run top during the benchmark (change update rate to 1 or 2 seconds by hitting 's'), and watch what sits at the top.
  • Read the log/production.log which says what percentage of time is in the database vs. rendering. Also tells you which actions are the most expensive.

TvE 07:42, 29 November 2006 (PST)

Please see Scaling the system for Project 4b results and graphs. --Tushar 00:00, 4 December 2006 (PST)

Personal tools