Archive by Author

Deleting all non-expiring keys in a Redis sever from command line

23 Dec

This ending year, we have been reading, installing and testing several in RAM key-value storage(NoSQL), engines.   We have decided to give Redis a try mainly because it can save the data to disk, in case you have to restart a server or it crashes, you have have your data back.  Our experience with Redis has been mixed, good and bad things, but that will be for another post…

Research article on Workload Characterization now Online

16 Dec

A couple of days ago, I was presenting our latest research article, entitled: Characterization of Workload and Resource Consumption for an Online Travel and Booking Site.  It was for the:  2010 IEEE International Symposium on Workload Characterization (IISWC).

The paper is about how we measure, describe the Web sessions, and characterize performance of each request type and applications for an Online Travel Agency.   If you are interested in more details, you can download it and read the abstract from: (more…)

Yottaa, a great tool for monitoring the performance of your site

22 Nov

Recently I have found this really cool and useful website: Yottaa, besides the curious domain name is a web 2.0 style monitoring and benchmarking site. You input what pages to monitor or benchmark and it uses an automated Firefox browser to load and process the whole page, including the CSS and Javascript. Then it reports you page load time, time to interactivity, and a full YSlow report, and their custom score.


Calculating 95th, 99th, 50th (median) with a single MySQL query

16 Nov

As a first post, let’s start with a basic tip for SQL queries I use very often.  First let’s start with some basic info on percentiles:

Percentiles are one of the best indicators of how well our web site is performing, unlike averages.  Averages tend to hide information on outlier values, and while they might be showing you adequate numbers, they might be hiding how a significant portion of  users are seeing your site.

For websites, a good percentile to base measurements is the 95th percentile, while for network equipment the 99th.  This is because you might have some pages that are inherently longer to process, such an availability search for a specific product, while network equipment doesn’t have this constrain.  You should really use the number that best represents your workload, 95th seems to be set as the standard but you can also use the 85th according to the type of site you are monitoring.  The median (not average) corresponds to the 50th percentile.  (more…)

Hello world!

8 Nov

Welcome to yet another blog on web server performance and optimization, but with a especial touch.  Expect several posts during the next weeks, in the mean time, you can check the About section with some more info on the Who? and Why? of this blog.

The Quotes section, with motivation and inspiring quotes on the topic

The blogroll to links of other blogs I admire and follow

And the books section, with some essential books on the topic, some that I read, and some are on the [long] reading list.

Any comments welcomed!  :D