My small contribution to the day was a lunchtime presentation on using Linux for back office stuff like web servers, mail servers, networking, etc. It’s a very general overview of the popular projects and tools out there for Linux. It’s mainly intended as a starting place for people looking to install a new service on a Linux platform. The slides from my talk are below.

Vertical Partitioning

Partitioning vertically is a database optimizing technique where tables with many columns are split into multiple, smaller tables.  Without partitioning, even normalized tables may be a mix of frequently and infrequently accessed columns.  This makes it difficult to keep things loaded in memory and can affect overall performance.

By splitting the table in two, you can have a table of frequently accessed columns and a table of less frequently accessed columns.  The tables can be merged together dynamically when needed, but most of the time you will just need the table with the frequently accessed data.  This approach also gives you the option of using different table types (storage engines) for each of the tables.  Of course, it’s probably easiest to implement vertical partitioning at the beginning of your project instead of coming back and making changes to the database structure later.

Horizontal Partitioning

Partitioning horizontally is a similar concept except that the table is split by rows rather than columns.  This works well when certain rows are accessed more frequently than others.  Depending on the size of the database, having a table of more frequently accessed rows can really speed things up.

The events registration system I created at MPOW can potentially benefit from this.  We have over 10,000 registrations in the system, but most of those are for past events that we rarely need to access.  Having a table of 100-200 registrations for upcoming events and a separate table for the thousands of past event registrations would probably provide a noticeable performance increase.

Query Cache

MySQL has a query cache that it uses to increase performance.  If a particular query has already been run on the database, MySQL can return the cached result instead of having to parse it, optimize it, etc.  I was already familiar with the query cache and we use it MPOW.  However, I learned a couple of new tidbits from Jay:

• “Any modification to any table involved in the SELECT invalidates the stored result.”  I guess I already knew this.  What’s cool, however, is that partitioning your tables properly can reduce this.  Making a change to a table containing upcoming event registrations would not affect the cache for the table containing past event registrations.
• Don’t use InnoDB tables when there are more than 5000 rows and you’re doing full table scans.  It’s slooooow.  Use MyISAM table types instead.
• If you are doing a SELECT on tables that change frequently, there is no point in storing the result in cache because it will soon be invalid.  Use SELECT SQL_NO_CACHE to prevent MySQL from storing highly dynamic results and filling up the cache.
• MEMORY table types are fast and good for data that can be thrown away or re-computed again.

ANSI vs. Theta Coding Style

Jay put two different versions of a SELECT statement up on the screen and asked the group which style they used.  One style used INNER JOINs (ANSI style) and the other used WHERE and AND clauses (Theta style or comma style).  I think most of the room (including me) raised our hands for the latter.  Jay made some good points about ANSI being easier to read and more portable.  I need to force myself to try ANSI and get used to it.  Theta style, however, is very natural for me and I suspect it will be hard for me to change.

The slides from Jay’s talk (including part 2) are up on his website.

Drupal 6.0 and Joomla 1.5 were recently released. These new versions make it even easier for libraries to get up and running with a CMS. Drupal is my personal favorite. We have several custom PHP scripts and Drupal allows us to easily integrate them into the content management framework. Drupal is also being used in cool libraries like Ann Arbor and Cleveland Public. Miami University is playing with Drupal as a faceted catalog and ALA has decided to move it’s website to Drupal.

There are growing options for Drupal support as well. Back in November, the DRUPAL4LIB discussion list went live. There’s a fair amount of traffic and of course there are several experts on the list to answer your questions. You can also keep up with Drupal in libraries and visit the forums at drupalib. There’s a post in the forum there on Drupal help websites.

If your library hasn’t considered a CMS yet, maybe it’s time. Even small libraries can benefit from basic CMS features like those found in WordPress. There’s even a free hosted version at WordPress.com that makes it easy to get started.

ColorZilla

This extension puts a little eye dropper tool in the lower-left corner of the browser. Clicking on the dropper and then on the web page gives you the RGB or Hex code of the color at that spot. You can also copy the code to the clipboard to be pasted into a style sheet, etc. I use this one a lot. It’s simple, but very useful.

Dummy Lipsom

I don’t use this as much as I should, but that’s because I often forget about it. It allows you to quickly fill in form fields with dummy Lorem Ipsum text. That way, you can quickly test a form without having to fill it out by hand. If you use a CMS, you can also use this to quickly create body text for a page.

Firebug

If I had to pick only two extensions, Firebug would be one of them. It sits as an icon on the bottom status bar until needed. It’s great for debugging JavaScript code, but I don’t do a whole lot of that. For me, the best feature is being able to click anywhere on the web page and seeing in real time what HTML and CSS is code is involved in that spot. Firebug also allows you to edit code and see the effects in real time.

HTML Validator

This extension is simple, but valuable. It validates web pages in the background and displays a check mark in the status bar. When viewing HTML source, it highlights the lines that are causing validation errors. It’s great for ongoing monitoring of a web site.

IE View Lite

Another simple extension. Load a page in Firefox, right-click on the page, click on “View this page in IE.” It will launch Internet Explore and load the same page you were viewing in Firefox. It’s nice for checking how pages look in IE and for pages that don’t look or work properly in Firefox.

Web Developer

If I had to choose just one extension, this would be it . It’s not quite as powerful as Firebug, but it’s very easy to use. It gives you a toolbar across the top of the browser that has the tools clearly organized and displayed. I like that I don’t have to think about how to use tool. It’s just second nature. Nifty features include a ruler, a page magnifier, a window resizer, and many more.

X-Ray

X-Ray puts an item on your right-click menu that acts as a toggle. When turned on, it surrounds a page’s content with the underlying elements and classes. So instead of a word showing up as word. It is displayed as <p><em>word</em></p>.

YSlow

This one is an extension of an extension. It works with Firebug to identify “problems” that may slow down the loading of a web page. The extension is Yahoo’s response to its Thirteen Simple Rules for Speeding Up Your Web Site. When run, it gives you a letter grade on all thirteen aspects of your page. Pay attention to the rules and what they mean. Some of the suggestions it makes can actually slow down your page depending how your site and server are set up.

I should note that all of the above extensions work in Firefox running on both Linux and Windows. I haven’t tried them on a Mac, but I bet many of them will work.

Of course, I’m not developing all of the time. Sometimes I just want to use Firefox as a browser. Having all of the above extensions loaded into Firefox can use up RAM and slow things down. So I created a separate profile in Firefox where I installed all of the developer extentions. So when I start the browser, I can choose to load my day-to-day profile with minimal extentions or the developer one.

If you develop web sites and don’t use Firefox, you should. If, however, you are stuck with IE, there is an Internet Explorer Developer Toolbar available. I haven’t used it, but I hear that it’s decent.

Hourly:

• check for odd things in system files using logcheck
• run some administrative tasks for Drupal (content management system)
• send out event reminders for Webcalendar (this runs every minute)

Nightly:

• backup all files to external disk using tar & gzip
• backup all files to an off-site server using rsync
• run web statistics software
• run a virus scan using clamAV
• sync up with time servers with ntpdate
• use grep to find any PHP script warnings/errors from the web server log

Weekly:

• use df to email me a report on disk space usage
• run the last command to see who has logged in
• verify ports in use using nmap

Monthly:

• rotate web server logs using logrotate

I also use cron with the PHP command line interpreter to run some custom scripts. Like emailing the staff each morning with current numbers from our event registration system. If you can do some scripting, you can automate almost anything with cron.