Tony Bourke

My name is Tony Bourke. I live in NYC and work in the technology sector. I'm also a freelance writer, writing on a number of subjects including Unix, networking, and load balancing.

Here are a few of my articles.

Published Works

Server Load Balancing, through O'Reilly And Associates.

Online

Have Your Layer Cake And Eat It Too on the O'Reilly Network website. This article appeared in the final issue of Web Hosting Magazine, and was reprinted on O'Reilly's site. The article discusses how network vendors are more and more adding features that would otherwise be seen in seperate boxes into a single device.

Parsing With Precision, which is based on PC-based load balancers versus hardware-based load balancers. It was published in Network World Magazine. It was one of my first articles published, and as it turned out, I ended up being wrong. I didn't take into account Moore's law, which meant general purpose CPUs were getting faster. At the time, general purpose CPUs weren't as fast as ASICs, and now that difference doesn't seem to be much of a problem. F5 is doing pretty well, using Penium IV processors.

Stop The Serial Madness!

When administering devices via serial console, finding the right cable at the right time can become very annoying, as anyone who's dug through a datacenter during an outage looking for the right damn cable can attest to.

This article discusses a method I've used to deal with such an issue. This was published in HostingTech Magazine, but they've since ceased operations, and recently, web operations, so I'm posting it here in its entirety.

It's Always The Load Balancer published on O'Reilly's website, which talks about how load balancers get blamed unjustly for many problems with Web installations.

Bridge-Path vs. Route-Path Server Load Balancing published on O'Reilly's website, regarding the placement of load balancers in a given infrastructure.

An article on console networks, my first published work. It was published in December of 1999, in Sys Admin Magazine. There is no online version of this article.

It should be noted that a couple of companies are now selling what I proposed in this article. Shoulda patented it, I guess :)

Load Balancers as Firewalls on O'Reilly's website, which discusses the ways in which a load balancer can be used for some firewall functionality.

Sun Versus Linux: The x86 Smack-down
This was my first article for OS News, and the first article I've written to be featured on slashdot. It pitted Solaris 9 x86 platform edition with Linux (RedHat 9, but Linux in general), and it was quite a contentious article. The flamings on slashdot taught me a lot about the nature of benchmarking and geek-player hating in general.

UnixWare 7.1.3 Review
Much has been made about the SCO lawsuit and their nefarious legal tactics. I wanted to contribute to the debate, so I decided to to do something that I hadn't seen any one do (at least, not recently) which was to take an objective look at what SCO was offering in terms of a Unix operating system. It turns out, it's pretty weak.

This was the second article I wrote to be featured on Slashdot, and I was spared flaming as the posters preferred to flame SCO itself.

The UltraSPARC Series

In January of 2004, I published the first in a series of articles on OSNews based upon a 5 year old Sun Ultra 5 that I had laying around. What was going to be a simple review turned into a major project, consuming quite a bit of my time. The series is still ongoing.

My Sun Ultra 5 And Me: A Geek Odyssey
This is the intro article, outlining the system and my motivations for doing such a review. It brings me back to the days of racing Sun Enterprise 450's down the hall (true story, unless you're someone who might want to hire me, in which case it's fabricated).

Are 64-bit Binaries Really Slower than 32-bit Binaries?
The conventional wisdom is that 64-bit binaries are slower than 32-bit binaries, but I couldn't find any benchmarks to back up that assumption. So I ran a few of my own. The results were interesting.

This one also ended up on slashdot. Much flaming ensued.

GNU GCC versus Sun's Compiler on the SPARC Platform
Another bit of conventional wisdom states that Sun's compiler suite produces binaries that are faster than GNU's hyper-ported GCC. The results were interesting, even refuting a conculsion or two in the previous article. As it turns out, the conventional wisdom is mostly true, although the difference between the two compilers is far less than many have speculated.

This ended up on slashdot as well, although only in the developers section. As a result, there were still flames, but not quite as many.