Securing the Client
Allen Holub
Thursday, March 18, 10:30am – 12:00pm
One of the largest roadblocks to secure computing is the client machine. The problem has several ramifications.

First, The client machine is itself vulnerable to infiltration, and most users don't know how (and shouldn't have to know how) to harden their machines. This problem affects not only the client machine itself, but the integrity of the internet as a whole. The MYDOOM virus, which caused millions of unsuspecting users to participate in a denial-of-service attack, is a case in point.

The client also is a major source of vulnerability in a corporate environment. For example: someone walks into your building and plugs in a virus-laden laptop; someone brings in a floppy and unwittingly installs a virus. The scenarios are endless.

The technical problems associated with securing a powerful general-purpose computing device masquerading as a consumer product are nontrivial.

Moreover, the problems here are not just technical, but social as well. Is it really reasonable to expect your average PC user not to click on an email attachment that made it's way successfully through the anti-virus filter and appears to come from a friend? Can you real prevent all your employee's kids from surfing the net or getting email on their parent's only-for-work laptop? You can't really address this problem if you just blame it all on the operating system.

This round table will discuss some of the ramifications of creating a secure client platform, discussing both the problems, the difficulties, and perhaps some strategies that might lead to a solution.

Agile Software Development: Principles, Patterns, and Practices
Robert C. Martin
Friday, March 19, 10:30am - 12:00pm
For the first four decades of software development we did not had a good definition of our craft. If you asked a software developer what he did, he'd have to say something vague like "I write code." Sometimes developers created something good, but could not repeat the process. They could do good things sometimes but didn't know how they did it. Quality was more an accident than a design. In the last decade this has changed significantly. We have identified enough principles, patterns, and practices to provide our craft with a solid definition. This definition gives us something we may not have had before—pride in workmanship. For though we may have been able to take pride in some of our projects before, we weren't often able to take pride in the way those projects were build. Now, we can take pride in *how* we build software, not just in the software we build.

Performance, Tuning and Testing Linux Clusters
Tim Strickland
Friday, March 19, 3:30pm - 5:00pm
Industry analyst IDC expects the cluster market to grow by nearly 35 percent to $2.27 billion by 2005, with Linux expected to be the most pervasive operating system used on clusters. Without the proper amount of planning, performance tuning, and testing, these cluster implementation are at risk, and may not deliver the level of flexibility, high availability and workload management desired. In this roundtable we'll talk about the importance and the options available.