How to Really Scare Microsoft
Open Source advocates and UNIX aficionadoes have long offered UNIX/Linux as a replacement for Microsoft's ubiquitous Windows(r) platform, hoping that somehow Microsoft's platform monopoly could be broken by software that was cheap, fast, and good. It's not going to happen. Aside from niggling details, UNIX/Linux have the same fundamental problems as Windows: system administration, lack of standardization, kitchen sink design, security vulnerabilities, and lack of fundamental protection for data and intellectual property rights.
To dethrone Microsoft, it is necessary to create a compelling vision of an environment that could lower the total costs of computing to never-before-seen lows. What might such an environment resemble? Do you really want to scare Microsoft? In this presentation, I will outline the high-level design of a marketecture that would stand a decent chance of dethroning the king.
slides (PowerPoint, 1.1MB)
Automating Linux installations
Wouldn't it be great to automate the installation and configuration of your Linux systems just like you can with Solaris and JumpStart? With Red Hat and SuSE you can.
This talk is an introduction to Red Hat's KickStart and SuSE's AutoYast installation tools. We will begin with a brief review of installation server requirements and network booting in a Linux environment. This will be followed by a more in depth discussion of the tools themselves with a focus on their features, configuration files and related utilities.
slides (PDF, 425KB)
Intrusion Detection and Prevention: Present and Future
Dennis Brown and Richard MacVarish
Viruses, worms, and other malware are a constant threat to any network. Technologies such as Intrusion Detection and Prevention systems help identify and mitigate these threats, and provide a picture as to what is actually happening on a network. This talk focuses on explaining what IDS and IPS are, where they came from, and where they is going.
The goal of this talk is to present a clear picture of what IDS and IPS systems are capable of, and how to properly deploy them. Tools such as Snort will be discussed and explained, as well as techniques and methodologies surrounding IDS and IPS.
Capturing System Crash Information for Postmortem Analysis
System crashes cannot be avoided, but steps can be taken to minimize the incurred downtime. This talk focuses on how to collect as much information as possible at the time of a crash to help developers fix the problem before the crash happens again (or at least reduce the number of test/reboot cycles required to classify the problem)
The talk begins with an overview of many of the tools available today, ranging from the basic panic/oops message deciphering to kernel crash dump facilities and kernel probes. Each tool is described at a high level, providing basic information on what it does, why it is useful, and whether it is supported by the various Linux distributions. The talk then focuses in on Netdump and Crash, providing a walk-through on how to set up netdump under Red Hat Enterprise Linux and Fedora Core. The talk concludes with a live demo of the crash program.
When Email Won't Do: Giving Presentations Both Technical and Mundane
The stereotypical geek talks faster with a keyboard than with the spoken word, and rarely emerges from the gentle light bath of the monitor to interact with humans in the flesh. However, actual geeks are sometimes called out of their natural setting and asked to speak in front of live people. You may have seen them, trying to hide in meeting rooms, user trainings, or on stage at LISA. Their wretched lack of experience and/or confidence can be telling. From conference talks to budget meetings, the poor geekly wretch displays afflictions such as the Silent Stammer, the Aimless Shuffle, and the Endless Ramble, clinging to slides, podia, and the clock on the wall, much as a drowning rat clings to a stick of floating wood.
Public speaking ills are not limited to geeks, whatever pop culture might lead you to believe. Even professional talkers, such as politicians, lawyers, and the dreaded salesgoons can make the most elementary mistakes. I'll cover general strategies for organizing and creating a talk, tips and tricks to condense jargon and technical details into forms that even the most pointy haired manager can grasp, and how to avoid the most common, and most annoying vocal and physical tics. I'll also address the ever elusive question of how to deal with stage fright, and the even deeper mystery of knowing when your audience is with you and when a hundred eyes are drooping -- and how to wake it up if the latter. Public speaking is like any other sysadminly activity: it becomes much easier given the proper tools, knowledge, and experience.
Chris Palmer is manager of IT production services at the Harvard Division of Engineering and Applied Sciences, and CTO/CIO of AskOnline Inc, none of which remotely qualifies him to deliver this talk. However, in his non-geek life, he has spent ten years volunteering as a public speaking instructor and coach for Boston area high school students. His teams and students have been consistently top ranked -- he has coached upwards of a dozen National finalists and champions. He began as a pitiful communicator, and attributes his coaching success to a history of having nothing come naturally, and so having to learn everything the hard way.
Systems Navigator for Configuration Management
Christian Pearce will be presenting a brief overview on the state of configuration management in the industry. Which will lead into how Systems Navigator (SysNav) was developed to manage data centers by leveraging open source tools like Cfengine. There are several techniques SysNav employs to accomplish the goals Systems Administrators set when attempting to automate the management of their systems. These techniques range from modeling the type of configuration management strategy, to writing custom Cfengine code that leverages SysNav's Cfengine architecture.
Christian Pearce is a Senior Software Engineer, who has six years experience developing system administration solutions. Currently he is the lead architect for Systems Navigator, which is an enterprise configuration management tool that leverages Cfengine.
“Show & Tell, Ask & Answer”
There was no formal presentation; instead, members talked about their current work and asked for help solving their current problems.
It’s Not (Just) About Bandwidth: Improving Application Performance over the WAN
Mark Stuart Day
Users and administrators are often frustrated by the performance of applications running over a wide-area network (WAN). These configurations are becoming more common as organizations consolidate servers in data centers, due to cost considerations or new regulations such as Sarbanes-Oxley and HIPAA. A typical response to a site with poorly-performing applications over the WAN is to increase the bandwidth of the connection to the WAN, but that change often has little or no effect on the user experience. In this talk, I outline three distinct contributors to poor performance over a WAN: bandwidth bottlenecks, TCP-level chattiness, and higher-level chattiness such as in Windows file sharing (CIFS) and Exchange email (MAPI). I compare and contrast some of the various product approaches available to address these problems, and demonstrate Riverbed's Steelhead product.
Mark Stuart Day is chief scientist at Riverbed Technology, where he focuses on patents, research, standards, and the future architecture of Riverbed products. His previous employers include Cisco, Harvard, IBM, Lotus, and MIT.
(something from/about Opsware)
System Administration and Sex Therapy: The Gentle Art of Debugging
Already debugged three things today and it’s not even breakfast? You must be a sysadmin. Our life is chock full of debugging “opportunities.” Not only do we have to fix problems in complex systems, we often find ourselves debugging the interactions between complex systems designed by other people.
To understand this process better and to get better at it, we’re going to turn to an unlikely source of information: sex therapists, counselors, and educators. With their help, we’ll explore why improving the interactions between complex systems when they go awry is so hard and what techniques and craft can be used to make the process easier. Come to this talk not for its mature subject matter, but for the chance to learn to be a better sysadmin through better debugging.