Presentations
The following presentations are confirmed
- Reliable, Repeatable, Reproducible Infrastructure - Steven Ellis
- Puppetising Kennards Hire - Lindsay Holmwood
- Tuning hosts for network performance - Glen Turner
- Linux Performance Tuning - Anshul Gupta
- High Availability, Replicating MySQL database - Oliver Hookins
- Disaster Management & Recovery Using Open Source - Ryan Fuerst
- OpenLDAP Enterprise Features - Bruce Huang and Tommy Yan
- A Special way to manage your Linux boxes by IM protocol - Tower Wan
- What's the time, Mr Wolf? - Lindsay Holmwood
- Wireshark - packet sniffing - Martin Visser
-
Reliable, Repeatable, Reproducible Infrastructure - Steven Ellis
How should we build and manage an enterprise ready scalable operational environment for Linux? Operational teams need to move beyond a simple "build and walk away" scenario into a truly "managed" model.
We will cover the key operational aspects of Linux in the Enterprise -
- Provisioning
- Patching
- User Management
- Configuration Management
- Security
- Logging
- Change Control
The key aims of our environment will be to make sure it is
- Reliable - Stability is key in the Enterprise
- Repeatable - Cookie cutter model usable across multiple accounts
- Reproducible - Rebuild rather than backup, simplifies problem resolution
- Scalable - Ability to rapidly scale at minimal additional operational cost
Based on work done for both Vodafone NZ and IBM NZ this will draw on a real world solution and tool set, but the standards, framework and methodology could easily be applied to your own environment and team.
Bio
For the last 2.5 years Steven has dual roles of Linux Architect, and Team Lead of the Linux Operations Team at IBM NZ. Prior to joining IBM he developed a scalable Linux solution for Vodafone NZ.
Length: 40m
-
Puppetising Kennards Hire - Lindsay Holmwood
"Don't Repeat Yourself" is a common principle in programming circles, and a foundation of sysadmin practice. Most sysadmins use simple automation techniques to handle daily tasks, but what happens when you start scaling beyond just a few machines?
Kennards Hire have successfully deployed Linux desktops and servers to all their branches, but the automation challenges they face in production have only just become apparent. Kennards have successfully started automating deployment and maintenance of their systems using the Puppet configuration management system.
This talk will detail Kennards efforts to build a Puppet infrastructure that scales from their desktop deployment to the datacenter, and how good design has allowed them to create componentised, reusable, and scalable infrastructure.
Length: 20 minutes
-
Disaster Management & Recovery Using Open Source - Ryan Fuerst
Using open source tools such as amanda, backup ninja, bacula, rsync and other tools to manage and implement a secure open source disaster recovery option. The topics discussed in this presentation will include the following.
- Analyzing your current environment and determining what resources and services you need a disaster recovery plan for.
- Determining the risk level you are willing to have with those resources.
- Create a disaster recovery solution using open source tools and maintaining that solution.
- Testing your solution.
Length: 25 minutes
-
Tuning hosts for network performance - Glen Turner
The bottleneck for network performance used to be transmission capacity. Huge amounts of capacity can now be had for moderate cost. The bottleneck has now moved to the host connected to the network and to the network protocols and their tuning.
This presentation gives Linux system administrators information to identify practices which lead to poor network performance. Common fault scenarios and the tools used to investigate those scenarios are explored.
The pros and cons of using Linux as a network middlebox are examined and a checklist of good practices for middleboxes are given.
Specific practices include: bandwidth-delay product and TCP buffer tuning; Linux buffer autotuning; congestion, link loss and TCP performance; common causes of link losses; choice of TCP algorithm, performance and fairness; applications design. Tools include: what do ping and traceroute actually measure; using iperf and ttcp, using Web100 kernel patches.
Middlebox issues include: participating in the control plane; anti-DoS measures; allowing fault diagnosis; queuing; minimising jitter; fairness; concurrent flow effects.
Glen is a network engineer with Australia's Academic & Research Network (AARNet). He has had long interest and experience tuning the AARNet network for high performance.
Length: 25 minutes
-
Linux I/O Performance Tuning - Anshul Gupta
Linux has evolved from being a hobbyist operating system to a production level operating system running front-line mission critical applications. Understanding Linux performance tuning has now become a major asset to systems administrators, IT auditors, consultants etc.
This session will discuss I/O tuning for Linux kernel 2.4 and 2.6 including:
- Gathering I/O related performance data
- Tuning the I/O subsystem for a given workload
- Tunables for elevators, filesystems, swap etc.
Anshul has been working on Linux for over 10 years. He is presently working with NAB as ESX/Linux/Unix Technical Designer. Before NAB, Anshul was working with HP as System, Security and Network Consultant/Linux Specialist where he was mainly involved in Linux and Security projects for various mid to large sized clients.
Length: 25 minutes
-
High Availability, Replicating MySQL database - Oliver Hookins
Making use of MySQL 5.0, DRBD, Heartbeat 2 and IPVS, some Perl scripts and other handy utilities we can build a high performance replicating database cluster. Usually the master server is the single point of failure so DRBD and Heartbeat 2 is used to add HA to this role, greatly increasing the availability of the overall system.
I'll go into information relating to:
- hardware choices
- replication technology/topology
- technical limitations or gotchas
- helper scripts
- backup strategies
- performance
Note: An extended version of the same talk will be given at the MySQL Miniconf by the same presenter.
Length: 25 minutes
-
A Special way to manage your Linux boxes by IM protocol - Tower Wan
When talking about Linux system admin, we usually think of using Telnet or SSH to control the machine, and using SNMP to monitor its status. However, if there are scores of Linux boxes to be operated, such as power off, it would be a hard work to shutdown these servers one by one. A cluster tool may help us to solve this issue but it is not flexible or convenient enough. Sometimes you may manage a group of machines; sometimes you may just operate one server. The popular IM application can be a special and efficient way to easily deal with such kind of cases.
Jabber, as an IM application, is the implementation of XMPP protocol for message exchange. Currently there are lots of open source Jabber server, client and library implementations, which provide a choice of building a private IM network. The Jabber client robot, acting as an agent on the managed node, can receive messages from other remote Jabber clients. Messages could be Linux shell commands, or user defined instructions to be executed on the Linux systems automatically, output or result will be sent back to the remote Jabber client. Also SSL and ACL will guarantee the data security. So both remote control and monitor functions could be achieved by IM
Jabber server, client and robot together make up of the IM network at the application layer to manage your servers in the LAN. It is a shortcut for the interaction between the system administrator and the machines.
Length: 25 minutes
-
OpenLDAP Enterprise Features - Bruce Huang and Tommy Yan
As an open source implementation of Lightweight Directory Access Protocol, OpenLDAP is a very good choice to store and manage enterprise directory data. This presentation is mainly about OpenLDAP enterprise features and enhancements which are valuable to construct enterprise directory system using openLDAP. As a global company, HP builds its enterprise directory website (http://directory.hp.com/) and DirectoryWorks tool based on OpenLDAP Enterprise Edition (symas CDS) to manage the enterprise data with different geographical distribution. In the first part we introduce the functionalities of HP enterprise directory system and its requirements. And HP directory structure is taken as an example to introduce large enterpise directory environment and applications. To meet these enterprise requirements, the challenges including enterprise grade robustness, Password Policy, Data Constraints, Group Policy and translucency, are analyzed. In this part we will also introduce how HP worked with Symas to implement these features using OpenLDAP and what is the current status. For these enterprise features, OpenLDAP provides overlays to implement them. The second part will cover the concept of OpenLDAP overlays, and how to configure and use the main overlays such as ppolicy, refint and constraint.
Length: 25 minutes
-
What's the time, Mr Wolf? - Lindsay Holmwood
Ever noticed that the year is missing from syslog timestamp? As a Unix sysadmin you've almost certainly dealt with logs - they're a crucial part of debugging and solving everyday problems. But what happens when you're sorting several years worth of logs? How do you process a log file with entries that span multiple years?
Having written a soon-to-be Open Sourced lightweight log processing and searching application, Lindsay will be examining some of the corner cases when processing logs, and discuss different methods for processing RFC3164-formatted logs.
The talk will be an extension of a recent blog post of mine
Length: 25 minutes
-
Wireshark - packet sniffing - Martin Visser
Wireshark 101 or how-can-i-figger-out-wot-the-flamin'-network-is-doing!!!!!
For many system administrators, what happens on the Ethernet and beyond is quite opaque and unfathomable. However, being able to efficient examine the network is a great boon to troubleshoot system issues, understand performance issues, and the like. Developers of network-centric applications should have a good understanding of the impact of their application on the wire. I have used Wireshark (formerly known as Ethereal) the packet sniffer and analyser many times to objectively finger-point at configuration issues and steer sysadmins in the right direction. This talk will show how to use the important features of Wireshark, how it teaches you about networking and how it makes your life better.
Martin has been involved in network engineering and consulting for nearly 20 years, working as a Network and Security Consultant with HP in Sydney for 10 years. He has been using Linux since 1992 and have been irregularly attending SLUG meetings since then.
Length: 25 minutes