Your current filters are…
Virtual hosting platforms promise easy, efficient access to computing resources on a whim. There are huge wins to these offerings, both economically and in terms of efficiency. However, many service types aren’t well suited to cloud environments. These services require special consideration in order to keep our sites, our businesses, or our organizations humming along nicely.
In this talk, you’ll learn how to identify and remedy bottlenecks in your production stack which are particular to the cloud. We’ll discuss how to get high performance out of variable, multi-tenant, and degraded (virtualized) hardware, including some of the important considerations in choosing a provider. Using experience from Urban Airship’s travails in the cloud, we’ll focus on getting the most out of your network, cpu, memory, and perhaps most importantly, your disk configuration.
by Ian Dees
In May 2008, five busy programmers in town for a conference sat down for dinner and decided to do a book together. We dreamed big dreams for the project scope, made big plans for technical collaboration, and yes, made big mistakes as we worked. Two years later, reality looks really different than we'd imagined--better, in fact.
In this talk, you'll hear about:
* Getting through the Trough of Sorrow
* What to do when you have all the responsibility and no authority
* Laughing at your estimates (and getting stakeholders to laugh with you)
* What worked for us (and what might or might not work for you)
* Lessons for freelancers, telecommuters, and distributed teams
While a terrific presentation may take talent, making a good one is a matter of science and practice. As generations of Toastmasters have proved, anyone can do it. Veteran conference presenter Josh Berkus will go over his tech talk tips in detail in order to help you improve your presentation skills. Programmer and slide-slinger Ian Dees will take on the specific topics of showing code to an audience and composing your slides.
Speakers who are giving talks later in the conference are especially encouraged to attend.
by Tim Harder
With funding from the National Science Foundation and other generous sponsors, OSU’s Electrical Engineering and Computer Science Department set about creating an ultra-mobile hand-held device for distribution to undergraduates. Envisioned as a cutting-edge computing platform that would encourage students to tinker with all the latest developments in the mobile space without fear of breaking their own gadgets, the "OSWALD":oswald project had some fairly grand and noble ambitions: design all hardware and software in house with the help of student developers, thus making the device that much more hackable and learning friendly.
Sadly, things didn’t quite goes as planned for OSWALD and the project encountered a number of critical points of failure. In this talk, Tim explores the design and deployment decisions made while constructing the OSWALD platform, with an eye to providing lessons learned to the open hardware and open source educational community. He’ll also discuss OSWALD’s future and the steps taken to provide similar, useful student experiences in a more efficient fashion.
by Steven Shiau, Chenkai Sun, Yao-Tsung Wang and Thomas Tsai
The Debian Squeeze (6.0) was released in Feb, 2011. It contains about "35k software packages in the repository":http://packages.debian.org/squeeze/allpackages, and a total of nine architectures are supported. The most common architecture, i386, need 52 CDs, or 8 DVDs to put all the software packages. This is really a rich package resource for a developer to use. Therefore, with this resource and using "Debian live build":http://live.debian.net, it's very easy for a developer to create a specific GNU/Linux distribution. We, as the developers of "Clonezilla":http://clonezilla.org, "DRBL":http://drbl.org, and the maintainer of "GParted":http://gparted.sf.net live, create the Clonezilla live, DRBL live, and GParted live in this approach. These 3 special live GNU/Linux distributions are created especially for system imaging/cloning, diskless linux, and graphical partition editor, respectively.
In this talk, we will introduce Debian live build, why live distribution matters, cover the steps to create such a live distribution, and those things one might have to pay attention to, e.g. when a package is not included in the Debian repository, how to deal with. A live demonstration about creating such a live distribution will be given, too.
by Lev Tsypin
There are already some very established "Drupal distributions":http://drupal.org/node/326175 such as "OpenAtrium":http://openatrium.com (project management), "Managing News":http://managingnews.com (aggregation), "Drupal Commons":http://commons.acq... (social network) and many more. The dominent business model evolving around these products is providing commercial support, much like RedHat does for Linux.
But there are other avenues as well, including getting a set fee each time a distribution is used on a managed infrastructure stack such as "Acquia's":http://acquia.com/products-services and even "through acquisition":http://developmentseed.org/blog/2011/feb/22/open-atrium-and-managing-news-acquired.
"ThinkShout":http://thinkshout.com has a distribution for conservation groups called "WaterShed Now":http://drupal.org/project/watershednow that is already in use by over a dozen organizations and we're working on another one with Shomeya called "RedHen":http://redhencrm.com/, an open source CRM built natively with Drupal 7.
This talk will cover some light technical details about Drupal distributions and how they work, how some of the more prominent distros are fairing in the wild, and how ThinkShout hopes to capitalize on the one's we're working on.
by Fred Wenzel
Django is a great web application framework that allows for rapid web app development out of the box. Since Mozilla picked up Django in 2009, they've started over a dozen Django-based projects. For these sites to scale to an international audience of millions of users, bells and whistles were needed that a stock Django instance does not offer.
Playdoh is the result of months of Django-based web app development at Mozilla and solves common problems that each (or most) of these projects had to solve to get from a clean Django instance to something deployable. It is not a high-level collection of django "apps". Instead, our biggest concerns were scaling to support Mozilla's massive traffic (fast templates, caching, etc), localization for everything, and bullet proof security.
Here's an outline of the topics we're planning on touching in this talk:
* Web apps at Mozilla: Scale, scale, scale
* What Django does well
* What's missing (for us)
* How playdoh solves these problems:
* The future of playdoh
* Using and contributing to playdoh
In 2009 I inherited a poorly documented tsunami evacuation simulator and used it to study the impact of proposed city planning on tsunami survival in Longbeach, WA. Half of the project was to learn how to use the pile of hacked-together proprietary software and script I was handed. The second half was to baby-sit a Windows machine on which the simulator would reliably take extended naps.
In 2011 we expanded the study into a new community. Facing new data formats, bit rot, mysterious crashes, expired software, and imminent project failure we re-built the simulator using open-source technologies in two weeks. In addition to simply completing the project, this change has liberated us from specific operating systems and per-seat software licensing.
Outcome? Faster results, less expensive setup, and a transition from a per-simulation pricing model to one focused on finding the best possible result.
Come learn how the time we spent prying ourselves open (and free) allowed us to deliver a better analysis in less time, guarantee future users less pain, and change the dominant paradigm from proving preconceptions to exploring alternatives. I will even run an exploratory simulation live, so we can all experience the power of asking good questions.
Urban Airship started using MongoDB in early 2010 and began migrating data off of it almost exactly a year later.
This talk will detail the wins Urban Airship gained with MongoDB as well as the shortcomings that eventually led to other databases being preferred. The goal is not to encourage or discourage anyone considering MongoDB but rather to share one group's experience with the database while supporting an ever growing amount of data.
For those who saw the first iteration of this talk at Update Portland, the slides will be updated, improved, and new data on sharding, replica sets, etc. will be added.
"Ganeti":http://code.google.com/p/ganeti/ is a robust cluster virtualization management software tool. It’s built on top of existing virtualization technologies such as "Xen":http://www.xen.org/ and "KVM":http://www.linux-kvm.org/page/Main_Page and other Open Source software. Its integration with various technologies such as "DRBD":http://www.drbd.org/ and LVM results in a cheaper High Availability infrastructure and linear scaling.
This hands-on tutorial will cover a basic overview of Ganeti, the step-by-step install & setup of a single-node and multi-node Ganeti cluster, operating the cluster, and some best practices of Ganeti. Finally, deploying and using a web-based management tool called "Ganeti Web Manager":http://code.osuosl.org/projects/ganeti-webmgr.
If attendees want to participate in the optional hands-on portions of the tutorial, there will be virtual machine "images available online":http://ftp.osuosl.org/pub/osl/ganeti-tutorial and at the tutorial itself. It’s recommended you download the image prior to the tutorial to save on setup time. We’ll be installing Ganeti on an Ubuntu VM and deploy instances using the "LXC":http://lxc.sourceforge.net hypervisor.
This tutorial will cover the following:
# Installing the base system and components
# Setting up the environment for Ganeti
# Operating a Ganeti cluster
# Deploying Ganeti Web Manager
21st–24th June 2011