Canonical Voices

Posts tagged with 'recent work'

Carla Berkers

OpenStack is the leading open cloud platform, and Ubuntu is the world’s most popular operating system for OpenStack. Over the past two years we have created a tool that allows users to build an Ubuntu OpenStack cloud on their own hardware in a few simple steps: Autopilot.

This post covers the design process we followed on our journey from alpha to beta to release.

Alpha release: getting the basics right

We started by mapping out a basic Autopilot journey based on stakeholder requirements and designed a first cut of all the necessary steps to build a cloud:

  1. Choose the cloud configuration from a range of OpenStack optionsChoose cloud configuration
  1. Select the hardware the cloud should be built on
    Select the hardware
  1. View deployment status while the cloud is being built
    View deployment status
  1. Monitor the status and usage of the cloud
    Monitor Cloud

After the initial design phase Autopilot was developed and released as an alpha and a beta. This means that for over a year, there was a product to play around with, test and improve before it was made generally available.

Beta release: feedback and improvements

Providing a better overview: increased clarity in the dashboard

Almost immediately after the engineering team started building our new designs, we discovered that we needed to display an additional set of data on the storage graphs. On top of that, some guerilla testing sessions with Canonical engineers brought to light that the CPU and the storage graphs were easily misinterpreted.


After some more competitive research and exploratory sketching, we decided to merge the graphs for each section by putting the utilisation on a vertical axis and the time on the horizontal axis. This seemed to improve the experience for our engineers, but we also wanted to validate with users in usability testing, so we tested the designs with eight participants that were potential Autopilot users. From this testing we learned to include more information on the axes and to include detailed information on hover.

The current graphs are quite an evolution compared to what we started with:
Improved dashboard graphs

Setting users up for success: information and help before the process begins

Before a user gets to the Autopilot wizard, they have to configure their hardware, install an application called MAAS to register machines and install Landscape to get access to Autopilot. A third tool called Juju is installed to help Autopilot behind the scenes.

All these bits of software work together to allow users to build their clouds; however, they are all developed as stand-alone products by different teams. This means that during the initial design phase, it was a challenge to map out the entire journey and get a good idea of how the different components work together.

Only when the Autopilot beta was released, was it finally possible for us to find some hardware and go through the entire journey ourselves, step by step. This really helped us to identify common roadblocks and points in the journey where more documentation or in-app explanation was required.

Increasing transparency of the process: helping users anticipate what they need and when configuration is complete

Following our walk-through, we identified a number of points in the Autopilot journey where contextual help was required. In collaboration with the engineering team we gathered definitions of technical concepts, technical requirement, and system restrictions.

Autopilot walk-through

Based on this info, we made adjustments to the UI. We designed a landing page  with a checklist and introduction copy, and we added headings, help text, and tooltips to the installation and dashboard page. We also included a summary panel on the configuration page, to guide users through the journey and provide instant feedback.


GA release: getting Autopilot ready for the general public

Perhaps the most rewarding type of feedback we gathered from the beta release — our early customers liked Autopilot but wanted more features. From the first designs Autopilot has aimed to help users quickly set up a test cloud. But to use Autopilot to build a production cloud, additional features were required.

Testing without the hardware: try Autopilot on VMware

One of the biggest improvements for GA release was making it easy to try Autopilot, even for people that don’t have enough spare hardware to build a cloud. Our solution: try Autopilot using VMware!

Supporting customisation:  user-defined roles for selected hardware

In the alpha version a user could already select nodes, but in most enterprises users want more flexibility. Often there are different types of hardware for different roles in the cloud, so users don’t always want to automatically distribute all the OpenStack services over all the machines. We designed the ability to choose specific roles like storage or compute for machines, to allow users to make the most of their hardware.

Machine roles

Allowing users more control: a scalable cloud on monitored hardware

The first feature we added was the ability to add hardware to the cloud. This makes it possible to grow a small test cloud into a production sized solution. We also added the ability to integrate the cloud with Nagios, a common monitoring tool. This means if something happens on any of the cloud hardware, users would receive a notification through their existing monitoring system.


The benefits of early release

This month we are celebrating another  release of OpenStack Autopilot. In the two years since we started designing Autopilot, we have been able to add many improvements and it has been a great experience for us as designers to contribute to a maturing product.

We will continue to iterate and refine the features that are launched and we’re currently mapping the roadmap for the months ahead. Our goal remains for Autopilot to be a tool for users to maintain and upgrade an enterprise grade cloud that can be at the core of their operations.


Read more
Robin Winslow

On release day we can get up to 8,000 requests a second to from people trying to download the new release. In fact, last October (13.10) was the first release day in a long time that the site didn’t crash under the load at some point during the day (huge credit to the infrastructure team). has been running on Drupal, but we’ve been gradually migrating it to a more bespoke Django based system. In March we started work on migrating the download section in time for the release of Trusty Tahr. This was a prime opportunity to look for ways to reduce some of the load on the servers.

Choosing geolocated download mirrors is hard work for an application

When someone downloads Ubuntu from (on a thank-you page), they are actually sent to one of the 300 or so mirror sites that’s nearby.

To pick a mirror for the user, the application has to:

  1. Decide from the client’s IP address what country they’re in
  2. Get the list of mirrors and find the ones that are in their country
  3. Randomly pick them a mirror, while sending more people to mirrors with higher bandwidth

This process is by far the most intensive operation on the whole site, not because these tasks are particularly complicated in themselves, but because this needs to be done for each and every user – potentially 8,000 a second while every other page on the site can be aggressively cached to prevent most requests from hitting the application itself.

For the site to be able to handle this load, we’d need to load-balance requests across perhaps 40 VMs.

Can everything be done client-side?

Our first thought was to embed the entire mirror list in the thank-you page and use JavaScript in the users’ browsers to select an appropriate mirror. This would drastically reduce the load on the application, because the download page would then be effectively static and cache-able like every other page.

The only way to reliably get the user’s location client-side is with the geolocation API, which is only supported by 85% of users’ browsers. Another slight issue is that the user has to give permission before they could be assigned a mirror, which would slightly hinder their experience.

This solution would inconvenience users just a bit too much. So we found a trade-off:

A mixed solution – Apache geolocation

mod_geoip2 for Apache can apply server rules based on a user’s location and is much faster than doing geolocation at the application level. This means that we can use Apache to send users to a country-specific version of the download page (e.g. the German desktop thank-you page) by adding &country=GB to the end of the URL.

These country specific pages contain the list of mirrors for that country, and each one can now be cached, vastly reducing the load on the server. Client-side JavaScript randomly selects a mirror for the user, weighted by the bandwidth of each mirror, and kicks off their download, without the need for client-side geolocation support.

This solution was successfully implemented shortly before the release of Trusty Tahr.

(This article was also posted on

Read more
Katie Taylor

Edges are special to us. We use them for finding apps, tools and system services, so using the edges will be second nature to Ubuntu phone users. By using the launcher, how to launch your favourite app will become ingrained in your muscle memory of the left edge.

The design vision behind Ubuntu for phones includes the use of fast and natural interactions, so taking that to the welcome screen means that if your phone is locked, you can still access the launcher, system services and the right edge. If you have a pin set up, you only need to enter your pin when accessing private data, in the Gallery app or the Dash for example.



If you’ve flashed your phone recently, you will be able to activate the lock screen for the phone using a temporary hack (love it!). You’ll notice that the blur has not yet been implemented, but will be added later. Thanks to Michael Zanetti for originally posting instructions to the Ubuntu Phone mailing list. Here they are:

To enable the pin lock, log into the phone and create a file /home/phablet/.unity8-greeter-demo, with the content: password=pin

If you want to see the password unlock screen instead, put this into the file:  password=keyboard
For now, the pin is hardcoded to “1234″ and the password is “password”. Note that this functionality can (and will) disappear at any time as we bring all the bits and pieces together. This is a temporary, simple way to enable the visual part of the lock screen for us all to have a play with.

Let us know what you think on the Ubuntu Phone mailing list and the IRC channel.

Read more
Anthony Dillon

We realise that changing Operating System (OS) is a big thing for anyone thinking of testing something out. That becomes a huge barrier for people trying out Ubuntu for the first time and seeing if they like it. As a member of the web team I decided to take on the challenge as a cool way to testing out some HTML5 and jQuery. The purpose of this blog is to talk about some of the challenges and thought processes I went through during the build.

Getting started

I started by breaking down the Ubuntu interface into DOM elements, similar to developing a web page from a design. So I took on the background first, each part of the build brought different challenges. In the case of the background, I needed it to stretch full width and height of the viewport.

The launcher

Once I had that nailed, I moved onto the launcher on the left of the screen. At the time of building (11.10) the menu would slide out of view when a window was fullscreen. By creating a DIV with a list of icons and styled with CSS to replicate the menu in the OS. Now I needed to add the interactive slide to the menu, which I decided to apply with jQuery because I wanted the animation to work on older browsers and not just browsers with CSS3. When a window is fullscreened I would trigger the menu system to slide the menu out of the viewport. Then binds a mouseover event to slide the menu back into view when the user needs it again.

Precise launcher

The global menu bar

Precise global menu bar

All that was left to make the web page look like the static Ubuntu interface was to add the global menu bar at the top for settings and window controls. This was again a simple DIV with a window control DIV floated to the left and a list of settings icons floated to the right.

Building applications

Now the interface looked like the initial Ubuntu screen. I turned to focus on applying the functionality to the menu icons, beginning with the folder icon. Created a Folder class and a File class, Folder class below:

function Folder($name, $location){
  if ($name==undefined) { $name='Untitled Folder';}
  if ($location==undefined) { $location='/Home';}
  var _name = $name;
  var _size = '6.2 GB';
  var _location = $location;
  var _in_bin = false; = function (){ return _name; };
  this.size = function (){ return _size; };
  this.location = function (){ return _location; };
  this.in_bin = function (){ return _in_bin; };
  this.rename = function($name){
    _name = $name;

  this.bin = function($in_bin){
    _in_bin = $in_bin;

  this.move = function($location){
    _location = $location;

  this.drawIcon = function($id, $type){
    return  "<div class='file-folder "+$type+"' data-type='folder' data-id='"+$id+"'><p></p><span>"+_name+"</span></div>";

  this.type = function(){
    return 'folder';

Both of these classes are used by the file system class that created them and stored them in an array to access and modify them later.
Now that we have a file system, the fun part comes when you connect it all up and see if it comes to life. This is how the connections happened:

Menu window workflow

When the folder icon on the menu is clicked it tells Base which triggers the open function in the folder system. I built the Folder and File classes with the aim to give the folder system full functionality, like rename, drag and drop to move, etc. But that was descoped to finish the tour in time for the release of 11.10.

Seeing this work I felt the buzz to jump to the next application, then the next, until all the applications I set out to develop were finished and working.

Have a go for yourself: Ubuntu Online Tour


One of the questions I was asked is `how long did it take you?’ In total it took about 4 months, from what started as a quick test to being given the time to bring the development to fruition. It was crucial that all elements of the demo looked and felt exactly like the OS as this will be most people first try of Ubuntu so we didn’t want it to be different if they install Ubuntu.

I plan to keep the tour up-to-date with the latest version of Ubuntu every six months. With every update I try to add features and application which I haven’t managed yet. A few features that come to mind would be to progressive enhance the movie player to use HTML5 video if available rather than Flash. I would also like to see the music player (Rhythmbox) be developed.

If you have HTML, CSS and jQuery skills and time to spare. Please feel free to contribute. The code can be forked from: Ubuntu Online Tour Launchpad Page.

Read more