The competition which is being run by taxiadvertising.com briefed us to communicate why we believe the eyes of the world should be on a brand that we represent by using a taxi-side in the most innovative and modern way.

We thought that the launch of the 2012 Chevrolet Camaro Transformers Special Edition was something the world should know about.

We designed a fully liveried design for the Taxi, mimicking the iconic rally yellow Bumblebee character in the Transformers movies. The livery itself is an augmented reality hotspot with a call to action for people to film the taxi side (linking with a service such as Blippar or Aurasma). When filmed on a mobile, the taxi would transform in to the special edition Camaro and using the geo-location of the users handset, prompt the customer to hop in to the taxi and visit their local dealer. Dealerships would then be asked to refund the taxi fare for customers arriving in special edition taxis.

Smart huh? If you think so too, we’d love it if you could vote for our design by clicking here:

The site is built using WordPress; the world’s leading open source publishing platform. To improve concurrency we used the W3 Total Cache plugin, in addition to HyperDB. These two plugins along with some other hardware choices mean that we can scale to serve virtually any level of traffic.

Due to nature of the client (live radio), we knew that we’d have to build out an infrastructure that could scale to meet demand at a moments notice. This needed to be automatic, seamless and most importantly transparent – both to the end user and the sysadmin. There isn’t much point being alerted to a high load at 6 in the morning if no suitable action can be taken. The brief from the client was simple – the sites cannot go down, regardless of how much traffic is thrown at them.

We looked around for a solution and Amazons Web Services was it: proven, reliable, flexible and cheap. The only hurdle came with its flexibility; how do we go about easily setting up a platform that is robust and can handle large (and I mean large) bursts of traffic? After looking at various articles, we chose Scalr. This affords us full control of the AWS architecture, all from the comfort of an easy to use and powerful web based administration panel.

Scalr has the concept of creating a Farm; a cluster of one or more Roles. Each Role contains one or more Instances and can also be configured with settings specific to a role. Scalr also has a scripting engine, meaning we can run custom scrips at various points during the creation of a new instance. For example we checkout the latest version of the source code from a GIT repository automatically before an instance is put into production. In addition to custom scripting Scalr allows us to easily and automatically create and restore from backups aswell as creating snapshots of server configurations.

Each role features at least two instances (apart from the cache role) so that in the event that one instance fails, there is another to server requests until a new instance is automatically created.

Database – a master/slave MySQL database that runs on a minimum of two instances. Replication happens automatically and in the case that the master fails, the slave is automatically promoted as master with a new slave automatically instantiated. This was a pre-built feature of Scalr and was one of the deciding factors on using their service. Snapshots are automatically created and saved to an S3 bucket every few hours. Using the HyperDB plugin, writes are routed to the Master and all reads are served from the Slave database(s).

Cache – a dedicated instance to cache database queries, session data and page fragments. With 7GB of RAM, it’s sole purpose is to improve the speed of the application and reduce the load on the database. We opted for Memcached – it’s in use in many of the worlds most visited websites, is simple and was easy to integrate.

Load Balancer – Nginx acting as a reverse proxy was the basis for our load balancer. In addition we installed Varnish to add as a cache. Very few requests actually make it through to the backend under normal circumstances as they are either served by Varnish or Memcached. With regard to the default role, in addition to adding Varnish we changed the configuration to turn off ip_hash in Nginx. When this is on, traffic is routed to the app instances in order they appear in the backend servers config. This doesn’t really distribute load evenly. If you need instance persistence I’d strongly recommend using the memcached role to store session data.

Application instances – these run apache and have been modified to add drivers for memcached plus other enhancement tweaks. The only traffic that gets routed through are cache misses (low) and certain traffic that we don’t ever want cached (user specific data and settings).

Admin – these sites run a modified WordPress install hosted on a dedicated Linode VPS. There are a number of cron jobs that need to run at regular intervals to consume and import data from a range of third party sources. As the import and export of this data was date and time critical, we couldn’t risk this being interrupted by scale event.

Email – all transactional email is routed through the wonderful Postmark service, taking away the worry of deliverability and sending consistency.

CDN - all images, javascript files, css and swf’s are served from an Amazon S3 bucket to reduce the load on the application instances. This also removes the need for us to network drives together on the application instances as all assets are served from a single central location.

We have used the scaling features provided by Scalr to set parameters to automatically add new instances to each role primarily based on load. This means that if a surge of traffic hits the application, load can be met by automatically adding more power. In most cases the caching will actually prevent this from even needing to happen.

Setting up a scaling solution is only half of the solution – you need to ensure that you can actually meet the demand and scale when needed. You could wait for a real load to come, or you could prepare and invest in load testing. We opted to use Blitz.io to simulate high levels of load towards our infrastructure. Blitz allowed us to simulate various numbers of users visiting the site at differing concurrencies ensuring that the metrics we use to scale actually work. We consistently managed to serve over 600 requests/second without any issues.

Hopefully this blog post has given you an insight into some of the things that go into making a highly reliable, highly available website.

To meet the challenge, we created a bespoke system that prompts users to confirm their location when they land on the home page. This enables the new site to retain the ‘localness’ of each station prior to the merger whilst offering consumers an improved experience in terms of layout, navigation and functionality. The site has also been designed to enable consumers to share content from the site more easily with their friends via social networks, particularly Facebook.

New features include news and events on music artists and links to videos and downloadable tracks. Additionally, as the site has been developed to integrate seamlessly with sources such as YouTube, Flickr, SkyNews, The Metoffice and TrafficLink, visitors are able to access far more content tailored to their interests than before.

Andy Price, head of marketing and digital at Free Radio comments: “We were keen for our new site to forge a deeper link between our on-air and on-line content. The ideas 383 came up with look set to do just that; from the additional content to the restructuring of the site so that presenter blog posts are displayed in a stream, which is a familiar concept to Facebook users. All of the ideas they have given us make the site a more engaging offering for consumers and we are excited to see consumer reaction to the new site over the coming weeks.”

The website of a Radio station, particularly a commercial station, is now critically important to its success.  The site is where the majority of interaction occurs and provides a significant brand building opportunity for both the station and its advertising clients. However it is important not to let the advertising get in the way of the page content. This is what we focused on when developing the new site for Free Radio and we hope consumers will enjoy what we’ve come up with.

We’ll follow this post up with a more technical one on how we approached the development specifically, but for now we hope you enjoy the new site.

Photo by Paul Clarke

Data as material

Each of the Hackday briefs came with plenty of raw data to play with. I knew that the Goverment had been hard at work at opening up it’s vast store of data, but the scale of stuff that’s now avaiable on Data.gov.uk is amazing. Over 8,000 datasets and APIs are now available to tinker with, even more are added each month. This overwhelming scale means that it’s difficult to get a handle on what is useful. It feels like I could spent a month just doing data exploration. “What’s available? What’s significant? What’s interesting? What’s the scale? What’s feasible?”. With more and more becoming available it’s as important as ever to spend time hacking with raw material.

Photo by Harry Metcalfe

However, with some of the datasets there seems to be some barriers to get up and running quickly. It may be correct to publish huge linked-data-sparql-xml-api-interfaces, but with hackers, simplicity will always win. A Nice Day Out by Thomas Shadwell was an idea to build a simple application to show nearby beaches, plans had to be scrapped but not because of lack of data, but the complexity of using it. Many of the popular mapping and location APIs provided by Google, Twitter and Foursquare all allow you to simply supply a point + distance = results. The beach dataset required developers to use complex and arbitrary handlers to translate current location to lat/long convert it to a bouding box then translate this back to regular lat/long. Although Data.gov.uk has done a fantastic job of opening these data sets up, I think more needs to be done with making them both ‘future proof’ and ‘hacker friendly’.

Getting excited and making things with strangers

At any developer meetup the most interesting thing to me is always the people, rather than getting wired in on my own and churning out code for my ideas. I opted to work with people I’d never met before. Our team consisted of three pre-existing-hackers and three total strangers. New people means new technology (I gained exposure to my first ever Grails/Groovy/Java application) as this was totally alien technology to me, I couldn’t actually get the app running on my own laptop.

It was an interesting perspective working with new technology and total strangers. With most projects at 383 Project I’m normally able to dive straight in and fix any problems, but taking a step back and being excluded from the build process shed a light on the role of the Project Manager and Client Partner within a digital agency. How can these roles all work together while speaking different languages?

Observing the process

Working in someone else’s office is always interesting, you want to snoop around, look inside the draws and see how they tick. With the hackday taking part inside the new Government Digital Service office we had a chance to see exactly what they were working on. GDS has been pretty great at publishing thinking on their blog, but that’s because they seem to be pushing thinking directly to the walls of their office.

Almost all of the main walls are plastered with personas, design mantras and ways of working. It was fantastic to see how “the design principles” appears to have directly evolved from a wall in the office. Rebuilding the many hundreds of services that the UK Government offers online into a single domain is no easy task, but having witnessed a tiny chunk of their thinking I know we’re in safe hands.

The hacks

Some of the best hacks of the day:

Our tiny app Happy House which attempted to examine the costs of utilities when buying a property. We made it easier to search for energy efficient properties in your area. Can more data exploration allow consumers to purchase more energy efficient housing? Can this be used to encourage more landlords in the private sector to install insulation? How does government make this type of data available to third parties?

Loftify, a group buying service for families to make older relatives’ homes more energy efficient, allows family members to pledge money to support older family members with some help from the government. Using a kickstarter style model the money is only released if the total is reached and families are connected with locally qualified professionals to complete the process.

Bycycle barometer This hardware hack use near-time weather and tube data to figure out if you should take the bike or tube to work.

Developers developers developers

A year ago I launched a project called HydraHack to bring more developers together around Birmingham. Since 383 Project’s involvement and sponsorship we’ve grown to nearly 100 developers from all technology backgrounds: Java, Ruby, Postgres and even .net. We meetup monthly to share technology and ideas. If you’d like to join us you can find us on Meetup, we’re doing our first ever show and tell on May 1st and 383 Project will be sponsoring the coffee.

Supporting hacker culture by attending hackdays and meetups is important, you never know what greats ideas or projects will come out of it.

This week we have a team from 383 out in South Africa undertaking a week long photo and video shoot for a client. With the great weather we’re having in the UK, spare a thought for our man Sam, who’s been working in 33 C heat since Sunday. Keep your eyes peeled for the fruits our labour in the coming weeks.

We’re pleased to announce that 383 Project are working with brandRapport London as the digital and creative global partner agency for Polar. Polar is the worldwide market leader in heart rate monitoring for athletes and sports enthusiasts, producing a range of training devices for runners, cyclists and other sports. The company has 18 subsidiaries in different territories and manages a distribution network supplying over 35,000 retail outlets in more than 80 countries.

Working in conjunction with sports specialists brand Rapport 383 Project are responsible for brand development along with the creation of all global marketing material for a range of product launches, in-store POS and seasonal campaigns. Online this includes the development of Polar’s digital offering including website design, digital strategy and activation of the brands presence in social media.

As a first view of some of what we’ve been up to, here’s the promotional video for Polar’s new RCX3 device which we shot with our friends at Vermillion in Portland last October.

Whilst we can’t give too much away until the Show, we can tell you that young Chevy fans can look forward to a packed schedule featuring the European premier of two exciting new concept cars, the Code 130R and Tru 140S, and a chance to vote for their favourite at the show and online. The Chevrolet Cruze Station Wagon will also be making its’ world debut, as well as the launch of the MyLink infotainment system, which integrates customers’ smartphones seamlessly with their cars.

All this means we can’t wait to pack our bags! We’ll have some photos and videos to share with you after the Show, but for now, here’s a quick teaser of what’s in store for fans choosing their favourite Chevrolet concept car. It’s a difficult choice!

Follow Chevrolet Europe for updates from the Geneva Motor Show:

Facebook

Twitter

The first major change is a visual one, a complete overhaul of the old pages layout which is now more in line with timeline for personal profiles. There’s a cover photo at the top with the body of the page split into two columns with a chronological dateline splitting the two. Key milestones attributed to dates will give you the opportunity to build a presence around a narrative and history in a way that hasn’t previously been possible. The New York Times for example have produced a rich visual timeline of their history and archives all the way back to 1857.

There’s also more prominence given to applications in the new format (there’s no concept of tabbed pages in the new format). Apps will no longer sit in a long list down the left of the page but will be displayed proudly right underneath the cover photo (by default, there’s only space for three, to see more, users will have to use a drop down option). Now that brand owned apps will be the thing that users are likely to stumble upon first, it’s important to think about which ones to hero. We think this will result in much more traffic to branded applications and that page owners will need to up their game in terms of the interactivity and narrative of them.

That’s not the only change to apps, from March 30th, you’ll no longer be able to set a default landing page. Facebook ads will be the only way to direct users directly to an application in future. It’s also important to remember that the new timeline layout effectively outdates existing tab content and applications. New application icons will need to be created and applications will need to be built to a wider format. Coca Cola have already changed which apps they hero as well as the more technical aspects of them.

Because directing users to a default landing tab is no longer an option, it’ll be important to present your brand in the most engaging way possible in the timeline. A new feature which will help do that is the ability to create ‘sticky’ pieces of content at the top of the timeline that are permanent pieces of content that don’t shift with date. This will be really important in giving focus to a big promotion or story.

The last big change puts a big tick in the customer service box. Brands will now be able to send and receive private messages with fans. This means that communication and conflict resolution/feedback can happen within Facebook without having to ask users to e-mail or call.

Sure, these changes represent a unified design standard across Facebook which is far more appealing than wall tabs which exist now, but more than that, it’s about storytelling. The new timeline helps brands to be better at this.

Over the past few months we’ve been developing the identity and online presence for a major new partnership between the Royal Shakespeare Company and the University of Warwick. Teaching Shakespeare is a global education programme designed to, as the name suggests, assist teachers in bringing Shakespeare to life more effectively in the classroom.

The overall look and feel has been designed to reflect the core values of the project – a vibrant, new approach to the modern day relevance of Shakespeare’s texts. As well as the ident and web design we’ve also helped weave some motion graphics in to the promo video. The overall design focus was on ensuring that the key information for each of the different types of courses was easily discovered and that the subsequent pages were structured in an engaging and logical way.

It’s been a really fun project to be involved with and we wish the teams at the RSC and University of Warwick every success with the launch.

To celebrate we created a video on a Facebook tab charting the journey of the page from its’ creation, up to the present. We wanted to thank the page’s fans so we pulled in avatars of Bullring’s most avid commenters to help us to create a narrative which represented the voice of the community.

Bullring are supporting the activity with lots of exciting giveaways over the next few days, so make sure you “like” their page to be part of it!

You can watch the video below: