There have been some subtle changes to the site that havn’t really been advertised. I’ve taken a moment to outline them for you:

Mobile Site Revamp
There were two major changes to our mobile website. The first was the outsourcing of the site to a third party called Quattro Wireless. They allow us to automatically reformat the mobile site depending on which device you are using. So the site you see on your Blackberry will be formatted for the Blackberry. While a visitor from a cell phone will have the cell phone version of the site displayed to them properly.

The second change we made was to automatically detect if users are coming from a mobile device if they visit www.cbc.ca. If they are, then they will automatically be directed to our mobile site: m.cbc.ca. There were a few bugs with this auto-redirection, but they have been ironed out this week.

This way, you have one url you need to remember: cbc.ca. If you visit from a phone, blackberry, or desktop, you will automatically be sent to either the mobile or regular website.

Flash Video
We have had flash video on the site for a while now. We’ve recently rolled this out to the news areas of the site, specifically The National.

This new video format allows us to provide higher quality video at a lower bit rate. It also allows us to almost guarantee that the video will be played on all platforms as the flash format is supported in most browsers.

You can expect to see Flash video to slowly replace all the windows and real media video on the site. We are currently testing live streaming with the flash player as well.

Anything Else?
There are some pretty cool projects going on that I will reveal over the course of this year as they roll out. So stay tuned!

	28 Comments »	Email This Post
	, CBC.ca web site, Under the Hood	Posted at 1:31 pm (06 Feb 2009)

CBC.ca has moved from the second floor up to the ninth floor over the course of the past two weeks. Us, along with the finance department were the first “guinea pigs” for the office space redesign.

Because this is going to be our permanent home, we were literally given keys to the blueprint. After playing around with little chairs, tables, and desks on a floor plan the workers starting building our “dream office”. In 7 months or so, they were complete!

You can see a few select photos of our new space from my co-worker Julian Dunn on his flickr page. Additionally what our old space looked like is also available on flickr.

	13 Comments »	Email This Post
	, CBC.ca web site, Under the Hood	Posted at 12:18 pm (22 Jan 2009)

Early Monday morning (01:00-05:00 ET Monday January 12, 2009) we will be taking the site off-line to perform some much needed maintenance on our storage device.

The CBC.ca website uses a lot of disk space. All of the media, news stories, podcasts, etc.. add up. We store all of this information on device produced by a company called BlueArc. All of the web and application servers “mount” the BlueArc to serve the website content. There is approx. 3 terabytes of data that is used for the website. 3 Terabytes is a lot of data, to put it in perspective:

1 Terabyte would be equal to 50,000 pages of printed material.
1 Terabyte of seconds would be 32,000 years.
850 Terabytes of storage is used for all of the pages in the Google search index
2,250,000,000 Terabytes of data in 1 gram of DNA

What will I see during the outage?
Because we need to take the BlueArc offline to perform the maintenance, all of the pages in CBC.ca will not be available. However, we are going to be providing an extremely limited site (what we’re referring internally as the “skinny site”). This site will be extremely light-weight and only offer the latest news.

A preview of the site during the maintenance.

All other content will be unavailable until the maintenance is complete.

If you have any specific questions on how the BlueArc works, or why we’re doing this maintenance, please feel free to post a comment.

	3 Comments »	Email This Post
	CBC.ca web site, News & Journalism, Under the Hood	Posted at 3:42 pm (08 Jan 2009)

In part two of two of my “Federal Election” series we are going to talk about two items. The first, which is the brain child of David Raso, is the postal code/riding look up. For that, I’m going to hand the keyboard over to him:

My Ridings
Hi, I’m David Raso a Senior Architect of Front-end Development at CBC.ca and Blake asked me if I could explain how one of our Election modules work. We wanted to build a module which would show you information and results for the riding you live in, so the My Riding module was created. This module allows you to enter your postal code and then displays information about your riding - it’s on every page of the Canada Votes site.

No Database Needed!
Sounds simple enough right? The simplest solution would have been to hook up the form to some sort or dynamic backed code written in Java or PHP which would search though the postal code database we got from Statistics Canada and display your riding info. Well because we get so much traffic on election night we have to build everything to be extremely stable and scalable. So using a dynamic application was out of the question - so then how do you provide a postal code look up database without a database? We could use some cloud computing (like Amazon or Google) or big complicated Akamai cached edge applications - but we like to keep things simple.

Everyone in meetings would say: “It can’t be done”, then you sit and stare at the white board. Few minutes later we came up with one of those “It’s so crazy it just might work” ideas.

When you enter your postal code into the form and hit search - we don’t connect to a database and find that postal code - we change your postal code into a url and using AJAX (a JavaScript programing technique) we fetch a flat HTML file which contains the ID of your riding. We then use that ID and AJAX, again, to fetch a flat HTML of your riding info. This is all done by your browser so all our web servers have to do is return those tiny HTML files your and computer does all the heavy lifting.

How It Works
So if you enter M9A2X1 we change that into “http://www.cbc.ca/news/canadavotes/myriding/postalcodes/m/m9a/2×1.html” we then download that file using AJAX. The files contents are stored as JSON (JavaScript Object Notation) which the My Riding JavaScript can easily understand and obtain the ID for your riding. Once we have your riding ID we then use AJAX again and download the file for your riding “http://www.cbc.ca/news/canadavotes/myriding/ridings/130.html” then all we have to do it display that HTML file in the My Riding module. Easy Peasy, Lemon Squeezie.

That’s all good and it works but you might be asking your self: “Does that mean you have a HTML file for each and every postal code in Canada?” - yup you’re correct. Over eight hundred thousand - 827,018 files to be exact. We took the database from Statistics Canada and created a script that would turn each postal code into a flat HTML file that contains the name and ID for the ridings in that postal code - some postal codes have multiple ridings, try out K0A1W0. We then uploaded the 3GB folder to our webservers - and because they don’t change we can cache all those files using Akamai.

On election night when the hundreds of thousands of users try to find their riding there is no database to go down, or application server to be bogged down. Just one small cached HTML file to download!

Comments and the Black Out
This is the first Election where commenting was allowed on CBC.ca. One of the issues we faced were users posting election related results in their comments.

This was an easy solution. We asked the 3rd party company that is responsible for moderating our comments (called ICUC) to not approve any comments for publication until the blackout period ended.

	Add Comment »	Email This Post
	CBC.ca web site, Under the Hood	Posted at 3:50 pm (05 Dec 2008)

The latest federal elections introduced some new tools for site visitors to use on our site.

You were able to use your postal code to get riding information, view real time results on an interactive map, view live video, and talk politics by posting comments.

I will be talking about these applications in detail in the next few posts.

Blackout
Providing these features proved to be a challenge. The Elections Act doesn’t allow us to broadcast results to areas of the country where the polls haven’t yet closed. Therefore we were not publishing any results on the website until 10PM ET. This is known as the “blackout” period.

Flash Crowd
Because of the blackout wasn’t over until 10pm ET, that’s when the majority of users came to the site. Let’s take a look at the traffic graph for that particular evening.

The date/time in the graph is in Eastern Time. To give you an idea of the scale of the graph the top of the Y axis would be a value 3.8 times higher than at the bottom. The red arrow indicates the peak, which lasted for approx 10 minutes. As the evening progressed the traffic subsided.

“Elections Mode”
Because of the blackout we had to figure out a way to display results on the site the instant 10PM rolls around. Here’s how we did it:

All of the site assets were uploaded to the webservers hours in advance. We were able to control the display of these assets using Server Side Includes (SSIs).

SSIs allows us to hide or display certain html based on the existence of a variable by using if statements. Here’s an example:

<!–#if expr=”${electionsmode} = on”–>
<results related HTML goes here>
<!–#endif–>

We include a file called “vars.html” at the top of every page which defines the “electionsmode” variable. When it’s time to go live, we just change the “electionsmode” variable in vars.html from “off” to “on” and the new content appears instantly on the site.

Getting The Data
The elections results are delivered to us from another IT department. It is the exact same data feed that is used on-air. We process the results every 30 seconds and spit out XML data for our applications to use. The process works like this:

1. IT FTPs the raw election data results to our server.
2. CBC.ca’s Java application sucks in the raw data.
3. The application applies business rules, calculates percentages, then outputs XML files.
4. Process is repeated again in 30 seconds.

The XML files are used for the flash map and riding result pages. Because this process only happens every 30 seconds, you might see a delay from what you see on Television. This is because TV does not have this post-processing and transforming that we do.

The Flash Map
The flash map was developed by Tom Klepl, a very talented flash developer. He worked closely with the system administrators to ensure that the map was designed in such a way that it played well with our CDN (Akamai).

The map allows users to zoom in to their riding and see real time results of who is winning the race. As well, it gives a general overview of all the ridings in Canada. One of our interface engineers, David Raso, produced the following time lapse video of how the country voted on election night.

In Part Two
In my next post, we’ll go into further detail on how the postal code search works (hint: no php, java, or database required!) as well as how we dealt with users who would post election results in comments!

	2 Comments »	Email This Post
	, CBC.ca web site, Under the Hood	Posted at 4:09 pm (20 Nov 2008)

Sadly this is the last Olympics that CBC will be covering for a while. That said, this is also the most well covered Olympics in CBC history. Partly thanks to the Internet and cbc.ca.

Don’t Feed The Animals
cbc.ca/olympics provides you with up to 12 online “channels” where you can watch Olympic events live. This is possible thanks to the folks in Television (specifically Web Presentation group), encoding software by Digital Rapids called StreamZ, and Akamai.

The Olympic feeds are back hauled by Television to Toronto where they are encoded by the Digital Rapids boxes. They are encoded in Windows Media v8 at 500kbps. This stream is then sent to Akamai for distribution.

In most cases the video you see are raw feeds from the venues. This means that there is no commentary. The only audio you here is ambient sounds from the event.

A lot of people like these feeds as they provide coverage for their sport even if it is not available on television.

Canada Only, Please
Due to licensing restrictions by the IOC, only Canadians are able to view Olympic coverage provided by the CBC. As such, we’ve had to use technology from Akamai to ensure that we follow the rules.

The Akamai method of “geofencing” (as we call it) is more sophisticated than what I described earlier. Multiple methods are used, some of them are:

• Your IP Address. Using a database of known IP blocks and locations.
• Which Akamai DNS server you use. When you look up a host name (like www.cbc.ca) you hit a specific Akamai DNS server. Akamai knows which DNS server is in which country and uses that information to figure out what country you’re in.
  BGP Metrics. In a nutshell, Akamai looks at what other servers/routers you go through to get to the stream. If those are in the country Akamai thinks you’re in, then the confidence level goes up.
• Timezones. Using a piece of javascript on the client side. We figure out what timezone your clock is set to. If it matches with one of the timezones in the country Akamai thinks you’re in, confidence level goes up.
• We have people who watch you.

The methods listed above, plus a bunch of others combine to give a Country accuracy level of 99%.

Trends
Because of the twelve hour difference (in the Eastern Time Zone) quite a few of the events happen late at night and early the next morning.

The above graph shows the number of people watching the streams over the past seven days. The majority of the watching happens between 8pm and 12am EDT, and 6am and 12pm EDT.

Some days are more popular than others (like August 14, versus August 16). The little red arrow you see on August 19, at around 12pm EDT is when the most users were watching the streams ever.

I suspect that the closing ceremonies will generate the most streaming traffic during the Olympics.

	12 Comments »	Email This Post
	CBC.ca web site, Specials, Under the Hood	Posted at 2:11 pm (20 Aug 2008)

This past week we released a new version of the CBC Program Guide. This was a much needed upgrade as the old version was virtually useless because it constantly crashed.

The Nitty Gritty
This new version is written in Java using Struts and Hibernate. These are enterprise frameworks that eliminate the need to create custom libraries. These “custom libraries” are what caused a lot of the instability with the previous version of the Program Guide.

The previous version stored all of its data in an Oracle database, with no expiry policy. This means that you could go back years to see what was on CBC Television in 2003, for example. The new version stores all of its data in a PostgreSQL database with an expiry policy. There really is no need to know what was on CBC.

The new guide is also extremely extensible. Able to output in multiple formats including HTML, XML, and JSON. This allows other CBC developers to leverage the data located in the guide.

Quick Rundown
Program Guide information is available for all CBC properties: Television, Newsworld, Radio One, and Radio 2. It also includes A to Z guides of all CBC Programs and Personalities. The new guide has been redesigned so that it is easy for you to view what is currently on air right now, which is highlighted in blue.

You are able to “segment” your day into early morning, morning, afternoon, or evening. So you only can only see 5-8 hours blocks at a time. You are also able to view the full day, or a schedule for the entire week. Clicking on a show title will bring you to the program page which will allow you to see air times, plus a description about the show and its personalities. You can also filter your schedule by program category. So you can only see Sports, Comedy, Drama, etc.

The print friendly version of the guide is well formatted and easy to read.

The Future
With the new Program Guide framework we are able to provide a lot of new features. Some ideas floating around include RSS feeds of your favorite show (air times, descriptions, etc..). The ability to include program/personality information in our search engine is also a possibility.

The new Program Guide will be used during the Olympics to allow you to know exactly what event will be on-air when. You will be able to access this information from the Olympics page or the Program Guide main page.

Expect to see more features and pages that utilize the new Program Guide in the near future!

	6 Comments »	Email This Post
	Under the Hood	Posted at 2:34 pm (22 Jul 2008)

And… farewell Blake.

The following is the last of the “Under the Hood” columns that have appeared on this blog for more than a year, courtesy of CBC.ca tech guru Blake Crosby.

Blake worked with CBC.ca for almost six years, first joining the team to babysit the servers during the Salt Lake City Olympics. He went on to work on other Olympic and elections sites, among others, and won an award for his work on CBC.ca’s Media Resource Locator tool (see his earlier column.)

Blake has moved on to a company called VerticalScope. Long term, he’s working toward a career in aviation – you can track his progress on his flying blog.

Thanks, Blake!
~PG
————–

Farewell Webtrends, Hello Hitbox!

There have been some behind the scenes changes in the way we process and crunch the web server log files.

Webtrends
The software we were using previously was called Webtrends. It processes the raw log files from the web servers and produces graphs and charts.

The main advantage to using Webtrends is the fact that it processes the raw web server logs. Anytime someone fetches content from our web servers, it is recorded in a log file. Whether this be a mobile phone, Internet Explorer, your grandma’s 386, or your text only browser - it’s all tracked.

Items such as your IP address, the page you were requesting, the type of browser you were using, and the date and time were recorded. This provided a solid source of data to process.

One of the downsides was the Webtrends limitation that the log files needed to be in chronological order. This is impossible with our website, as we have many different log file sources that are all out of order. There was a lot of overhead to merge all these log files into a chronologically correct source of data for Webtrends.

Changing Business Requirements
With the recent “upgrade” of the internet to Web 2.0, CBC needed to upgrade their website with more “Web 2.0″ features. This included items such as the most viewed stories, or most e-mailed stories. This real time data was available from the web server logs, but Webtrends couldn’t process the data fast enough for it to be useful.

This is where Hitbox, our new system, shines. The Hitbox product comes from a company called Visual Sciences (formerly WebSideStory.) It works the same way as Webtrends, except it offers real time data of people visiting the website. This is not done using log files, but javascript instead.

For every single page you visit on CBC.ca, a cookie for “a.cbc.ca” will be set. This cookie is used by Hitbox to track your movements throughout the website, and is recorded in real time. Although no identifiable information is recorded, we can see how individual users use the website.

That means content producers can track the performance of various areas of their sites in real time - understanding what stories are most popular, the times of day with heaviest usage, the most common navigation paths through the site, what links users follow to and from stories, and so on. By watching specific live stats instead of waiting for a report, the content itself can better reflect users’ actual behaviour.

	7 Comments »	Email This Post
	Under the Hood	Posted at 11:16 am (28 Nov 2007)

It seems that computer programmers like to use Hawaiian names to describe their applications. We use two such applications:

1. Media Wiki (”quick” in Hawaiian is “wikiwiki”)
2. Akamai (Hawaiian for “intelligent” or “smart”)

CBC has their own wiki and uses the same software that powers The Wikipedia called Media Wiki. Anyone and any department is welcome to use it. We currently have 259 articles and growing.

CBC Wiki front page. Click for larger

Some of its uses:

• Training guide for new hires
• Personal pages for employees who use them as note pads or a place to keep track of project progress
• Information about infrastructure, policy, phone numbers, etc..
• A listing of standards to be used by content producers

The wiki is only available to CBC employees at wiki.tor.cbc.ca . It is pretty technical as only the IT departments are currently using it but should be useful to other departments as well including Radio and TV.

	1 Comment »	Email This Post
	CBC.ca web site, Under the Hood	Posted at 2:51 pm (07 Aug 2007)

Manging multiple machines or accounts involves remembering quite a few usernames and passwords. Every member on the team also has to use the same usernames and passwords as well. The challenge is to find a secure way of storing these login credentials so that everyone has an easy way to access them.

Originally we stored all the usernames and passwords in what we called the “Password Book”. This book was a simple notepad with the pertinent information stored in a cabinet close by. The main problem with this system was that the information was not available unless you were actually in the office. This made on-the-road troubleshooting almost impossible.

Currently we store all of the information in a blowfish encrypted database. The database is stored on a Wiki page (which I will talk about in another post) where it can be downloaded and stored on a usb key.

This database is then read by a program called Password Gorilla. The advantage to using Password Gorillia is that the client is available on a multitude of platforms, including mobile ones like Windows CE (but not Black Berry).

In reality, we just need to remember one password: the password to the encrypted database, and we have access to all the usernames and passwords we need in order to do our jobs.

	10 Comments »	Email This Post
	, CBC.ca web site, Under the Hood	Posted at 9:54 am (10 Jul 2007)

… 82% of the time.

One of the cooler features of the new site design is the weather “widget” in the top left hand corner of the page. Some people might have noticed that the weather report being displayed is for the major city that is closest to you. How did our website know what city you are in?

It’s All About The IP Address
Every computer on the Internet has an IP address. These set of numbers allow networks to send the data you requested to your computer and not the one down the street. The IANA is the organization that keeps track of IP addresses on the Internet. They assign a block of addresses to 5 different portions of the world: Africa, Asia/Pacific, North America, Latin America and the Caribbean, and Europe, Middle East, and China. The fact that this division is made is the first step in figuring out your physical location based on your IP address. At this point, we can narrow down which region of the world you are located. Obviously not accurate enough to use in our application.

In each one of these regions there is a organization that then redistributes IP blocks allocated to them by the IANA to various people or companies in that region. For our discussion, we will focus on ARIN (American Registry for Internet Numbers) which handles IP addresses in North America. ARIN would be similar to a telephone company that distributes telephone numbers when you move to a new house. Except in this case, they hand out IP addresses to ISPs and Corporations.

It’s Public Information
ARIN (as do all “registries” as they are called) require that “owners” of IP addresses have a physical address on record. This information is available to the public at various locations.

Lets take a real world example: My IP address for my computer at home is: 24.137.199.17. If I type that address in the ARIN website, I get the following result:

Aurora Cable Internet ACI2 (NET-24-137-192-0-1)
                                  24.137.192.0 - 24.137.223.255
Aurora Cable Internet HE-24-137-192-0-19 (NET-24-137-192-0-2)
                                  24.137.192.0 - 24.137.223.255

Great! I know that a company called “Aurora Cable Internet” is responsible for my IP address and any other IP address that is inside the ranges listed above.

The string after the company name (”ACI2″ and “HE-24-137-192-0-19″) will be able to provide us with more information about that company. Plugging in “ACI2″ into the same search box on the ARIN site reveals (among other information):

OrgName:    Aurora Cable Internet
OrgID:      ACI-38
Address:    350 Industrial Parkway South
City:       Aurora
StateProv:  ON
PostalCode: L4G-3L6
Country:    CA

Now we all know that I live in Aurora, ON! More importantly though, we have just associated a physical location with a IP address.

What’s this 82% business?
The database that we use is only 82% accurate within 40kms of true location.

So what happens if you are in the 18%? The weather defaults to Toronto (further proof that we’re the centre of the universe). From there you can click on the “change city” link to save a new location.

I live in Canmore, but the weather for Calgary shows up. Why?
We’ve only discussed one part of how this system works. We can now identify which city you are probably located in. However, we have weather data for hundreds of cities across Canada and we need to determine which city with a weather station is closest to you.

Currently the weather widget only knows about 21 cities in Canada (we will be expanding this shortly). To find the closest city with weather station to Canmore we use a calculation called “great circle distance“. Put simply, its the distance between two points on a sphere (the sphere in this case is the Earth).

Holy Math Batman! The Great Circle Formula

Using the latitude and longitude of Canmore we calculate the great circle distance to the 21 supported cities and return the one that is closest and display that data in the widget. This calculation only happens once and the information is saved as a cookie in your browser.

Tip of the Iceberg
The weather widget is just a sample of some of the possibilities with the location based website customization. Some of the possibilities can include:

• Biasing your searches on the website to rank items that are physically closer to you more important than others. For example if you search for “Conrad Black” and you are located in Vancouver, show news stories of Conrad Black in Vancouver at the top of the results list.
• Customizing the news page. Similar to the above, display your local news before or above the national news.
• Display the proper radio and television schedules for your area automatically.
• Show you more “relevant” advertisements based on your location.

I’m sure you will see more of this type of customization as cbc.ca evolves over the next little while.

	15 Comments »	Email This Post
	CBC.ca web site, Under the Hood	Posted at 10:37 am (05 Jun 2007)

One of the great things about the Internet world (that our TV and Radio departments are envious of) is that we can get statistics about our visitors in real time. All of these statistics are anonymous, they don’t contain information such as your address or your name. We can, however, mine some pretty interesting information from the log files that our web servers generate.

It’s a Windows World
I’m sure its no surprise that the majority of our users are browsing the website using a windows machine and Internet explorer. In fact, 75% of hits to CBC.ca were done from users running Windows XP. Similarly, 72% of the hits to CBC.ca were done from Internet Explorer. Take a look at the breakdown for Jan 1st, 2007 to April 1st, 2007:

Visitor By Browser:

Visitors By Operating System:

How We Get This Information
Every time your browser fetches a page from CBC.ca the web server tracks that “hit” in a log file. Here is what an example log entry looks like:

69.17.178.81 - - [24/Aug/2005:21:03:17 -0400] "GET / HTTP/1.1"

200 4532 "http://www.google.com/"

"Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rev:1.7.0)

Gecko/20050716 Firefox/1.0.6"

Without going into too much detail, you can see the following information:

• The date and time the request for the page was made
• If the user came to this page from another one.
• The browser version, type, and operating system
• The size of the page the user downloaded

Using the information found in these log entries, we can come up with quite a few types of statistics, such as:

• What country the user is coming from
• What the busiest time of day for CBC.ca is
• What the most popular pages are
• What 3rd party websites are generating the most traffic to us
• If there are any broken links on our site
• What language the user ’s operating system and/or browser supports

How CBC.ca Uses This Information
Understanding who our users are and what type of browsers and operating systems they use is an important part in designing the services that are offered on CBC.ca.

We also use this information to do something we call “dayparting”. If we discover that the majority of the traffic to Business/Money section of the site during the lunch hour, then we may change the way items are displayed in the line-up on the front page. For example, we may promote more “business” related stories.

	11 Comments »	Email This Post
	CBC.ca web site, Under the Hood	Posted at 4:22 pm (25 Apr 2007)

[This post is courtesy of one of our developers, Keith Murray]

As you are probably aware of from the other articles on this site, the folks at Radio Two have rolled out their new lineup of shows. What you may not realize is that at the same time, a small group of folks at CBC.ca have been working overtime to set up a landing page to support the new look and feel of Radio Two.

Whats New?
Some of the cool things on the new landing page are detailed schedules, program descriptions and host bios, and playlists for the shows. The playlists come from an internal system called INews that I know nothing about. The show listings and bios come from another internalsystem called Program Guide, and that’s something I know about.

Before I get all technical and tell you about how the listings and bios get to the new Radio Two landing page, let me describe my relationship with Program Guide.

Some History.
Way back in the 1980s, I would ride my bike to Greenly’s Bookstore in downtown Belleville once a month and buy a copy of The National Radio Guide. It was a glossy magazine published by the CBC that had articles, detailed show descriptions, and bios of on-air personalities. The best thing is that it also had the complete Radio and Stereo (now known as Radio One and Radio Two) schedules. Never again would I miss an interesting episode of Ideas, with Lister Sinclair.

Check out these scans from the February 1989 edition. Ahhh, Morningside, Brave New Waves, Night Lines…

Click for larger…

The CBC stopped publishing the Radio Guide in the late 1990s, and they now publish listings on the Program Guide.

When I started working at CBC.ca, I found out that the Program Guide application was in need of an owner. Nobody wanted to maintain the code. But for me it was like meeting an old friend, so I gladly took it over. It didn’t take me long to find out why people avoided it.

Mouldy Code
Program Guide is suffering from bit rot. It’s old, and too many features have been hacked on. As well, every time somebody accesses a schedule, the application dynamically looks up the schedule in a database, and creates a nicely formatted page, just for you. Every time. It has to, because the local morning show in Tofino isn’t the same as in Montreal. But, that’s just not very efficient, and Program Guide crashes often.

Ok, I’m finally going to talk about how the schedules on the new landing page work.

When the Radio Two folks approached me with the idea of the new landing page pulling data from Program Guide dynamically, I originally said “No way, it will never handle the load!”. However, it was mentioned that the new Radio Two schedule had no local programming, so it only had to look up the schedule once per day, and it would be the same for all time zones (”half an hour later in
Newfoundland”). That I could live with.

So once per day, the Radio Two schedule, the show descriptions, and host bios are retrieved from Program Guide and stored in a file format called XML. XML files are an easy way to store lots of related information in a simple file that can be read with any text editor. It’s also easy to write other programs that read XML files and do something useful with them. For example, our podcasts are served up via RSS feeds, which are really just XML files.

Here is a snippet of the schedule for Saturday night in XML.


<schedule>
<timeslot>
<startdate>2007/03/25</startdate>
<starttime>20:00</starttime>
<endtime>22:00</endtime>
<title><![CDATA[Canada Live - With Patti Schmidt]]></title>
<episode></episode>
<shortdescription><![CDATA[Listeners will be transported to concert halls, music clubs and festival stages across the country for live performances]]></shortdescription>
</timeslot>
</schedule>


The XML files with the schedule information are transformed into HTML files on the landing page via XSLT. XSLT is a fairly efficient way to take XML files and transform them into something else, usually while adding formatting or other display components to
the data.

Here is a snipped of the XSLT file that correlates the programs with the host bios.


<xsl:stylesheet xmlns:xsl="http://www.w3.org/1999/XSL/Transform" version="1.0">
<xsl:output method="html" indent="yes" omit-xml-declaration="yes"/>
<xsl:param name="param1" />
<xsl:variable name="titleKey">
<xsl:value-of select="$param1" />
</xsl:variable>
<xsl:template match="/">
<xsl:for-each select="personalities/personality[./title=$titleKey]“>
name: <xsl:value-of select=”concat(firstName, ‘ ‘ , lastName)”/>
</xsl:for-each>
</xsl:template>
</xsl:stylesheet>


This is what it ends up looking like on the new landing page.

The core schedule data is still coming from Program Guide. However, with the new Radio Two landing page it’s much more efficient, as the information is only read once per day, for everybody, in any time zone.

Note from Blake: The program guide is slated to be revamped with performance increases and easier to use interface. For all your Radio 2 needs you can go straight to the Radio 2 website and not have to use Program Guide at all.

	6 Comments »	Email This Post
	CBC.ca web site, Under the Hood	Posted at 10:21 pm (26 Mar 2007)

[Note: I've updated some content since the broadcast]

Incase you don’t know, Test The Nation is a show on CBC Television that will allow you to take an IQ test and see if you’re smarter than the panel of surgeons, tattoo artists, or radio DJs.

There are two ways you can take the test. First, is online at cbc.ca/testthenation. The second is while you are watching the show by keeping track of your score on a scorecard.

Who Made It Happen at CBC.ca?
A quick glace at the credit page yields names and titles, but what exactly do these people do?

Project Management ensures that the project runs on time, meets all deadlines, co-ordinates the resources, schedules meetings, and is the link between the client and the project team members.

Platform Support is the team that I work on and provides support to the Technical Lead, Programmers, QA Team, and Design team. When things aren’t working as expected with the webservers, databases, ftp server, or mailing lists the Platform Support team is there to keep everything running smoothly.

Design Lead/Front-End Programming deals with anything that is the front-end of the site, that is, what you see and interact with. This role invovles integrating dynamic content, such as flash applications, or ajax widgets into the site. This also includes programming these applications.

The Research department takes all of the raw numbers from various sources and computes official and audited statistics for tv, radio, and cbc.ca. Their role in Test The Nation is to ensure that the data from the online IQ tests is accurate and in a format that can be digested by other departments for inclusion on TV or the website.

Copywriting is the term given to a person or department who’s role is to ensure that the text content is accurate and error-free. This includes checking spelling, grammar, and ensuring that the copy (their term for “text”) is correct.

Design, like the name says, involves designing the look and feel of the website. This includes colours, fonts, and layout. With projects that are integrated with TV there is usually a style guide to follow so that both TV and Online look unified.

Technical Lead is the person that all non-technical people go to in order to answer to their tech question. This person utilizes all of the resources at his disposal (Platform Team, IT, External Vendors) to get the answer he’s looking for. In the case for Test The Nation this person also did the back-end programming (written in Java of course).

Quality Assurance ensures that everything works properly. This includes testing the website on a multitude of platforms and browsers. As well as testing the back-end components of the site. With Test The Nation their role was to ensure that there was no bugs in both the front-end (Flash) and back-end (Java) code.

A view from the set

So How Does It Work?
The online test is a flash application that delivers the question and records your answer.

When you have completed the test data, such as: how you scored in each section, the time it took to complete the test, your demographics that you inputted at the start of the test, and your computed IQ are stored in a database.

All of the data is funneled through our Content Delivery System. Using their system, we are able to throttle the number of test responses that reach our database. This allows us to ensure that the database never goes down because it’s overloaded. If, for whatever reason, the site goes down, our CDN will queue up and test responses for later delivery when the site returns to normal.

At certain intervals during the show, the statistics department will deliver the latest results from the online IQ test.

Interesting Trends
Keeping an eye on the website traffic during events like this is always fun. Take a look at the following graph:

The red line indicates when the broadcast aired in the Atlantic region (7pm ET). As the show airs in each timezone you can see a corresponding spike in traffic. The next “blip” is eastern time. The following central, mountain, then pacific. If you were to drill down into each spike, you could see corresponding “mini spikes” during commercial breaks as people browse the site during TV commericals.

Keeping It Fun
During events like this, I like to spice things up a bit by having a pool of some sorts. A few of us in the office are making predictions as to the number of people who will do the online IQ test. Some people guess, others try to make an educated guess by using wacky formulas (like, taking the percentage of the estimated number of tv viewers) or “insider information”. Whatever the case, it will be interesting to see who won.

CBCers were able to take the IQ test before the show aired. I’m ashamed to admit that I scored an IQ of 96. I’m going to use the excuse of being distracted because I was doing the test while watching TV

	18 Comments »	Email This Post
	CBC.ca web site, Under the Hood	Posted at 6:33 pm (18 Mar 2007)

Daylight Saving What?
It’s amazing how something invented so long ago can cause grief today. The United States passed a law that would change the rules that we follow for daylight savings time.

As you know, computers keep track of time and can automatically adjust their clocks for daylight savings time. There is no need to “spring forward” or “fall back” your computers’ clock.

All of these DST rules (and all time zone rules) are stored on your computer as part of your operating system. If there are any changes with these rules your computer must be patched or updated in order for it to know about them.

How CBC.ca tracks time
Time is kept in sync with all of the servers using a system called “Network Time Protocol” (NTP).

NTP allows all of the servers to sync their time from really accurate time sources like atomic clocks. This is extremely important in the TV and Radio world and equally important at CBC.ca.

The majority of CBC.ca web servers keep track of time in Eastern Time. The eastern time zone observes DST. This means that if the DST rules change, all of the servers that run CBC.ca must be updated as well.

So what’s the big deal?

Newsletters
The news digests that are sent out are done so on a schedule. This would mean that the emails would arrive an hour earlier (or later).

In fact, twice a year, we need to update this schedule manually because Saskatchewan does not observe DST. In order to keep the Saskatchewan news digest going out on a regular schedule, every spring and fall we need to adjust the schedule on our server manually.

For you techies out there, that means adjusting the cron job each time we switch from daylight savings time to standard time:

###### Sask is in pre-cambrian time. No Daylight saving time:
###### One hour behind ET from Oct until April
#00 13 * * 1-5 perl /sites/cbc.ca/bin/digestmail.pl sask-am-headlines
#00 21 * * 1-5 perl /sites/cbc.ca/bin/digestmail.pl sask-headlines
###### Two hours behind ET from April until Oct
00 14 * * 1-5 perl /sites/cbc.ca/bin/digestmail.pl sask-am-headlines
00 22 * * 1-5 perl /sites/cbc.ca/bin/digestmail.pl sask-headlines

So yes, we will need to update those comments to refelct the new DST rules.

Log Files
Every time you visit the website, certain information is recorded in a log file. We use this data to see which pages people are visiting and how popular certain sections of the site is.

The time of the visit is also recorded and needs to be accurate in order for us to produce accurate statistics.

Program Guide
Accurate time is necessary in order for the Program Guide to be useful. You wouldn’t want to miss your favorite CBC program because the Program Guide is an hour off.

If Only…
If we were using UTC time instead of eastern time for timekeeping, this would be a non issue and we wouldn’t have to patch four different types of operating systems.

	12 Comments »	Email This Post
	CBC.ca web site, Under the Hood	Posted at 11:44 pm (19 Feb 2007)