The KPN CDN
This summer we helped KPN to deploy their CDN. KPN has licensed our VideoExchange CDN software suite.
Like all telecom operators KPN has multiple strategical, technical and commercial reasons to deploy an on-net CDN.
Let’s start with some details about KPN for those who only know the US market players:
Europe
Tier-1 operator KPN is active in Western European countries including Germany, Spain, France, Belgium and the Netherlands. The company offers land-line telephone (+6 million subscriptions), mobile telephone (almost 25 million subscriptions), broadband access (over 2 million subscriptions), IPTV, DVB-T and DVB-H television services. The company also offers wholesale telecom and carrier services to many internet providers and service providers.
USA
In the USA, KPN owns the majority of wholesale VoiP service provider iBasis, handling roughly the same international volume as AT&T.
Pitch
KPN invited me in March to demonstrate our latest VideoExchange CDN solution. I pitched the solution to a small executive management audience of innovation, system engineering, video services and marketing divisions. A few weeks later I had another demonstration for technical and sales staff.
A nice quote was: “This is my dream. This is the technology I’ve always wanted to have’. KPN had considered developing their own CDN technology, but after having seen the demo, it was clear that KPN would benefit from licensing: a proven, feature rich and mature technology with a short time to market, compared to an expensive and uncertain internal path.
Q&A
KPN had many questions, about technologies, scalability, reliability, features, usability, API’s, deployment process, required skills, technical requirements, and so on. Many difficult and in-depth technical questions. Since we had already deployed a great number of CDN’s before, most of the questions were covered in our documentation. And VideoExchange is mature technology. Virtually all required features were built-in and there are so many additional features in there as well.
KPN also had a lot of marketing, sales and strategical questions. I had answered those in my pitches and used our experience with other customers and our own StreamZilla CDN to make sure that the CDN was not only a technical project, but could be sold to the market as well.
April: deal
In april KPN asked me to do a formal offer which was agreed to within a few weeks.
Project team
End april we setup a project team:
- First layer: me and a KPN executive for strategy and overview.
- Second layer: a Jet Stream project manager and a KPN project manager.
- Third layer: 2 Jet Stream software engineers and system engineers, plus four KPN system engineers and operational engineers.
Architecture
A great benefit of VideoExchange is that you can deploy the CDN independently from the network layer. There are no dependencies on specific DNS hacks. You don’t have to rip out switches and routers and replace them by even more expensive ones. There are no black boxes or appliances involved. Buy some off the shelve servers from HP, SUN, IBM, Dell that meet the CDN specs and you are ready to go. Storage can be anything: DAS, NAS, SAN.
Based upon our documentation, KPN had designed a first prototype architecture. Which was discussed in the formal kick-off meeting early May.
A few days later Jet Stream and KPN agreed to a final architecture design. KPN made a list of required hardware, which we double checked.
4 Weeks or 8 weeks?
We could have deployed the CDN within 4 weeks after the kick-off meeting if our own engineers would have deployed the CDN. Normally we ask temporarily remote access and can fully install the CDN remotely. Most of the work is making sure that all settings and config files are correct and tuned.
But KPN preferred that we trained their team so they would be able to deploy, operate and scale the CDN themselves. So, two engineers of KPN came over to get on-the-job training by our engineers: how to prepare the hardware, how to setup the software, how to integrate the logical parts together.
The longest delay in the project was to wait for the hardware vendor. Fortunately VDO-X runs on common hardware, so we only had to wait for a matter of days.
Testing
By the end of May the core of the CDN was up and running. All basic logical components were running and were integrated. The project was now split into parallel paths for deployment of all the extra servers (scaling), tuning and testing. KPN has a very intense testing procedure before taking any service into production. Every button, every feature was tested by specialized service testing staff. Testing is split into user interface and API testing, system integrity checks, performance tests and security tests. Fortunately all tests were passed, besides some minor issues.
New features
We had some spare time within the project timeframe to solve the minor issues and to make the cluster shared storage feature more robust. Since KPN runs on other OS trees than we do ourselves, we encountered some incompatibility issues. They were solved by adding code, allowing the VideoExchange software to run in multiple modes, depending on underlying software trees, versions and architecture design. Some feature requests were added to the road map and will be in next versions of the software.
Operational
By the end of June, the few critical and most small issues were solved. KPN and we mutually declared the project to be done. In july, the CDN went operational in internal mode, with a limited set of customers. This is to make sure that operations gets accustomed to daily routines before going live for the larger customers.
KPN did not have to hire additional staff or to further train their staff. They do not depend on us for operating their CDN. Of course we do offer 3rd line support. The VideoExchange software automates most routines. And the management interface is very intuitive. Adding servers or setting up new customers is done within a few clicks. Compare that to other CDN solutions which require heavy training, run in multiple complex applications, or are not even offered as a self-service system by the vendor.
KPN had deployed their CDN in under 2 months, from scratch. The process was a breeze for both us and KPN, thanks to the great staff on both sides. Kudos to the project managers and the engineers! And many thanks to the KPN staff who called the project a big success. They said that their experience was that most complex IT projects need much more time and custom work, and are tough. So thanks again!
Competitors?
Some people ask me whether we helped a competitor enter the market, since we also operate StreamZilla. No.
First of all, CDN technology licensing is our CORE activity. StreamZilla is extremely popular but it is a (great) second revenue model. And a great learning project. All that we learn is translated into better technology and better strategic consultancy. It is all about focus, remember! Focus, focus, focus.
Second, KPN has a different market approach. StreamZilla is a horizontal CDN, offered in Europe via e-commerce and European partners. KPN will offer their service to another market segment, vertically integrated with other KPN services like DRM, web hosting and connectivity.
Third, by licensing our technology to KPN, we automatically position ourselves as the experts while KPN can build further upon their brand name.
Fourth, we see KPN as both a customer and a partner. We can actually team up to block non-European competitors out of the market. The VideoExchange software raises the bar because of the features, innovation, ease of use and powerful API’s. When multiple service providers offer a similar, high end service in the same market, it becomes a de facto standard, making it harder for international competitors to enter the market.
Overflow, Overlay
Two of the possible cooperations between StreamZilla and KPN are overflow and overlay. VideoExchange allows operators to tie multiple CDN’s together. KPN and StreamZilla can temporarily overflow traffic towards each others CDN’s.
Our strategy is to license VideoExchange to as many operators as possible. Not just to make money (actually, did you know that we are the only CDN tech company in the market that is actually making money instead of burning money?) but also to help the operators to build a larger, virtual CDN. And that will become an interesting challenge for the non-telecom (‘pure play’) CDN’s who cannot delivery as ‘deep’ as the telecom operators. More about overflow and overlay in an upcoming article.
No comments yet.