Date: 14th August 2014
Book makers know about betting, but they need not gamble on their content delivery.
It doesn’t seem long ago I was watching the World Cup finals and here we at the start of another domestic season. When I say I was “watching the World Cup finals”, I was really looking at how 7 UK-based online football gambling websites were performing!
The results are in and I have to say, they all played well… with an average availability of 99.59%. There were peaks and troughs though. Interestingly the day before the first semi-final, the Brazil v Germany game, we saw availability drop to just over 97% and on the day of the game the delivery time was 23% slower than the fastest recorded day. It would seem betting on the top two ranked teams playing each other was more interesting than any other game.
Overall quite an impressive World Cup from the betting boys, but the slowing delivery time got me thinking about betting online and in particular, in-play betting.
I was amazed at the amount of content on in-play websites: Some offered text commentary, graphic representation of the games, live audio and some showed the game itself! Most, presumably sourced from third parties. This richness of information and content may all have an impact on the user experience. Throw in a few thousand punters betting in real-time and the bookmaker’s may have a problem.
So betting sites depend on many third-party suppliers, cloud services or content delivery networks (CDN), in fact a myriad of platforms and other service providers. Cloud and CDN performance has a huge impact on the online betting experience, as the very content being delivered drives the highest interaction from their customers. The combination of third-party content and time-sensitive data might mean critical delivery would seem to be out of the bookmaker’s control.
It’s very difficult to test & measure cloud and CDN suppliers, because by their very nature it is dynamic: The servers and the IP addresses that are serving the in-play website’s content change and they can change every few minutes. Further complications are that CDNs are, by definition, designed to serve content to end-users via local servers to ensure high availability and high performance – so different users may have different experience via different servers.
All is not lost though, as there is a technical way to keep track of one the most important CDN factors and that way is to capture the ‘cache headers’ from the content served. Basically a “Cache=Hit” is good, whereas a “Cache=Miss” is bad. In essence, a “cache miss” means the CDN serving your content was forced to make a return trip back to your central, origin servers. Uploading content to/from the origin is not ideal as the gambling house would want its own servers to be handling bet transactions rather than CDN content.
With gambling websites so dependent on third-parties, they need to ensure the relevant parties are alerted in real-time for issues that they are responsible – “FineGrained” alerting.
“FineGrained” alerting only sends warnings to the relevant third-party or team member of an issue arising that they are responsible for. These alerts are achieved by filtering problems detected via their host domain or detecting slow delivery of specific components by their full URL. By logging their hosts “Time-to-First-Byte’” (TTFB) will also provide an indication to any possible slowdown.
Combining TTFB in the monitoring system can then trigger additional network MTR (trace route) tests in real-time. These smart triggers can then identify if the poor TTFB is due to a fault of third party servers or a network issue nearby.
The complexity of a gambling site can now be matched by an array of clever alerts, narrowing down the problem. Agents running in the background can be set to alert when certain issues are detected and the relevant parties alerted to fix the fault. With the new football season upon us, surely it is better to play safe by keep the punters happy and the content flowing.