Ticketscript: Scalable and available
Ticketscript offers the organiser of events and concerts a full solution for the online sale of tickets. The proposition encompasses integration with Social Media and traditional marketing, behavioural analysis of visitors, profiling, and more. These activities are the direct cause of two very strict demands.
- Extreme scalability. At the start of sales of a popular event / concert, tens of thousands of clients want to buy their tickets simultaneously. People are literally waiting for the opening hour, thus generating up to twenty times the normal workload on the environment.
- Extreme availability. Especially during peak hours, a technical failure will spell disaster. Apart from the obvious loss of trust of the consumers, the organisation immediately loses revenue since the consumer will simply order the tickets elsewhere.
DOWNTIME ABSOLUTELY UNACCEPTABLE
Downtime equals massive losses and the consumer will quickly lose trust (and might then permanently turn to another sales channel). To help counter this threat, Ticketscript develops its applications using ‘Feature-based Development’, an amalgam of light-weight Agile methods. When combined with the CI/CD environment of Sentia, this makes for function-rich software that is quick to adjust to the end-users’ wishes, including the Infra- as-Code on which that software runs.
A change that could lead to a malfunction, is nearly always pre-emptively detected by the automated Test Environment. If an error however slips through the net, the Cloud reacts with automatic mechanisms to apply Workarounds, or even solve the issue altogether.
With our technology, our clients can organise their events even more efficient and therefore sell more tickets. Frans Jonker CEO Ticketscript
PLANNED SCALING AND AUTOMATED SCALING
The Public Cloud offers possibilities for Autoscaling (scaling up within minutes to withstand peak loads). To this end, Sentia measures many parameters including CPU load and messages in the Messaging Queue. But for Ticketscript, even that was not enough because of the massive peaks that appear in mere seconds.
Through Sentia’s Halloumi suite, the organisation can adjust a scaling parameter that will have the environment grow in anticipation of an event! On King’s Day 2017, our client used this mechanism to scale up the environment twentyfold in preparation of a truly huge event. Over the first minutes, the environment processed 35.000 tickets.
With a few simple lines of code, Ticketscript had the freedom to scale up the environment near-endlessly. That is the power of the public cloud and our management suite, Halloumi.
TIGHT RELATIONSHIP BETWEEN APPLICATION, MIDDLEWARE AND OS
Joining Cookbooks
The nature of the individual application sets high demands for middleware and OS. Containerising is very well suited to the purpose, yet Ticketscript chose not to use this yet (for practical, operational reasons).
Ticketscript has its own Chef Cookbooks that define, among others, the relation between EnginX and the application; Sentia in turn maintains the Infrastructure-as-Code with its own Cookbooks. We combine both sets of Cookbooks in the CI/CD street and test the outcomes with an automated Test Environment – including the integrity and functioning of our Ticketscrit’s portion!
Test, Staging and Production shared: optimal Quality Control
Development takes place at Ticketscript’s offices; Test / Staging / Production are a joint operation with Sentia on the Public Cloud. We purposefully keep the sub-environments functionality equal, and send all software and infrastructure releases through the Development Street without any exceptions. This ensures that the live applications continually comply with the two key requirements.
TECHNICAL DETAILS
- The virtual machines have no ‘direct lifeline’ with other elements; many virtuals may fail without severe impact on the applications (stateless configuration, communication through message bus).
- CI/CD environment in the Cloud, except for Development (on-site).
- Central logging with a tamper-free Event logger. The logs are sent with Serverless
- Functions to a managed, tailor-made ELK stack and there analysed.
- Complies with various Best Practices and Sentia’s own implementation standards: AutoHealing, AutoScaling, Encryption, Design for Failure.
