Scaling With Demand: How to get your University ready for clearing
Clearing is a high-pressure time for both students and Universities. Clearing can quickly shape the next 3 to 5 years of a student's life. And for universities, the financial benefits run into millions. It’s the single biggest university enrollment event.
We take clearing very seriously at Codiance Higher. Ensuring universities and students have the confidence they can quickly access the most up to date information is critical. You need to ensure that:
- Pages load quickly, without errors, even under heavy load.
- A backup is almost-immediately available in the instance a component fails
- If an underlying platform fails (such as Azure), a replacement is available.
There are three key areas that we believe you should think about when preparing your university for clearing.
Response Times
Under such heavy load, response times are critical. We find that users will not wait for pages to load for more than ~0.5s! As a first step we would double the power of your servers specifically for clearing. In practice, this will provide x8 the power of your day to day setup. For Azure, we usually upgrade UK servers to P3v2.
We’d also recommend the site running across multiple app services in multiple data centers. As well as UK-based data centers, we’d recommend having servers in key markets for your university, such as APAC.
Resilience
Resilience is the ability to ensure the site can meet unexpected demand and recover from failures. As mentioned, ensuring you use at least 4 data centers is a key component to this.
Load balancing the data centers is key. If you use Microsoft Azure, you can place an Application Gateway (their load balancing technology) in front of the app services, ensuring requests are distributed evenly.
Application Gateway has its own resilience and redundancy solution built-in - a single Application Gateway is composed of 2 separate instances.
Redundancy
With the elements above in place, a failure in a single region would not be catastrophic. If any given region fails, there are 3 other servers in this design.
Finally - and although extremely unlikely - there is still the possibility that services such as Azure could fail. For those that use Azure, this would affect every region.
In this very unlikely event, we’d suggest setting up a static page on an unrelated server that simply displays your telephone number and a link to applications. So even if everything fails, there’s still a clear path for those applying.
Load Testing
We of course recommend load testing this configuration. We would use 400 concurrent virtual users loading the home page and clearing page constantly. It shouldn’t flinch at even this high load!
We hope this has been useful. We know how stressful clearing can be for anyone responsible for the smooth running of clearing.
If you want to find out more, or understand how we can help you manage clearing even better, please get in touch!
We’re always keen to speak to people about their exciting ideas.
Please get in touch for a friendly, no obligation conversation with one of our engineers.