Black Friday may have begun in the USA, but it’s become a massive phenomenon around the world. South Africa is no exception – for 2018, Black-Friday.Global tracked an increase of 1,952% in sales compared to an average day of shopping. Online sales in 2018 increased by 700% from the previous year, and there’s every reason to believe they’ll increase again this year.
Transaction volumes in November 2018. Courtesy of PayFast.
The same survey suggests that 66% of South Africans will shop for Black Friday deals online and offline in 2019. The stampedes and bottlenecks in physical stores have their equivalent in the online world – last year, Takealot, Makro and Superbalist all had site problems on Black Friday. Besides a loss in sales due to problems with access to or payments on these online stores, there’s reputational damage to contend with, which may have more long-term effects.
But there are a few strategies and lessons you can apply to minimise bottlenecks and maximise uptime:
1. Start with evaluating systems performance
Using historical data and projected growth, assess whether your hardware can shoulder the extra load – you may need to upscale hardware or storage capacity. If you’re not using cloud computing, consider the advantages – it eliminates a lot of the systems performance headaches, allowing rapid, automatic scaling.
2. Implement failover and disaster recovery for high availability
Consider all possible Black Friday scenarios, like serious hardware or software failures occurring in conjunction with peak demand. It’s essential to simulate failure and disaster recovery with dry runs – the ability to recover quickly from unexpected failures is critical to surviving Black Friday with minimal revenue and reputational damage.
3. Develop a scaling strategy
Vertical scaling (e.g. adding extra memory to a hardware) and horizontal scaling (e.g. adding additional servers to a cluster) are potential approaches, but there are multiple factors that contribute to successful scaling – service provider capacity, choice in hardware, and how applications are written are all potential points of failure. Microservices architecture can be implemented to improve scalability, but it won’t necessarily limit these failures. A scaling strategy is essential.
4. Develop a load balancing strategy
Reverse proxies help to improve performance on your web application, since internet traffic is offloaded from the application server and static caching can be used. You can only achieve a truly robust set-up by using multiple web and/or application servers and load balancing between them. A round-robin strategy is a good start but smart strategies allow you to direct traffic to under-utilised servers, keep specific clients on a particular server, and add resources during peak times. Cloud-based load balancers are a great way to optimise traffic.
5. Do some stress testing
On 27 September 2019, Pick n Pay ran a Black Friday trial. For customers, it was an early taste of the great deals that will be available on 29 November, but for Pick n Pay it was an invaluable opportunity to stress test their systems. It helped them identify weaknesses, which they can work towards minimising. An alternative is cloud-based testing – it’s easy to implement, and allows you to scale to a level that models real-world traffic patterns.
6. Use centralised transaction monitoring
Early detection of potential issues is key in managing the risks associated with Black Friday. Visualising key metrics in real-time, with a tool like Electrum’s Console, allows your team to be proactive in its responses.
7. Deal with timeouts correctly
Transactions typically pass through many external systems – processors, acquirers, associations, issuers, national payments systems and 3rd party providers. If any of them fail during peak traffic, it puts your business at risk. One way to mitigate against this is to have secondary relationships in place with other providers.
8. Use a trusted partner that can scale and keep up with volume
Cloud-based services like Electrum’s come with considerable advantages for rapid, on-demand scaling. By leveraging the latest high-availability, failover and disaster recovery technologies, our service can ensure your business is up 99.999% of the time.
As the annual Black Friday shopping activity grows, it’s inevitable the deluge of transactions will cause problems for some South African stores. The complexity of payments systems, and the immense strain they’ll be under with the traffic involved, will kick up some unexpected problems. But with the right strategy – and partners – you’ll be doing all you can to minimise the issues, and make this Black Friday your best yet.
If you’d like to talk to Electrum about preparing your business for Black Friday, get in touch.
.png)
.png)
