Uninterrupted Service When Campaign Traffic Jumps Tenfold
The story of how an e-commerce infrastructure whose traffic jumps tenfold during campaigns handles that surge without buying capacity.
The story of how an e-commerce infrastructure whose traffic jumps tenfold during campaigns handles that surge without buying capacity.
In e-commerce, the most important day of the year is the day the infrastructure is pushed hardest. At campaign time traffic can reach ten times its normal level, and the surge happens within minutes, not hours.
On fixed-capacity infrastructure this equation has two solutions, and both are bad. Either you hold capacity for campaign day all year round — which means paying for hardware you don’t use for three hundred and sixty days — or you knowingly put that day at risk. In the second case you lose not just that day’s revenue, but the chance that the customer who abandoned their cart ever comes back.
What’s more, the decision can’t be undone. By the time you notice the system slowing down on campaign day, it is too late to intervene.
In the architecture on portvMind, application servers run behind a load balancer. During campaigns the server pool is grown; the load balancer runs health checks on the new nodes and adds them to traffic automatically. When the campaign ends, the pool is shrunk.
The second half of this model works on the cost side. Because metering is based on the time a resource is running, cost drops the moment capacity is reduced. Capacity opened for a campaign doesn’t keep showing up on the invoice for months.
Having the infrastructure located in Turkey brings a third gain. Access time to end users is noticeably shorter than with infrastructure serving from regions abroad. Considering that page load speed directly affects conversion rate in e-commerce, this creates extra value on campaign day.
When traffic jumps tenfold during campaigns, the infrastructure expands automatically and the service continues without interruption.
The more important change was in the nature of capacity planning. Capacity is no longer a purchasing decision made months ahead according to the campaign calendar, but a configuration adjusted within the day. This removes both the time the IT team spends preparing for campaigns and the risk of a wrong forecast.
Every e-commerce infrastructure that faces campaign traffic, launch peaks or seasonal swings meets the same question: when, and by how much, should capacity be increased?
On portvMind you don’t have to answer that question in advance. You increase capacity when you need it and reduce it when the peak passes. You can see the cost of the campaign load you expect in the calculator before making any commitment, and track your spending day by day by defining a separate budget for the campaign period.
We review your current environment together and design your campaign architecture. Migration consulting is free of charge and requires no purchase commitment.
Retail From stores to the e-commerce warehouse, many systems were moved; with GPU-backed Kubernetes clusters, product images are processed in real time.
Read the story
SaaS With multi-region architecture and self-service resource management, applications are deployed quickly and costs stay predictable.
Read the story