EN
Back to the archive

The encyclopedia · Software & IT · Technical decision · 2014–2016

IBM predicted Wimbledon's web traffic and cut cloud-computing hours ~50%

Instead of manually scaling for tournament spikes, IBM's Predictive Cloud Computing forecasts demand and provisions automatically.

IBM · All England Lawn Tennis Club (Wimbledon)

The solution

Wimbledon's digital platforms see traffic that can surge by around 55,000 times during the championships. Reactive cloud platforms needed human intervention to scale computing resources, wasting both compute and labor whenever demand moved.

A small team in IBM Global Technology Services built PCC: advanced forecasting, discrete-event simulation and numerical methods model Internet traffic demand in near real time and allocate cloud resources automatically. In 2014 it was applied across tennis and golf tournaments.

PCC cut IBM's cloud-computing hours by about 50 percent at those events while automating labor, produced 16 patent filings, and strengthened IBM's analytics patent portfolio. IBM remains Wimbledon's official technology partner, running fan experiences on AI and hybrid cloud.

Why it worked

  • Forecasting turned a reactive scaling problem into a scheduled one
  • Simulation plus numerical analysis handled non-stationary spikes
  • Automation cut labor along with compute
What it achievedProvision cloud capacity by forecast, not reactionclever

What can be applied

For workloads with huge, predictable spikes, don't pay for peak capacity or wait for alerts; forecast the spike and let provisioning follow the forecast.

Aftermath

PCC expanded within IBM's portfolio and brand; IBM's Wimbledon partnership continues, with fan features such as Match Insights and Win Factors running on IBM Cloud and hybrid cloud technologies.

Sources

spotted an error? The archive wants to know.

Related cases