Big Compute and high performance computing are now international fields, as anyone who has been to conferences likes Supercomputing and ISC can tell you. Six countries have 10 or more clusters on the latest Top 500 list and 29 total countries are represented. It’s clear that the need for large-scale computing is growing globally with the news that China now is the country with the most clusters on the Top 500 list. In support of this growing globalization of high performance computing and big compute, the major cloud service providers are continuing to spread spread their datacenters around the world.
At Cycle Computing, we understand this, too. Our customers have many reasons for selecting which region to use for their workloads. Some customers prefer to use the closest datacenter to minimize the network latency. Other customers choose to go further away in order to mitigate the risk of natural disasters. Some have teams spread around the world. Whatever the reason, we keep up with the new regions from CSPs because we know our customers may need them.
Sometimes adding a new region is simply a matter of copying our images and software. Other times, a new region means supporting new APIs or other provider interactions. As an example, we have a customer that uses CycleCloud to power their software-as-a-service offering. They recently had requests from their China-based customers to be able to run within the China region.
Adding support for this region involved some development work on our part, including patching upstream libraries and finding reliable CentOS mirrors in China. But once completed, that work enabled our customer to deploy their SaaS product into the China region with very little effort. And more importantly, the transition was almost seamless for their end-customer.