According to a press release issued today, Hewlett-Packard, Intel and Yahoo have partnered with three research groups to create an open source test bed for research in cloud-computing, the concept of using globally distributed computers as a platform for delivering services.
As many of you know we've been working with Intel for the last 18 months on several "cloud" projects and our reoccurring problem has been how do you test a cloud platform designed to manage upwards of 100,000 servers without access to a real world test bed. This is a good first step, but with less then 1000 servers it still doesn't adequately address the problem.
I wonder how others developing large scale cloud platforms are handling scalability testing?