Blog Post

RightScale sizes up Google Compute Engine

The availability of Google Compute Engine has certainly shaken up the cloud infrastructure world.  GCE (s goog)pretty much zoomed to number 2 on most people’s lists of most scalable public clouds. But there’s been precious little third-party data about it.

RightScale hopes to change that. In a blog post Wednesday, Brian Adler, RightScale’s senior cloud architect outlines tests RightScale performed with Apica,a website testing and optimization company, that give some indication of the scope and performance to be expected from GCE.

In the testing, Apica drove traffic to a three-tier web application running on GCE with RightScale Cloud Management Platform configuring , monitoring and scaling the deployment.

From Adler’s blog post:

During the test, we scaled up to 330,000 page views per minute from 200,000 concurrent users, maxing out at 42 servers on GCE during the peak load. To put these numbers in perspective, Evernote states that its application on average receives 150M requests per day. Our testing on the GCE platform nearly doubles the load that Evernote typically experiences.

Now keep in mind that RightScale is supporting GCE — indeed reselling it. I would recommend reading the blog for the full result summary, but net, net, net, Adler said GCE exhibited “extremely high performance, low complexity, and great flexibility.”

Now here’s hoping that RightScale will run some comparative tests of GCE and Amazon(s amzn) Web Services to see what sort of race we’re looking at here.

rightscale gce

2 Responses to “RightScale sizes up Google Compute Engine”

  1. Certainly an interesting baseline but I’m not entirely sure what they’re trying to prove, because they’re not comparing it to anything. Agreed with Barb it would be much more interesting if they could run the same simulation across all the clouds they support. But that probably wouldn’t go down well with the losing clouds after all, Rightscale are resellers of them!

    Benchmarks can be dangerous. They’re useful for baselines or to get a relative idea of what performance to expect but as for providing actionable data, they’re more just a show off.

    • sebastianstadil

      You are right about the politics of this.

      As a follow up to a previous post (, we created a community project on github ( to benchmark cloud performance for some common use cases encountered (mysql, cassandra, or analytics workloads, for instance), for which we’ll publish results once we’ve gotten community validation.

      Benchmarks are hard to get accurate though, so we’d love to get your input to make sure we cover a diverse set of use cases.