They go through separate endpoints. The endpoint for compute (EC2 api) has a form: http://192.168.23.71:8773/services/Eucalyptus
And the endpoint for storage(Walrus, which exposes S3 API) has a form: http://192.168.23.71:8773/services/Walrus We use the same environment variable as EC2 to expose the URL (e.g., EC2_URL, S3_URL, EC2_ACCESS_KEY, etc). The rightscale's aws gem was able to speak to Eucalyptus and I recently tested 'aws' gem and confirmed it speak to Eucalyptus without changes. -----Original Message----- From: David Lutterkort [mailto:[email protected]] Sent: Thursday, December 16, 2010 2:20 PM To: [email protected] Cc: Sang-Min Park Subject: Eucalyptus and dynamic switching of driver/provider Hi Sang-Min, Since we are going through some pain with (a) rewriting the EC2 driver and (b) making it possible to select the driver and endpoint dynamically, I was wondering if you could explain what the best approach for Eucalyptus is. IOW, with Eucalyptus, do storage and compute requests go through the same endpoint, so that users could set API_PROVIDER=http://euca.example.com/ and be done or are there separate endpoints for compute and storage ? David
