Personally I would just go for Virtualenv, but you'd be surprised if you know people do with virtualization. I've mostly seen it in Windows shops, but it's by no means limited to that. A common "anti-pattern" I've seen is to build a vmware image, or set of images, that run your "environment". Developers then spin up copies and request changes that they need be made to the "master copy".
When it's time to test and deploy, the code is deployed to a vmware image, which is then sent to the testers. When everything checks out, the code is once again deploy to a new copy of the image, which is then promoted to production.
On argument we've seen is that it makes it easy to do a rollback of the system, using vmware snapshots. It might be a sensible idea in some cases, I just thing it's a bit weird and some overhead.
The worst use case of this I've seen is a Telco, where you needed to spin up as much 12 vms, depending one what you and your teams was working on. But then again that's the same company that read the Perforce guidelines on recommended setup and still decided to just pill all the projects in to one repository.
When it's time to test and deploy, the code is deployed to a vmware image, which is then sent to the testers. When everything checks out, the code is once again deploy to a new copy of the image, which is then promoted to production.
On argument we've seen is that it makes it easy to do a rollback of the system, using vmware snapshots. It might be a sensible idea in some cases, I just thing it's a bit weird and some overhead.
The worst use case of this I've seen is a Telco, where you needed to spin up as much 12 vms, depending one what you and your teams was working on. But then again that's the same company that read the Perforce guidelines on recommended setup and still decided to just pill all the projects in to one repository.