We are considering supplementing our engineering lab via the use of cloud services such Amazon’s EC2 or Rackspace’s Cloud Servers. Currently, we use a large set of VMWare servers to host our QA testing environment, where we perform installations of Zenoss servers as well as setup target systems (Windows, Linux, etc.) to monitor. In addition, these VMWare servers are connected to the same private network as various routers, storage devices, and other types of OS’s, such as HP-UX, Solaris, etc. All of this is done to simulate what Zenoss users have in their IT environments, and for the most part, the setup works quite well except that there’s never enough. The QA engineers are constantly finding new combinations of OS’s, applications, etc. that require more VM guests to be created. Here is a quick brain dump of the comparisons thus far.
|External cloud||On-premise virtual infrastructure|
|Guest platform flexibility|
|Access to monitoring targets|
|Access to install artifacts|
Obviously, VM sprawl is something that many IT organizations are dealing with, so our situation is far from unique. While some of these comparisons may be generic to lots of development organizations, the systems management aspect brings many additional requirements due to the variety of targets to be monitored. Even though this analysis is in its infancy, having the thought process published will generate ideas from others. Look for further posts as we get closer to making a decision.