Does automated failover defeat the purpose of virtualization?

820 pts.
Failover Clustering
Server administration
My understanding of the purpose of virtualization is to reduce server sprawl. In theory if you get one physical server to do the work of two servers you increase the percent of utilization of the server while cutting down on the number of servers in the pool, thus reducing power consumption, cooling needs, data center space, hardware and software costs. But if you then utilize automated failover that requires that you have a powered-on stand by server that takes up space, needs to be cooled, does nothing until the primary server fails, doubles the hardware cost, and possible doubles the software costs, doesn’t this defeat the purpose of virtualization?

Answer Wiki

Thanks. We'll let you know when a new response is added.

Double-edge sword. If you have a small environment, you can host alll your virtual servers on a single physical server. However to provide high availability, you should add an additional host. With this in mind, you are provided all your virtual servers high availability function as compared to utilizing either Microsoft clustering per supported application. Not all your virtual servers would normally support a MIcrosoft or 3rd party high availability application.

In one environment, I was able to reduce 40 WWW physical servers to 2 physical servers hosting all 40 WWW web servers. Consider the reduction of the power, cooling, and of course space.

Discuss This Question: 2  Replies

There was an error processing your information. Please try again later.
Thanks. We'll let you know when a new response is added.
Send me notifications when members answer or reply to this question.
  • Cklunt
    Jim - the basic truth is that fail-over provisioning is never really is very cost effective until it's n+1. Even given the situation Aquacer0 gave above - I one of the *physical* servers failed - he would have been in deep as the other would be unlikely to be able to double its load ( all being treated as equal). A better choice might be to distribute the load across 3 active servers in such a way that any the 2 remaining servers might be able to pick up the guests from the failed server with out grossly significant impacting the other resident guests VM's. Granted this isn't as "cost-effective" but it's much better protection against outages if you consider possible lower performance for a short duration ( until the failed server is restored) better than loosing 1/2 your apps.
    45 pointsBadges:
  • Jim4522
    CKlunt, I had already reached the conclusion that failover to a standby backup system can not economically be justified, see the question I added this morning, and I agree that a more logical solution would be as you propose, to failover to other active servers in a cluster. But if one were to follow your advice what is the impact on those using the failing server out on the network. It is hard to believe that the failover you propose would be accomplished instantly and seamlessly. Is there any failover theory that utilizes independent active servers that will not be felt by users of the failing server? Jim4522
    820 pointsBadges:

Forgot Password

No problem! Submit your e-mail address below. We'll send you an e-mail containing your password.

Your password has been sent to:

To follow this tag...

There was an error processing your information. Please try again later.

Thanks! We'll email you when relevant content is added and updated.


Share this item with your network: