Is Cloud Computing More Energy Efficient than On-Premise Data Centers?

This blog is based on the cloud computing panel session at the third annual SVLG Data Center Energy Summit.

The panel was moderated by Teresa Tung of Accenture, with Jon Koomey of Stanford and LBNL and James Staten of Forrester.

From left: James Staten, Jon Koomey, and Teresa Tung

First, let’s define what they were talking about. There are public and private clouds, and they can be placed in-house or outside at your favorite provider’s. So there are three possibilities: (1) public and outside, (2) on-premise and private, and (3) hosted private. Public and internal clouds do not make sense and do not exist. This discussion compared public external with private on-premise (the panel did not categorize an enterprise’s own infrastructure as private cloud).

  – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – –  
White Paper: Effective Backup and Restore
– – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – –

So, under this assumption, is it more energy efficient to execute a task at your internal data center or to use a public cloud? That was the theme of the conversation. Without defining the basis of the conversation, it is hard to argue the conclusion either way.

Before the summary of the panel discussion, here’s the conclusion:

  • It is relatively easy to compare the cost of the two scenarios, and there are many examples of that. But the energy-related data is scarce, mainly because it is harder to measure and cloud providers do not want to make it public yet.

The summary of the conversation follows.

  • Some fundamentals about cloud computing:
  • There are a huge number of instances of tasks of a homogeneous nature.
  • Economies of scale spread cost (thus, power use).
  • Going from physical servers to virtual servers reduces cost and energy (strict redundancy no long necessary).
  • Using cloud bypasses internal complex work processes for implementing a system (use credit card vs. lengthy work order approvals).

SaaS and IaaS both exploit economies of scale:

  • For SaaS, homogeneity is implemented with multitenancy, processing a huge number of customers with one instance rather than using a separate process/thread for each.
  • For IaaS, homogeneity comes at the level of the same hypervisor.

Cost of transactions need to be studied:

  • If you use cloud computing for a given task, the task is initiated and terminated by sending and receiving signals or transactions between your premises and the cloud.
  • The cost of transaction needs to be studied to compare the two environments.

Concrete examples of cost reduction or energy reduction based on cloud computing:

  • Cost reduction examples are abundant, but energy reduction examples are not well published yet. Accenture will publish a report with some details of the data on cloud computing providers (CCPs), without naming them, in November.
  • University of Massachusetts’ DNA analysis cost them around $200 for four hours using Amazon’s services, but if they had had their own, it would have cost them more than $10M over a few years.

Overprovisioning trap for CCPs:

  • CCPs are better at provisioning their services and infrastructures
  • The organization is flat and does not need to interact with many departments for forecasting.
  • CCPs are profitable if they can achieve 60% utilization. They play Tetris to fit as many VMs as possible on a physical server. Usually, VMs do not stay very long, and economies of scale work very well.

Will we have our own data centers in the future, maybe 15 years from now?

  • Most commodity-type applications will move to clouds. But some using a cloud may leave the cloud to provide more specific features, like high performance, and enterprises will still run data centers.
Zen Kishimoto

About Zen Kishimoto

Seasoned research and technology executive with various functional expertise, including roles in analyst, writer, CTO, VP Engineering, general management, sales, and marketing in diverse high-tech and cleantech industry segments, including software, mobile embedded systems, Web technologies, and networking. Current focus and expertise are in the area of the IT application to energy, such as smart grid, green IT, building/data center energy efficiency, and cloud computing.


No comments yet.

Leave a Reply