Foundation of Green IT: Consolidation, Virtualization, Efficiency, and ROI in the Data Center [Paperback]

The book describes many possible changes in a data centre, that addresses the increasing cost of these centres. Of these, the biggest advantage appears to be a shift to virtualisation. This springs from industry surveys that found typical server usages of 5-15%. Most machines in a data centre sit idle most of the time. But their power consumptions and hence the centre's cooling requirements remain the same. A tremendous waste that virtualisation directly tackles. Of all the issues mentioned by the book, going to virtualisation, non-trivial though it will be, is the largest saving.

In this sense, the book's title may be regarded as a pun. The green is not just an environmental saving, but a monetary one. Albeit, the pun reveals an American bias, since the US has most of its currency in one off-green colour, while most countries don't associate a single colour with money.

But I do have a disagreement with the text. Chapters 9 and 10 speak of a method of replacing offsite tape backup with a disk library that is also offsite. The savings claims look impressive. But the text makes a stunning statement that "Tapes are inherently less reliable than disk"?? Where the heck is the basis for that statement. Tapes have different failure modes from disks. This in and of itself does not mean less reliable. And one thing about a tape is that it has no moving parts, when it is not being used. A disk, if it is powered up, has a bunch of moving parts. The book acknowledges the latter in a different chapter, that talks about replacing disks with flash memory that emulates a disk, for storage that will be frequently accessed. One correct advantage of flash over disk is that flash has no moving parts. But by the same token, tapes are also generally acknowledged to have this advantage over disks.

Going further, it is unclear from the example in the text if the disk library that is offsite has most of its disks spun down or not. If the disks remain powered up, this gives rise to the chance of head crashes. Plus, the book makes much about efficiencies and energy savings [the latter mostly thru virtualisation]. Yet it doesn't seem to address this basic thing about a disk library versus tape. When tapes are archived, they are stored, and no power is used. When a disk is spinning, electricity is being used. More to the point, for archival data, what is the need to keep it constantly rapidly accessible by having it on a system of powered disks? This seems really wasteful of energy. Historically, one well regarded advantage of storing to tape instead of disk was this very feature of no energy consumption. Which also reduces the cooling requirements for a room full of tapes, as compared to powered disks.

Another aspect is the cost breakdown offered for the different savings. One is that for offsite tape storage, the cost of sending a courier 260 days per year with the tapes to the offsite is $19 500. But there are two objections, which the book doesn't seem to address. The first is that that cost can be halved simply by deciding that it is worth the extra risk to hold more tapes for a few days longer at the main data centre, by halving the frequency that the courier is used.

The second is that for disks, there might be no courier cost. But this is offset to some extent by the extra cost of a [presumably] high bandwidth connection between the main data centre and the secondary one. How else is the data going from the first to the second? Another way of looking at this is to say that yes, for disk libraries, there is a data connection between centres. Then imagine for tape backup, the main data centre has the same bandwidth connection to the secondary centre that holds the tapes. In this case, the tape drive that writes the tapes can be at the latter, and there is no need for a courier to carry tapes from the former to the latter.

The point I'm trying to make is that if you have any experience with data centres, you can analyse the book's cost comparisons and possibly find them wanting.

Granted, there are cost savings from going to disk libraries. Notably by reducing labour costs. To do retrieval, this can be largely automated if the needed disks can be put online by machine, instead of having to be manually found and turned on. So less need for a low level data clerk to be at the offsite facility. But this does not necessarily lead to a net cost saving.

It may well be that there is a place for offsite disk libraries in certain contexts. But perhaps the book should have argued its case by addressing the above points in more detail. The chapters do offer spurious accuracy in the cost estimates tables. What those omit are estimates of the errors in the provided numbers. The book does say that the numbers could vary, but nothing about by how much.

Mr. Poniatowski covers all the hot topics of "Green IT" which is relevant to the times we live in today - The book looks at what every IT professional is dealing with, and that's power or lack of it due to the ever increasing demand, e-waste and it's plague on our planet, going paperless, TCO (Total Cost of Ownership), ROI (Return on Investment) and other related topics, all centered around the "Green" movement. This book is well thought out and addresses the key issues in a manner which provides a clear understanding of various choices that can be made to help make the environment better. Information Technology (IT) is the lifeblood of every aspect of our civilization. By considering the importance of Green IT as one of the parameters to be addressed during their decision making and making some green IT choices, Executives can help minimize the impact of Information Technology installations on the environment. I found this book very informative.

Foundations of Green IT: Consolidation, Virtualization, Efficiency and ROI in the Data Center is a pick for any college-level computer collection strong in networking issues. It comes from a team of leading Green IT consultants and provides technical information and case studies for consolidation, better design, and reducing PC waste. Technical specifics make this a powerful recommendation.