The attention and need for green and sustainable organizations has been happening for quite some time – both because it’s the right thing to do and because engineers really know how to approach it. There is an extreme and very dedicated focus at SAP to help companies in the area of Green IT and Sustainability. I am delighted to support the efforts and help educate as much as possible how customers can take advantage of the work coming from the Green IT Community. What I want to do is go through what a fully engineered approach to green IT looks like. The reason I want to do this is to show you how many moving parts there are, and how, if you have them all synchronized and moving together in service of green IT, you can achieve tremendous savings. But first, let me set the stage a little bit …
It seems like only yesterday that the dot-com boom was in full swing, and companies from stock exchanges to overstock outlets were buying servers left and right to keep up with the demand for compute capacity. Hardware manufacturers were only too happy to oblige. Cool rooms were built, and servers were configured to vertically support one application on each server.
When the bottom fell out of the dot-com bubble, it was generally assumed that demand would slacken and there would be a great deal of excess capacity on the market. Well, demand for certain things, like competing online grocers, did kind of fade away. But by and large, companies did not stop innovating, and they continued to look for new revenue opportunities, and they were using more compute resources to do it. With the wave of corporate acquisitions and right-sizing that happened throughout the decade, even as the human chatter began to get quiet in some office suites, the machines were still being put to ever-more taxing applications and calculations down in the server room.
Now, organizations are reaching a kind of tipping point. The proliferation of applications spread across multiple servers, floors, and even cities and nations, has made for an inefficient crazy quilt of computing power. Some of the powerful hardware acquired in the early part of this decade throws off a great deal of heat and consumes a lot of energy. A word that was on the periphery ten years ago – “sustainability” – is now at the forefront of every corporate board member’s agenda. How can companies get greener? Sure, recycling all that waste paper certainly helped. But the energy bill is where the real impact is still to be made.
Enter “Green IT.”
Most server arrays are severely underutilized – they typically only use about 20% of their capacity. Even though we are increasingly turning to virtualization and cloud computing to handle the heavy processing around and as those machines announce their capacity has become available – decade-old machines are still overstraining the cooling systems of existing data centers and running up enormous energy bills.
Virtualization works to an extent – but it also increases the complexity of the systems you manage (probably on a reduced budget with reduced staff – have I guessed right?) – and it ultimately can’t change the fact that running some of this dated hardware is like driving the equivalent of a ’79 Buick when you could be driving a Prius.
So that’s why the experts are increasingly concluding Green IT really starts to afford benefits when you deploy an operating system that supports virtualization, ALONG with hardware that is configured to support virtualization from the start. Virtualization, in my opinion, is after all about making the most efficient use of hardware resources, and therefore the most efficient use of energy.
But going “green the hard way” also refers to the fact that some hardware is designed explicitly to consume less power and run cooler.
In support of SAP and it’s focus for Green IT, we are pushing optimization for hardware platforms. We have started to see some very impressive SD benchmarks for our Chip MultiThreading line but want to really push the limit with optimizing for this next generation platform. The processing power alone delivers 4x higher performance with half the space.
So that’s just the processor. When you start looking additional technologies built in at the hardware level such, as logical domains, then you start looking at better consolidation and scalability.
A complete engineered approach is to also optimize the Operating System to take advantage of the hardware capabilities. For example, if you are a Solaris user today, simply upgrading to Solaris 10 can help you shed some dollars from your energy bill. A study by Crimson Consulting Group determined that running Solaris 10 on Sun hardware represented an average power/cooling savings of approximately 60%, with the average space savings clocking in at 57% over Solaris 9. Another example is the engineering efforts with Intel. Optimization for Solaris to take advantage of the intelligent performance and automated energy efficiency built into the Intel Xeon processor 5500 series also delivers energy savings. I haven’t even touched on Network Virtualization but I encourage you to look at technologies such as Cross Bow in the latest release of Open Solaris. This new feature will be in the next release of Solaris.
What I am simply stating is that one needs to look at how all these pieces of technology work together to help you save on power, space, cooling, administrative and support costs. This is not an easy task but certainly one that SAP is working on to simplify. The future is Green. We applaud SAP in this area and will continue to support their efforts through continued collaboration.