Written by Michael Sheehan on Aug 12th, 2008 | Filed under:
Cloud Computing,
General,
GoGrid,
Hosting
Kenneth Brill, executive director of the Uptime Institute, wrote an eye-opening commentary on Forbes.com titled: “Servers: Why Thrifty Isn’t Nifty” yesterday. What really grabbed me was his introductory sentence: “We are currently in the biggest data center construction boom in history.” He postulates that this is partially due to Moore’s Law which states that the number of transistors on a chip could double every 24 months which has translated into a boom in IT and consequently an increase in global productivity.
Strangely missing from his article, however, is the mention of Cloud Computing, but more on that later.
Granted, his commentary is truly targeted towards the larger corporation or enterprise that are looking to build or use large data centers and understanding the financial and environmental impact of such. He summarizes the dramatic growth in a paragraph in a manner that is almost scary to read:
“The number of servers in the U.S. has grown from 5 million in 2000, to 10 million in 2005, to a projected 15 million in 2010. More servers eat up more electricity and energy costs go up. To avoid future energy shortages caused by increasing IT demands, 10 more power plants need to be built to the tune of $2 billion to $6 billion each and their cost is ultimately going to get passed on to IT through increased utility bills.”
Power is a concern for everyone, especially those who run large data centers. ServePath, the parent company of GoGrid, operates a 20,000 square foot facility in San Francisco, where real estate alone is expensive. Many large corporations (such as Google) whose livelihood is server hosting are building tremendous data centers near rivers in order to capitalize on more environmentally-friendly hydro-electric power.
But not everyone has the luxury or financial wherewithal to be able to do this type of massive construction. So there must be another solution. Many have chosen traditional hosting to accomplish this, but, as Brill points out, the CapEx for simply hosting a dedicated server is large (and growing). Brill estimates that a $2500 Servers (what he calls a “low-end server”) hosted in an optimal cost location in the U.S. will actually cost between $8,300 and $15,400, depending on the Tier level of the hosting facility.
Source: Forbes.com - "Servers: Why Thrifty Isn't Nifty"
What is particularly interesting about Brill’s analysis are the points that are learned from this table. But this is where we need to start thinking outside the “physical” box and look to alternatives, namely Cloud Computing (through GoGrid, for example). So, I will step though his points (quoted from the article) and put things into perspective from a Cloud Computing standpoint.
- “Spending $2,500 on a server really means spending between $8,300 and $15,400 in facility capital to provide the necessary space for housing the server and powering it.” My Response: From a hosting provider’s viewpoint, cost and operation efficiencies can be gained through the use of virtualization/cloud computing. A handful of high-end servers, clustered in a Grid with management servers/software/devices, can be used to create hundreds of Cloud servers. This reduces the provider’s costs, assuming initial R&D and development are not factored in. Innovations such as pay-as-you-go (utility-like) billing deliver a much more cost effective solution to the end user (SMBs, Corporations, Enterprises, etc.) if hosting is outsourced to the Cloud.
- “Facility asset utilization is low at 55%. Because of the length of time required to justify, fund, plan and build a major data center (typically 24 months or longer) and the large incremental cost and risk of subsequently increasing its capacity, data centers are typically over-built at least in the early years. As a result, average facility utilization is low relative to capacity.” My Response: While Cloud Computing is still a new computing technology and service offering, it is the future that the Enterprise must start evaluating now. A 55% data center utilization is scary. Think if the used space alone were to be replaced with a Cloud infrastructure. A multi-fold space savings would be gained if traditional hosting space were converted to Cloud Computing infrastructure space. If one physical high-end Cloud Computing server that is the host for 8 Cloud Servers is used, data centers would suddenly have even more available space. Is that a good thing? Well it could be, assuming that the space could be sub-leased to other Cloud providers, for example. Since the physical infrastructure shell is available, build-out costs would be lower and the time to do build-outs would be significantly faster. It would almost be self-perpetuating.
- “Depending on tier level, depreciation ranges from 42% to 50% of total facility costs. Facility depreciation per server is not typically calculated because depreciation is a single lump into a big cost bucket and this doesn’t motivate companies to conserve.” My Response: Developing a Cloud Infrastructure will help with depreciation. From a hosting providers standpoint, depreciating several hundred high-end servers running a Cloud Infrastructure as compared to several thousand dedicated servers is much easier. The jury is probably still out in terms of CapEx and depreciation of the high-end Cloud environment, but from the standpoint of a corporation that outsources their dedicated servers to a hosting provider and then transitions their dedicated servers into a Cloud, the savings are immediate. But it does bring up a question: How do you depreciate a virtualized or Cloud server from a corporate standpoint? Do you look at the high-end boxes? If you have outsourced to the Cloud, you probably can’t. Since Cloud servers can be created and destroyed on the fly, it may be hard to amortize or depreciate over time. Accountants may start really scratching their heads with that one I believe.
- “Just the electricity required to provide power and cooling will exceed the cost of the server in six years.” My Response: This is an easy one. “Go Green” with virtualization or Cloud Computing. The ability to run 5-10 Cloud servers within a Cloud Infrastructure saves power and cooling. A single high-end server within a Cloud Computing node is hosting numerous cloud servers may consume a higher amount of energy versus a single dedicated, lower-end server, but if you stack it all up and do the analysis, the Cloud wins.
- “Facility operations include mechanical and electrical technicians, physical security staff, facility equipment maintenance and repairs, property taxes and other physical facility expenses. Within a wide range, these costs are largely fixed and on a unit basis would come down significantly if spread across a larger base that would result from increased facility utilization” My Response: To extend that thinking a bit further, if you can produce multiple virtual units from a larger single unit, economies of scale are multi-fold. You get “more out of less” physically and if you need to grow to gain increased facility utilization, your output increases in multiples as opposed just one by one. Furthermore, the human resource aspect is controlled due to the fact that there are fewer physical boxes to maintain. This may be partially offset by the need for a software team (from the cloud infrastructure provider’s perspective). However, for the Enterprise Clouds if outsourced, this would be a win.
- “Greenhouse gas emissions per server (based on a coal fired electricity source) are 4 tons per year. “ My Response: Basic math. 1 real server = 4 tons. 1 high end server running 8 Cloud servers = 8 tons (this is non-scientific…I doubled the number just to make a point). That would translate into 1/2 ton for the equivalent Cloud server. It would be interesting to get some actual numbers, perhaps later.
- “For an organization with 5,000 servers, the industry rule of thumb is that up to 30% are technologically obsolete. This means that up to 1,500 servers can just be unplugged with no negative impact on data-center production. The savings: $12 million to $23 million recovered in data-center facility capacity, $700,000 in annual electric savings and 6,000 annual tons of reduced greenhouse gas emissions. These savings result merely by telling the “kids” to turn off the “lights” when they leave the room. If we did this on a broad national scale, do we really need to be building all the new data centers, or could we defer a large portion of this investment into the future? Our companies and economy would be far better off if that money went into new application development instead of bricks and mortar!” My Response: EXACTLY! Move the Enterprise to the Cloud and you can’t lose. With Cloud Computing, under-utilized or dormant servers can be easily managed. Delete or destroy the instance of that server in a few clicks of the mouse (at least with GoGrid, it’s that easy). The instant horizontal scalability (both up and down) of Cloud servers makes the setting up of Cloud or Virtual Data Centers a much more financially viable solution. True, you can’t build a hugely complex data center in the clouds…yet. (Some would say that you can now though.) It will come and with it, incredible efficiencies.
In terms of servers being obsolete, this becomes less of an issue with hardware virtualization within Cloud Infrastructure providers. The ability to create “images” of newer servers, hardware or technology through virtualization allows for obsolescence becoming less of a mitigating factor.
Of course this transition from traditional data centers to data centers in the Cloud has to start on a smaller scale, beginning with bootstrapped startups, then Web 2.0-ers, then SMBs and finally the Enterprise. Savings on both the Hosting Providers side and the Corporation will allow for this new “application development” (e.g., reinvestment of money and resources by the Cloud Infrastructure providers into the self-same technology). And the end-user savings will enable innovations within the Corporation, eliminating the need of maintaining a bloated and archaic IT infrastructure.
I encourage you to read through Kenneth’s article very carefully and then read it again with Cloud Computing in mind. In many ways, it is a loud wake-up call to those companies with “a rack in their closet” or those thinking that Cloud Computing is just another trend. I say it is much more than that, an actual evolution of the data center. It is probably one of the only ways that Enterprise can scale and grow in an era of extreme cost and environmental consciousness. When you do move to the Cloud, GoGrid will get you there quickly, easily and efficiently.
