Welcome, and first, I would like to thank all the wonderful people I have engaged about the Billion Node Cloud concept. With very few exceptions, they share the enthusiasm of this seriously ambitious endeavor and think it is kinda cool. You know who you are.
BILLION NODE CLOUD
The Billion Node Cloud is a concept and vision that incorporates existing technologies and dramatically expands their application to achieve a new type of cloud computing ecosystem.
A Billion Node Cloud does not refer to Smart Phones or hand held devices like tablet computers, but someday it might. The computing hardware platform it employs are servers. All sizes of servers and in particular low-wattage, compact MicroServers to achieve the massive volume needed to reach a Billion Nodes.
BNC blends three existing technologies to create a computing environment with global financial and productivity implications. The three technology areas are –
- High-performance, low-wattage and compact MicroServers.
In the last year, the industry focus on low-wattage, server-based computing is finally yielding significant results. An example of this is the US Micro PC, CriKit MicroServer and the CriKit Desktop Private Cloud – http://www.crikit.info . For the first time, end users can have significant compute power in a small form factor and combine those compute engines to create low-wattage cloud and cluster solutions on a desk. Today, one CriKit MicroServer is the compute equivalent of an Extra-large instance in Windows Azure or Amazon Web Services. This is significant because it also means that these MicroServers can run many small or extra-small virtual machine instances very adequately. This compute capacity in such a small form factor with low energy consumption dramatically changes the dynamics of computing.
- The maturity of Cloud Computing software solutions.
Cloud computing software has become sophisticated enough to be able to manage a large number of nodes regardless of where those nodes reside. A great example today is the Enomaly offering from VirtuStream. http://www.enomaly.com With the SpotCloud service, organizations and even individuals can “rent out” their hardware resources and hypothetically generate revenue from their participation in the SpotCloud environment. This capability will be adopted by other cloud providers over time to create extremely large cloud computing environments. Other cloud software vendors have been talking about the capability to have one large integrated cloud and are in varying stages of providing something that could accomplish the concept of a Billion Node Cloud.
- High Speed wireless bandwidth.
There are many new technologies and solutions being rolled out that provide adequately high speed data transfer through cellular networks and also satellite communication environments. An example of upcoming satellite broadband technology that makes high speed networking ubiquitous is Kymeta Corporation – http://www.kymetacorp.com . This connectivity speed is crucial to access cloud system compute resources that could very well be mobile or “remote” locations, where land lines don’t exist, or are impractical. Current 4G capabilities are becoming widespread and have the speed necessary to allow large numbers of nodes to participate in a cloud environment for a limited type of cloud processing. Namely, those applications that don’t require huge data sets to be transferred to cloud nodes. Fortunately, there are many cloud application types where non-ethernet speeds are adequate for data transfer of light applications and small data sets. Current, high speed land line connectivity will certainly play a major role to achieve a Billion Node Cloud, but it is the mobile aspects which give the concept highly disruptive and innovative qualities.
Bringing it together
OK, so let’s combine these resources into a cloud solution. The CriKit MicroServer draws around 50 Watts and .8 amps when fully exercised by applications. There are many environments that can provide 50 Watts of electricity as a matter of operation. Autos and trucks are two examples. Imagine if a long haul truck was configured with a small datacenter of, say, 6 nodes and it was connected to the “cloud network” via wireless. The “truck datacenter” would draw 300 watts and provide 6 extra- large compute instances as part of the overall Spot Market cloud. An extra-large instance costs approximately $20 per day with additional costs for bandwidth and transactions. If utilized, a trucker could make $120 a day for simply having the data center in his truck. Multiply that by 10,000 and a large trucking firm would have the potential to make $1.2 Million A DAY while their trucks are on the road being used as “cloud nodes”. They have the space and the electricity, so why not make some money by implementing a few MicroServers and wireless in the truck? Think for a moment of all the cars and trucks on the roads of this planet. Now imagine they contain compute power and are integrated into a global cloud infrastructure where anyone can buy and sell compute resources. We are much closer to that reality than you may think.
Now, let’s take it down to Taxis and regular commuters. One or more MicroServers in the vehicle could generate revenue while moving or sitting in traffic. Further, the concept is certainly not limited to mobile vehicles. Homes, educational institutions, cash-strapped government entities and businesses of all sizes could take advantage of this revenue generating solution without significant electrical or space investments. In addition, because MicroServers draw such a small amount of electricity relative to their compute power, they could be powered by a few solar cells. The future could hold small, secure cinder-block huts that contain solar-powered mini-data centers generating cloud revenue from a home owner’s back yard. There is nothing “far-fetched” and futuristic about that scenario – everything exists to do it today.
If all the components exist to actually begin this new era of cloud computing, the question really becomes, “Why not?” Why not create a computing environment that allows anyone with the required minimal financial resources and expertise to share in the “gold rush” of revenue that cloud computing is ushering in? Why not distribute some of the massive cloud revenue to “the people” instead of just large institutions? To simply entertain and investigate those questions and consider the massive benefit to society through thought experiments starts us down the road to implementation because the answers are overwhelmingly positive and the effort is so minimal.
Super Green Computing?
With the vision of using Solar power and existing electricity generating entities – like cars and trucks – to provide the electricity for the compact compute engines, it stands to reason that the additional investment in electricity generation would be minimal. Compare that method of powering compute engines with creating massive data centers with huge backup generators and incoming grid power and it is easy to imagine the electrical savings which translates into huge reductions in the consumption of any resource that is currently used to generate electricity, like goal and gas.
For a quick overview of how much coal could be saved by using low-wattage MicroServers, please see the video at
Financial Overview of this preposterous proposition of a Billion Node Cloud.
First, the MicroServers. For a general example, at $1,500 per MicroServer, the total cost for just the hardware of a Billion Node Cloud is $1.5 Trillion Dollars. If there are services components for installation, maintenance, training, etc., the number is significantly higher. That looks like a lot of cost, but savvy business people know it means massive spending and that means JOBS – lots of them. People will be needed for all aspects of this huge endeavor and the ecosystem it creates will be massive across the board. Further, this money is not a “spending sink”. It is a direct investment to generate revenue with a minimal payback period which we shall see next.
That was the cost, now for the revenue generation. This calculation depends on utilization of the installed compute resources in a SpotCloud market scenario , so, let’s say the compute resources are 50% utilized as Extra Large instances and one Billion of the compute nodes exist. This would be roughly 500,000,000 times the cost of an Extra-Large instance today. So, $20 times 500,000,000 equals $10 Billion PER DAY in theoretical revenue generation from One Billion Nodes– again, at 50% utilization of a Billion compute nodes. At this rate of utilization, there is a theoretical 150 day ROI on the investment of $1.5 Trillion.
As a thought experiment, imagine One Billion Nodes of cloud computing happening all around you – cars, trucks, homes, neighborhood condo associations ( “Hey, lets turn that unused closet in the exercise room that nobody uses into a cloud mini-datacenter to pay for landscaping! ” Why not?), etc. Then expand the scope of thought to imagine the ways people make money on the periphery of the concept. Networks, Software, Installation, Repair, Solar Panels, Alternators, Batteries, Mounting Hardware and maybe even gizmos and dangly things on rear view mirrors that say something like ” My Car Computes” . The point being that massive revenue generation also exists around , and because of, a Billion Node Cloud for the every-person to participate in directly.
The road to a Billion Node Cloud, and all the revenue it will generate, will take many years. On this journey, there will be substantial new developments and at the current trend of chip development where many, many compute cores are being placed on a single piece of silicon, there is the real possibility of an acceleration of the installation of compute nodes. Today, there is some vehicle modification required, where in the future, it may simply be an option from the vehicle dealer, or a server compute unit may simply be plugged into a standard connector on the vehicle. Once general society and not only technologists are engaged with this ambitious endeavor, the engineering around the solution-set will change to make it easy for the average citizen to provide server compute resources from anywhere they are – car, home, boat, business, wherever.
Jobs, jobs, jobs
The amount of employment around an ambitious concept like the Billion Node Cloud puts other job creation strategies to shame. From a global perspective, the number of people to implement this vision is staggering. We must look at this from an entire supply chain perspective, from planning, design, manufacturing, implementation, maintenance and additional services. Using that broad spectrum of revenue generation and job creation as a foundation, if there is one job created for every 100 nodes in this Billion Node Cloud, that would mean approximately 10,000,000 jobs would be created worldwide from this single initiative. And many of them would be net new jobs because it is a completely new category of applied computing.
There actually is a national security play here. When DARPA first funded the research that became TCP/IP networks and eventually the internet, one of the design goals was resiliency in case of attack. A Billion Node Cloud made up of discreet compute nodes spread all over the country and possibly the world would ensure that compute capability is available in case the major computing centers were taken down. In many ways it achieves a similar goal to what the original DARPA design was for the internet. Ubiquitous, resilient connectivity and ubiquitous compute resources in one cloud.
The Billion Node Cloud initiative does not replace existing and planned cloud computing environments. It augments them. It brings cloud computing out of the monolithic data centers controlled by large entities and brings it down to the realm of the masses. With the Billion Node Cloud, anyone who is capable can participate in the revenue generation aspects of cloud computing at a nominal amount of investment. Imagine, in a time of crisis, organizations being able to quickly utilize a huge cloud infrastructure that is spread out across the country and operates basically without a single point of failure.
General Security Concerns
Security of applications and data while in vehicles or in homes is a valid concern that will have to be addressed. Initially, encryption of data will be a priority for most situations. However, in the “cloud bidding” software for the spot market, various security levels can be advertised to customers and they can select which security level is appropriate for their compute needs. Not all cloud computing jobs need tight security and some need very little at all. For example, academic organizations and researchers may find the security level of the Billion Node cloud quite sufficient. There will be conscious trade-offs that weigh security, required compute volume and price for consumers to make the best choice for their particular cloud application. For those that need security, strong data encryption and network encryption can help with the operational security aspects of the solution. Hardware design can also assist security with the MicroServer devices being created with minimal access mechanisms and secure logins. Further, an ecosystem of certification of the security of a given installation can be developed, which again, creates jobs.
The ambitious concept of a Billion Node Cloud is not the realm of computing science fiction. The key components are available today in various stages of sophistication for this specific endeavor. It is simply a matter of having the will to turn this concept into a huge beneficial reality. Imagine how much computing can be accomplished with 8 Billion compute threads, 16 Billion GigaBytes of Memory and somewhere in the vicinity of 240 Billion GigaBytes of storage. In reality, by the time the world reaches One Billion Nodes of cloud compute capacity, those numbers will most assuredly be larger because of the persistent march of Moore’s Law.
When examining this vision, it is important to see it with the eyes and mind of the future. In one or two generations, the computing world will be substantially different. Students that get answers today from a powerful calculator or personal computing device will routinely submit huge questions to the vast cloud computing ecosystem that will exist. Things that only scientists do today with supercomputers, will be routine for children in the future. Today, we are working with computing that will be the philosophical equivalent of a slide rule of the past. In that future world, students will have routine discussions about the best way to use gravitational forces to get to the outer planets and beyond because they will have massive, inexpensive compute power at their disposal and know how to use it. Applications that do fantastic amounts of computing around weather modeling, or seismic predictions, or galaxy mapping will be readily available to anyone that wants to experiment with them. The concept of a Billion Node Cloud is just one part of this fantastic future where people have massive compute resources at their fingertips to help answer the most entertaining, instructional or vexing questions for humanity.
To realize the tremendous societal benefits of this concept, we need to begin today.
For more information on the Compute Nodes of the Billion Node Cloud concept, please go to http://www.crikit.info
Copyright 2012 Paul Morse – All rights reserved – may not re-publish without prior written consent from author.
Update – 11/14/2012 – The pace of creation of low-wattage MicroServers is accelerating. Since the posting of the Billion Node Cloud concept on 11/4/2012, there have been two major announcements of large organizations creating low-wattage processors. Intel and Texas Instruments both announced very capable processors that will run at less than 15 watts of electrical consumption. These announcements are part of the trend that makes the Billion Node Cloud a near surety for the future.
Update – 12/11/2012. Today Intel announced the first Low-Wattage Server Processors based on the Atom Architecture. This is a significance milestone for the Billion Node Cloud and Mobile Server Computing. At this time however, the low-wattage Xeon Processors are far more capable and provide significantly more compute horsepower for applications. The march is on and the bricks are being created for the Billion Node Cloud.
Update – 12/17/2012 According to El Reg, DARPA is working on 100 Gb/s wireless technology, which will help create the wireless environment for the Billion Node Cloud. http://www.theregister.co.uk/2012/12/18/darpa_100g_wireless/ If entire metro areas and highways get 100 Gb/s wireless speeds it would then be even easier to transmit applications and payloads to mobile data centers running low-wattage MicroServers. For any naysayers, based on recent trends of technology development, it is patently obvious that it is only a matter of time before the concept goes from semi-fiction to reality.