Big server farms that can crunch big data are huge and they consume enormous amounts of electrical power. Servers now account for 10% of global electricity consumption, and the prospects are for continued escalation. “It’s the same amount of electricity that was used to light the entire planet in 1985” [i]. Usage of power by computer increases about 7% per annum, or doubles every decade[ii]. Server farms also have a major environmental impact. Cooling systems to cool server farms use about as much power as the servers consume[iii]. This represents enormous cost as well as challenges in siting server farms that must be located near large rivers or other massive bodies of water. Then there is the problem of thermal pollution.
Space offers solutions to this problem. A server farm at GEO can receive unmetered power from the sun 24/7 with the depths of space serving as an ultimate heat sink. If the architecture of the computing facility can be designed to use a constellation of small satellites, costs can be lowered further avoiding the cost of launching major facilities. Launch costs of small satellites are dropping and the ConnectX satellite modules are roughly the size of cantaloupes.
ConnectX is a startup that has announced readiness to offer big data processing capabilities at more than an order of magnitude reduction in cost to corporate users. Currently average data storage prices are $900 per terabyte per year. ConnectX projects their delivered could potentially be a fraction of current prices.
The significantly lower cost opens opportunities for many new uses of big data as well as for a much broader range of customers that can afford to buy data that they can use at a fraction of earlier costs. ConnextX has assembly an end to end technology platform that can create, process, encode, encrypt, decode, transmit, receive, decode, and decrypt big data of all types. The Company is developing modules for specific applications: fraud detection for the financial industry, churn for telco, product recommendations for retail, statistical process control for manufacturing, advertisement placement for marketing and other applications.
ConnectX sees its technology as competing with emerging quantum computing solutions, but with technology that does not require basic scientific advances. The technology is an engineering solution that uses known science but with data architecture that is uniquely suited to the tasks of big data computing. The platform that ConnectX is faster and more efficient at transmitting Big Data than known alternatives. At the much better price-performance that is offered users – whether companies, research organizations or governments, can store their entire data store and have it available for analysis. ConnectX has also developed new techniques for data encryption that provide needed security for sensitive data. The information will be accessible anywhere through any mobile device as well as stationary systems. The service will be a web-based service where the client can upload their data and receive back the analysis.
The ConnectX team includes spans a broad range of experience and unique knowhow and specialties – big data architectures and computer science, computer security, satellite launch and space operations. Lance Parker, the CEO, has a business and economics background, has multiple patents in mobile communications, data security and satellite systems. More about ConnectX can be found on their website – www.connectx.com.
Opening the power of computing in space
ConnectX is not the first firm to propose servers in space with the idea that energy and cooling, both of which are costly on Earth, are abundant and cheap in space. Skyserver is a concept developed by Keith Lofstrom whose concepts not only anticipate server farms in space but also construction of the servers to take full advantage of the microgravity environment to reduce mass by eliminating structural supports[iv]. While Skyserver is a fascinating technical concept, ConnectX has developed a full business model that is integrated with the novel technology to provide a very promising value proposition.
Space-based supercomputing opens exciting possibilities. Why transmit terabytes of data from satellite remote sensing to Earth that can be processed in space and only transmit the information requested by the user? Minute changes in topography could be analyzed and immediately direct higher resolution imaging of an object of interest. Only information that is relevant would need to be transmitted to Earth. While such computer architectures can be built with constellations of small satellites the development of such systems will drive the development of multiple technologies. Developing space-based supercomputing to serve terrestrial demand will also create the capabilities for advanced computing facilities in space that can take full advantage of the vacuum, microgravity and abundant solar energy available in space
Download PDF.Big data computing above the clouds
* * *
Vid Beldavs is a futurist now based in Riga, Latvia who has been involved with space policy since the mid-1970s when he was inspired by Gerard K. O’Neil and developed and taught “Space industrialization and national priorities” at Coe College on a developmental leave from his job in forecasting future trends for Cummins, a global manufacturer of diesel engines and power systems. Vid can be reached at firstname.lastname@example.org
[i] Bryan Walsh, 2013 “The Surprisingly Large Energy Footprint of the Digital Economy” retrieved 2014-09-29 from http://science.time.com/2013/08/14/power-drain-the-digital-cloud-is-using-more-energy-than-you-think/
[ii] Philip Bump, 2012 “The N.Y. Times tells us how much energy the cloud uses — but not why it matters “ , retrieved 2014-09-29 from http://grist.org/news/the-n-y-times-tells-us-how-much-energy-the-cloud-uses-but-not-why-it-matters/
[iii] Justin Lee 2012 “Energy usage of server farms”, retriewed 2014-09-29 from http://large.stanford.edu/courses/2012/ph240/lee1/
[iv] Keith Loftstrom, “Server Sky – Data Centers in Orbit” http://news.dice.com/2013/04/26/is-it-time-for-servers-in-space/