Cloud computing simply transfers the burden of IT service delivery onto the cloud service provider.  Not surprisingly most of these providers are currently in the USA – with data centers in the USA.

Sadly most regions in the USA depend on dirty carbon-fired generating stations (oil, gas, coal) to provide electrical power, so it comes as no surprise that power-hungry data centers are dependent on greenhouse gas (GHG) emitting coal-burning power plants.

The most popular form of “clean” energy generation in the USA is to use nuclear power to heat water to drive steam turbines.

Although the power-generation part of the nuclear power story is arguably clean, there is still that pesky detail of how to dispose of the radioactive waste that results from the process.  Since that problem has not been solved and is literally “buried”, nuclear power is actually dirty.

Isn’t it strange that in the 21st century, IT is largely dependent on coal & steam?

Greenpeace recently did a survey of some of the largest and better-known Internet sites to raise an alarm about the dirty side of cloud computing:

• Apple’s largest data center is in Lenoir, NC (500,000 Sq Ft) with a dependence on 96% dirty power.  Apple is building an even larger facility nearby that will have the same dependency on dirty power.

• Yahoo’s 190,000 Sq Ft data center in Lockport, NY is 72% dependent on dirty power, while its largest facility in La Vista, NE (350,000 Sq Ft) is 93% dependent on dirty power.  Yahoos’ dirty power index is 86% based on the weighed average of these two facilities.

• Google’s two largest data centers are in Lenoir, NC (476,000 Sq Ft) and Dalles, OR (206,000 Sq Ft).  The Lenoir facility depends on 96% dirty power and the Dalles facility depends on 49% dirty power.  The weighted dirty power consumption index for Google is 82%.

• Microsoft’s 700,000 Sq Ft data center in Chicago is 99% dependent on dirty power, while its 470,000 Sq Ft data center in San Antonio, TX is 89% dependent on dirty power.  Microsoft’s 470,000 Quincy, WA data center is 100% clean energy powered (hydro).  The weighted average dirty power consumption index for Microsoft is 68%.

What Greenpeace doesn’t tell you is that these industry giants are all trying to improve their GHG-emissions.

Although Apple has been visibly reducing the carbon footprint of its products and has taken the high ground in responsibly accounting for its total product life cycle impact, it appears that its IT department has not yet focused on this problem.

Overall, Apple reports that its facilities, including data centers, account for 3% of its total life cycle GHG emissions.  In other words, Apple has a massive reduction challenge to solve in the manufacturing and use of its power-hungry products before it shifts its focus on internal IT impact.

Apple’s focus on lifecycle impact will have a larger collateral benefit on reducing GHG emissions globally.

Yahoo is building a large facility in Buffalo, NY that is expected to be hydro-powered, so we can expect that its dirty power footprint will fall somewhat in the near future.

Google is moving aggressively by limiting it’s power waste and investing heavily in renewable energy sources through its RE < C initiative.

Meanwhile, Microsoft appears to have the leadership position with 25% of its total energy consumption coming from renewable sources.

All of this leaves considerable room for improvement and Greenpeace is rightfully keeping the heat on cloud computing.

The definition of “unreasonable” under traditional legal interpretation means that it is unreasonable for you to be searched without probable cause. Under the Charter of Rights, a police officer who had reasonable cause to suspect that you were going to blow up an airplane would be justified to search you via a pat-down or, after arresting you, via a strip search.

The Charter protects us from being searched without any reason to do so. Boarding an aircraft is not a valid reason since all travellers have no intention of blowing up the aircraft.

In fact, given that attempts to blow up an airplane occur less than once a year, all travellers are innocent virtually all of the time. This is hardly probable cause for strip-searching all passengers.

Jennifer Stoddard, the Privacy Commissioner of Canada, believes that the ends justify the means. In a recent letter to the Ottawa Citizen and posted to her website, she outlined the 4-point test that she applied to this question.

The 4-point test applied by Ms. Stoddard starts with (1) “Is the measure necessary to address a specific risk?”. In other words are the means necessary to achieve the ends?

If so the ends justify the means as long as (2) they work, (3) the loss of privacy is proportional to the identified need (i.e. loss of privacy caused by the means is proportionate to the ends that are to be achieved), and (4) there is no less privacy invasive way of achieving the same end. It is all about the ends justifying the means.

Perhaps the reason why the Privacy Commissioner of Canada does not defend our privacy rights under the Charter of Rights and Freedoms is because she has no mandate to do so.

According to the Privacy Act that defines her office and duties, the Privacy Commissioner is limited to reviewing situatations only pertaining to the privacy of information about an individual and not the individual’s inaliable rights and freedoms. The letter on her website confirms that ”… it is neither our duty nor expertise to assess the aviation threat and risk assessments…”

In other words her office has no business making a decision on CATSA’s request to strip search Canadians – whether it is done virtually or otherwise.

Just because the Privacy Commissioner says it’s OK to do so doesn’t change that fact that full body scanning and pat-downs without probable cause is a violation of our Charter Rights.

The technical solution to this problem has been around for years in the form of IPv6 which uses a 128-bit address space. The size of this number is difficult to comprehend. For example, there are 80,000,000,000,000,000,000,000,000,000 IPv6 addresses for every single IPv4 address.

Or, looking at it another way you could assign 70,000,000,000,000,000,000,000 IPv6 addresses to every star in the known universe. Clearly address exhaustion with IPv6 is not a problem.

However for a variety of technical reasons, the interoperabilty of IPv4 and IPv6 is complex and in some cases very difficult. And despite almost two decates of research and experimentation, the best minds in the world have failed to figure out a seamless transition path. At best it seems that every system connected to an Internet backbone or access network will have to switch over to IPv6 at some point and, since this infeasible to happen overnight (more likely many years), this requires simultaneous mapping of IPv4 addresses that will still remain in use behind firewalls. 

Some solutions require the equivalent of having two addresses for everything (also infeasible).  For example, your favourite website would need a dual address for each of the IPv6-equipped and IPv4-equipped clients that visit it.  Imagine the overhead of assigning a second address to the 50 M+ Internet accessible websites!  The DNS (Domain Naming System) is of no help for this because of lack of widespread implementation of compatible mechanisms in thousands of currently deployed DNS servers.

Local networks will also have to be upgraded to IPv6 to assure a smoother transition.  That means that the cut-over to IPv6 will require a massive network investment over a small period of time (1 – 3 years).

Sounds like Y2K all over again doesn’t it?

Many countries (China, Japan, India, USA, France, Germany, UK, Norway, Netherlands, Russia, Ukraine, Australia) already have public IPv6 inter-networks in-service on either an experimental or production basis and are gearing up to interconnect them into a new IPv6 Internet.

Canada is notably absent in preparing for IPv6.  There is no official government policy on IPv6 adoption.  Worse yet, Canada’s leading communication research group, CANARIE (Canadian Advanced Network and Research for Industry and Education) is actually dragging its feet on pursuing IPv6.  CANARIE is a nonprofit corporation funded by IT and telecom vendors, research organizations and the federal government and the void created by the deadly combination of lack of public science policy and the collapse of Nortel is telling.

Our official national stance is to wait and see what the USA does!  This seems to be our national policy on just about everything these days.

In view of this, the recent proposal by a former Nortel executive to urge the Harper government to actually do something by investing in creating digital infrastructure jobs takes on new meaning. 

Why not invest the billions that we are about to otherwise spend on so-called “shovel ready” projects to actually prepare  Canada for the new Internet reality?

So if you were planning to win the war for the data center it seems obvious that you would have a product that has:

• high speed bus-based network with an ability to tag and stream traffic flows by priority (aka tag-switched Ethernet with Netflow and other QoS)
• ability to support a wide variety of densely packed mass storage (because nobody other than fruit flies want to actually be in the disk drive business)
• have a multi-core blade architecture featuring as much memory as possible (maximizing VM capacity)
• be able to pack lots of blades into a small physical space with an environmentally low footprint (green is not only good for the planet – it’s much cheaper to operate and your components live longer)
• avoid totally minimal margins on the blades by using proprietary network QoS to assure superior I/O performance for the same unit cost (since everyone has an Intel CPU you cannot gain strategic advantage from a superiour CPU performance)
• support a data-center wide control program consisting of distributed hypervisors running Windows and Linux guest O/S (aka VMWare, Hyper-V, or Citrix).  This allows you to have a data-center wide network-centric solution that interconnects all storage to all blade complexes and all blades within those complexes. (You want to source your hypervisor from others because it too will become undifferentiated commodity – that’s why most are already free.)
• have an advanced distributed workload management solution that can deal with both SOA-services and batch jobs (yes these still exist in real life and still do real work).  You want to own this because this is where a huge part of the value and differentiation is.
• have a centralized agentless manager to help you administer all these moving parts.  Being agentless it must rely on standards to actually perform the administration, but the level of intelligence built into it will directly lower the operational cost of management (i.e. wetware is expensive).  Hence it is key to own this part too.

With the exception of the last 2 points, we have just described Cisco’s UCS product plan.

The impact of Nortel on the global economy across the 115 year history of the company is impossible to count. 

Every time you pick up a touch tone phone, use digital communications of any kind, experience broadband Internet access enabled by optical technology, or DSL, or high speed wireless – you are using technology invented by Nortel.

Every time you access your bank or brokerage account online, or use your mobile phone, you are riding on one or more protocols designed by Nortel. 

The first corporate email system in the world was built by Bell Northern Research.  So was the first use of digital packet communications, high-speed fibre optic rings, etc.  These are the very foundations of the Internet.

Nortel’s impact on the tech sector extends far beyond communications.  Engineers at Bell Northern Research contributed enabling technology to the electronic design community, distributed computing, advanced man-machine interfaces such as speech recognition, visualization graphics, dignital signal processing, etc. 

Nortel’s patent portfolio extends across Wireline, Wireless, Datacom, Enterprise and Optical technologies and services.  As of December 31, 2007, Nortel had approximately 3,650 US patents and approximately 1,650 patents in other countries. In fact Nortel has consistently ranked in the top 70 in terms of number of granted U.S. patents since 1998. 

Nortel has received patents covering standards-essential, standards-related and other fundamental and core solutions, including patents directed to CDMA, UMTS, 3GPP, 3GPP2, GSM, OFDM/MIMO, LTE, ATM, MPLS, GMPLS, Ethernet, IEEE 802.3, NAT, VoIP, SONET, RPR, GFP, DOCSIS, IMS, Call-Waiting Caller ID and many other areas.  The term “standards-essential” means that the technology would not be viable without the contribution of Nortel’s engineers.

My own career at Nortel was relatively brief, but in the less than 10 years that I was there I personally witnessed meetings where Nortel’s engineers educated IBM, HP, Intel, Cadence, Mentor Graphics, Microsoft, and a hundred other companies on advanced technology.  The spin off impact of those meetings alone on the tech industry was incalcuable.  Intel actually modified silicon designs, HP introduced new products, and Cadence & Mentor acquired new technology to rev up their revenues.  These were non-patent related discussions.

Nortel was the largest spender on R&D in Canada through both direct investment in its own labs and through leveraged investment in university interaction.  Literally thousands of doctoral degrees in Canada were made possible though collaborative research with Nortel over the years.  Even the scaled back Nortel of today spends more than 1/3 of its salaries on R&D jobs for Canadians.

Yet McGuinty is proud of denying Nortel’s call of distress?  Shame on him.

We get what we vote for.  Our politicians both federally and provincially have demonstrated that they would rather prop up the resource sucking industries of the past than enable a modern Canadian economy of the future.

The fact that the digital economy can create more numerous, more interesting, and higher paying jobs for Canadians compared to the back-breaking and mind-numbing jobs of the resource and manufacturing sectors is completely lost on our politicians. 

Perhaps it is because we elect lawyers and not engineers to parliament?

Is the real problem with Canadian voters who sleepwalk their way to the polls if they bother to vote at all? Do Canadian parents not care about the quality of jobs that will be available for their children? 

Why do we tolerate this ineptitude from our politicians?

Yes Nortel’s management laid the seeds for its destruction.  John Roth in particular is to blame, as is his successor Frank Dunn who is now facing charges for misleading shareholders and gross stupidity. 

Nonetheless, allowing Nortel to die is the wrong policy decision for both the Canadian economy and the high technology sector of Canadian industry.  Write your MPP and MP and give them a shake!

Clearly the industry can do better and Microsoft’s recent Azure announcement certainly raises the bar in a way that might wake up some other interesting players. 

For example, Oracle appears to have been on the Cloud Computing (CC) sidelines because of their belief that CC is fashionable gibberish.  Looking at Amazon and others it is easy to understand this point of view.  However Azure changes the playing field substantially.

In true MSFT fashion, Azure consists of many things – a clear sign that MSFT is serious about winning market share in the cloud.  Some of the more interesting elements include

• the use of SQL Server databases (yes Virginia a real database with a real query optimizer and a real I/O optimizer),
• redirection of SOAP and REST interprocess communication via or into the cloud,
• a general purpose distributed cache manager (with cache coherence strategies built-in) and
• deep integration with a development toolset that in fact is already used by over 100,000 developers.

While Amazon relies on revolution and using EC2 necessitates a re-write of the I/O layer in most applications, Microsoft is promising smooth transition and evolution of many existing applications into the cloud. 

It’s not hard to imagine which will be more appealing to all those IT departments with more work than people!  If MSFT can acutally make Azure work as advertised sooner rather than later, Amazon will rapidly resemble the former Netscape (anyone remember them?). 

However the battle is far from over.

Although there are indeed benefits from aggregating primitive resources such as CPU and storage, the economics of this are entirely based on economies of scale.  Cloud computing solutions that offer undifferentiated access to commodity resources will compete solely on cost – not on value.

Small wonder that the early players are those with excess Internet-accessible CPU and storage.  Amazon has already invested in large computer farms to run its online retail business, so why not sell it’s excess capacity at marginal cost via EC2?

Doing real work on the cloud, however, requires using higher level resources such as websites, databases, and other middleware.  Social networking sites such as Facebook and various blogsites, already offer higher level resources (such as websites and media storage management) on a cloud basis.

Offering the necessary APIs and tools to provide access to higher level resources so that custom applications can be migrated to run on the cloud is a natual step forward – as demonstrated by Azure’s easy support for migrating entire SQL Server databases into the cloud and by Oracle’s support for live migration into Oracle Virtual Server.

Although there are a greater variety of higher level resources, ultimately this approach is still a supply-side resource aggregation play where the service is sold at its marginal cost.  Small wonder that Oracle offers it’s enabling virutal server for free.

Inevitably Oracle (or another established player) will enter the game with an equally competitive solution as MSFT is promising. For example, Citrix is already offering a cloud product suite that is far more compelling than Amazon’s. 

The only difference right now is that Oracle and Citrix are deciding to offer cloud-enabling software products and not cloud services, while Microsoft and Amazon are offering their cloud-enabled products to users of their proprietary cloud service.

The real end game will be fought over how well these competing cloud solutions can dynamically match the massive supply of resources to real-time workload demand.

From prior experience in creating and working with massively scalable grid computing products (such as Platform Computing, DataSynapse, Globus, etc.), dynamic resource management is far from trivial.  It involves managing both the demand-side as well as the supply-side of the virtualized environment. 

On the demand-side it requires rich scheduling of millions of requests for complex resources in seconds.  Workload is specified based on the resources required to complete it (e.g. this version of O/S, that version of database, etc.) in the form of Jobs or Transactions (depending on the type of workload).  Scheduling is based on matching resource requirements to resource availability in real time.

On the supply-side vituralization is used to segregate resource consumers and it requires the dynamic re-provisioning of supply in response to shifts in workload demand for different resource types.  For example, if you need more Apache and less IIS at any instant in time and the virtual pool of resources is re-provisioned accordingly by loading the necessary virtual machines on the fly.

By offering dynamic distributed resource management, cloud computing will move from supply-side cost economics to demand-side value economics.