Thread by @2youngBadazz

cloud/fog/edge:

i havent forgotten about the original target, i am drawing a big circle around elon musk and dont want to rush the process. besides, the benefit of 'exposing' elon musk comes from placing him into context

'cloud computing', 'fog computing' etc are terms that sound gimmicky and annoying. 'cloud computing' itself has been overloaded to mean multiple unrelated things, which is alienating and discourages people from seeking a concrete understanding.

these terms, cloud or fog or edge, refer to a system of distributed computing. one of the main arcs in the history of high technology is the shifting around of computational power, particularly back and forth between end user device and centralized facilities:

in the early years computational power was centralized. to allow many users to access it, systems of time sharing were developed (see MULTICS). this is one significant thrust in the arc, access to centralized computing power.

in the early to mid 60s, the idea of centralized computing as a public utility was widespread theatlantic.com/past/docs/unbo…

and already the issues with achieving this were well understood

then came a shift to microprocessors, and the ability to deliver computational power in the user's own device

and this led to the rise of personal computers in the 1980s. where computational power is available, is subject to various concerns from the ruling class perspective, for example a shift towards end-user devices meant less surveillance but also wider faster adoption.

the history of distributed computation is a balancing act of these concerns and goals. re: goals, think again to the kill chain. pc and commercial internet improve early links in the chain, but there are limits to their utility because pc are in a fixed location.

without mobile devices, there is no way to truly track or monitor targets. mobile devices must sacrifice computing power to achieve small size. but they also need to be engaging to encourage adoption:

the entire 'silicon valley' gambit is to disguise a global military transformation as a consumer industry, to place the entire world in the crosshairs and launder this as consumer choice. people may see through it but its been very successful regardless.

one company that is famous for addressing mobile issues in the 90s is an apple spinoff called General Magic. 'cloud' predates 94 contra to the image but the company was nonetheless important in advancing centralized computation in personal devices.

the company was another 'get investors to pay for it' scheme, labbing out issues with mobile devices on investor dollars, and then its staff spun out and created the REAL mobile device industry

in the 90s it was already recognized that to conserve battery and deal with computing limitations, mobile devices would need to offload computation to an external location. mobile faced the same problems as the utility computing paradigm, lack of adequate network infrastructure.

the infrastructure laid during dotcom boom solved this problem, per early cloud computing book 'the big switch'

remember after the bust the infrastructure began to come back online in earnest around 2006. this is when they said the fiber glut was over. this roughly inaugurates the era of 'cloud', when this term began to be used in earnest.

06 through 08 marks the big initial push of PR related to 'cloud computing'. google started to offer an online document software, and companies like IBM started to push towards utility computing.

its worth separating what is encompassed by 'cloud computing' into separate ruling class goals. one major thrust was to push everything into centralized locations in order to capture more data. centralized 'cloud' software storage *and* infrastructure all enable

unprecedented gathering and processing of data. i will cover 'big data' as a separate topic, the focus here is the maturation of the ruling class' distributed computing paradigm, the delivery of computing power.

so in the 'beginning', there was the 'cloud'. offloading mobile device computation was something examined since the 90s, as shown by general magic. here is a 2001 paper referring to this as 'cyber foraging'.

a more widely used term now is 'mobile offload' or 'cloud offload'. use cases include image processing (face rec) and natural language processing (siri).

these things, and indeed any pure 'mobile to cloud' offload i think still relates more to the gathering of 'big data'.

so what do i mean by saying that distributed computing is a separate thrust? pure cloud offload alone has significant problems when it comes to achieving *low latency*.

the surveillance system exists to enable some kind of kinetic response. *thats why* its important to achieve low latency. this is still one of the greatest weaknesses in the kill chain. 'cloudlets', or 'fog' computing are the solution to this problem.

'fog' computing refers to computational power delivered *closer* to the target than central cloud processing. it is an extension of the cloud, *closer* to the device that requires offload.

there is deliberate confusion of the terms 'fog' and 'edge' computing, but roughly speaking:

cloud refers to centralized, powerful and highly secure computational infrastructure.

fog refers to outposts of cloud computing power closer to the target.

edge refers to computation

done on the device itself.

it should be becoming very clear that the 'military' and 'consumer' paradigm for delivering computation are indeed the same and solve the same problems

interestingly, in the 2005-2007 era of early cloud computing there were several projects to modularize 'data centers', put them in shipping containers.

a lot of hype aside this never took off though it saw military usage even dating back to the ~05 era.

this was i think a case of a technology being trialed before its time, the concept is being revived as a flexible way to deliver fog/edge computing power (only a relevant paradigm since ~2014), notably by azure:

'azure space' is a very new microsoft offering that includes azure modular datacenters with an option for satellite internet powered by starlink.

as the picture mentions, aws is also expanding its space efforts. i highlighted iridium because we will come back to this company at some point, they were one of the companies that trialed LEO internet back in the 90s.

so the 'cloud', the distributed network of computing power, is extending up into space and out towards the very edge where most of the sensors and devices live, and this period marks a very significant degree of maturity i think. many of the pieces of the puzzle are now in place.

in addition to the possibilities of LEO internet, 'cloud providers' are partnering with telcos to place fog/edge computing power closer to the mobile devices requiring computational offload.

one of these offerings is aws wavelength:

another is 'azure edge zone'

these imo are more usefully thought of as 'fog' nodes because they are static and not distributed on edge *devices* themselves. again the jargon isnt as important as understanding the paradigm of distribution

in addition to shipping containers, which could conceivably be transported by autonomous vehicles, UAVs themselves have been studied as possible fog nodes:

using other UAVs in a swarm to distribute computational burden in conjunction with cloud connectivity has also been studied:

again to me this seems to be a paradigm reaching maturity where delivery of computation is entirely flexible and liquid, and sensors/devices can use complex reasoning to determine where to offload computation to. 'big data' that is less urgent can go towards highly secure clouds,

whereas data requiring low latency decision making can be sent towards fog or edge nodes. this is a major step towards the realization of what i imagine has been the plan since the 1960s.

the sloppiness of the earlier eras has given way to something that i think will more or less remain the shape of global computing architecture well into the future.

the 'next wave' of technologies will all be enabled by this flexible and low latency computing