How to build #smartplanet.Part 1. Infrastructure
“There was a time when every household, town, farm, or village had its water well. Today, shared public utilities give us access to clean water by simply turning on the tap; cloud computing works in a similar fashion. Just like water from the tap in your kitchen, cloud computing services can be turned on or off quickly as needed. Like at the water company, there is a team of dedicated professionals making sure the service provided is safe, secure, and available on a 24/7 basis. When the tap isn’t on, not only are you saving water, but you aren’t paying for resources you don’t currently need.” –Vivek Kundra, Federal CIO, United States Government, 2010.
I was not planning to talk about #infra the first thing in the scoping #smartplanet. Still, given how fast everything changes, I have decided to put it as a placeholder that I can revisit in six months and see if anything has changed.
When you start thinking about the equipment needed for the #smartplanet, what is already available ( radars, Starlink, satellites, fiber, towers, etc.), what will be needed, and what will NOT be needed, it gets muddy very fast. And I am not talking about FCC regulations ( and every counterpart of it in every country), but the full “workstream” development. Since this is a separate discussion, I will briefly cover “the cloud” part of the #infra.
So, first question: cloud or on-prem? What is on-prem for the planet version? I am getting dizzy just from calculating the high-level cost of the infrastructure. The ROI of #smartplanet will be discussed in the nearest future, but for now, let’s assume the price is not an issue. The location of the hubs/nodes is also irrelevant for now, as by default, it will be in some form of a grid covering the entire planet.
There are currently cloud providers such as Azure ( MSFT), AWS ( Amazon), GCS ( Google), and IBM. Of course, there are lesser-known companies, but if you are a startup, you are most likely to start with those four. AWS was the top choice in the past decade, but after COVID, Azure, and GCS are eating AWS’s lunch, IMHO.
As each of those offers AI/ML platforms and services, standing up the tech across the globe will be rather simple once the use case is defined. I do have a note for the Product Managers of those systems, though; try using it yourself, please, and see how clunky some of them are. From the TCO ( total cost of ownership), no system admin/manager wants to have to deal with 1500000 different services, which should have been wrapped under one ( and with one straightforward bill).
Okay, so now we have the theoretical data pipeline set up to the cloud, how much data will we transfer, and at what rate? For example, one of the use cases is to monitor the health of senior citizens and predict and prevent things like trips and falls ( that would be a fascinating implementation of the close loop — drones to deploy predicted fall?). Do we do it in nanoseconds, seconds, minutes, hours, or days? Now, each tracking variable would have it’s own time system. I just checked how many parameters ChatGPT has, and Google says it is 1.7 trillion. 1.7 trillion is fine because they do not update it in real life. They used it to train the network but do not constantly run updates on it. Or do they?
“Parameters are the numerical values that determine how a neural network processes input data and produce output data. They are learned from data during the training process, and they encode the knowledge and skills of the model. The more parameters a model has, the more complex and expressive it can be, and the more data it can handle.”
The #smartplanet would constantly evaluate itself for the “balance” state and adjust when it is not in balance. The Earth is doing it by itself right now if we look at the system from the #Daoist perspective, the yin and yang framework, but the discussion if we already have #smartplanet implemented by the #ancients is scheduled for another time.
So, to sum up, some (cloud) platform will be chosen to have at least partially cloud infra with the services such as
- Compute resources ( VMs and containers, this is a job for a pro architect to define the structure — or can AI do it itself?)
- Storage ( i.e. satellite video, etc.)
- Networking ( another task for AI to self-optimize the process?)
- Databases ( I still have some reservations about how AGI will be able to get out of the database constrain)
- Monitoring ( will it be a human or a robot doing this task?)
- Security and Compliance ( AI can probably generate ISO 27001 standards in a nano sec for the #smartplanet and even comply with it, but boy, there is so much to unpack here at a later day)
I probably forgot something, oh well…
Previous: https://medium.com/@tanyatalks1/can-we-humans-build-smartplanet-intro-8f735c13a943
