Data Decay: TIME

image
image

Starting point:

  • Energy use in cloud computing rises by approximately 10% to 30% annually. ( Digital Reality ) SCALE
  • Data centers consumed 7.4 gigawatts of electricity in 2023, up 55 percent from 4.9 gigawatts in 2022. U.S. regulators forecast that nationwide demand from data centers could rise from about 19 GW in 2023 to 21 GW in 2024 and reach roughly 35 GW in 2030. (Cushman Wakefield)
  • In the USA, data centers consume around 1.7 billion liters of water per day, a fraction of the nation’s total daily water use (1,218 billion liters). ( Researchgate )

——

Must points:

  • This is the clock. How much time does it take for our data to decay?
  • Plan to have it on the floor and keep the live counter of how many days have gone by since this data decomposition process has begun.
    • There’s compost and 24 worms inside
  • SCALE vs Locality of the server vs illusion of the cloud.

The hyperscaling of data centers around the world relies on large-scale extraction of resources, especially power and water, from locales, most often locales which are not beneficiaries of the computing happening. We would like to introduce to you, Data Decomposition, a new essential process to scaling up data centers which involves… Decomposing them.

FORM & TECHNOLOGY

  • Raspberry Pi hosting a website - TAP here - Live counter
  • Cloud
  • Data Center made up of cardboard.

Issues:

  • Tom Igoe = computing how many resources. He advised against it.

END GOAL

  • Have it on the floor and see how much time it actually takes to decompose, thus having the clock served its purpose.

Cloud computing = illusion of timeless, frictionless, infinite scale.

Decomposition = grounded, entropic, finite process.

The counter = makes it measurable and irreversible.

  • Tailscale = what lets us connect to the NYU Sandbox.
    • You can’t connect directly because of port forwarding.

Office Hours with Jeff

Breaking down the walls

  • When you request Solar Protocol, we can measure how much solar energy is being used.
    • Would this replace the networking we have? No.
  • What would happen to the Raspberry PI? It might just break.
  • Parking behind Navy Hard - New Lab - station where we can mount projects to.
  • Can design with how much energy we’re getting.
  • Voltaic:
  • Full sun unobstructed can capture a certain amount of sun. You can calculate in what time of the year or day the amount of energy / sun hours to expect for that location. You can model for expected weather.
  • Different sources of power: bio-based. Collecting methane.
  • First step would be to estimate the amount of energy we need. Rasperry Pi is especially difficult because it draws power in peaks.
  • MIT Robotics project - getings their energy from slugs. Robotic snail.

Voltaic documentation.