/ eos

The granular information we would like to see in the requirements specification for a Block Producer.

The estimated requirements for CPU,RAM,Disk and Bandwidth based on transactions per second:

TPS Cores RAM Disk IOPS Bandwidth
1000
5000
100000
1000000

Reasoning: Helps with our future expansaion plans.

Requirement: R/W Latency requirements for RAM and fractional reserves (storing contract states).
Reasoning: Most of us are planning on starting small and then increasing amount of ram as the need increases. Also many are looking ahead to beyond 4TB solutions and having the R/W, latency requirements will help us plan for and test alternative solutions such as M.2 SSD drives.

Requirement: R/W Latency requirements for EOSIO storage (ipfs).
Reasoning: Although not being released until later this year, having some indication of the requirements will help us build base solutions that can scale without having to replace existing equipment.

Requirement: Parrallel processing cluster R/W & latency requirements.
Reasoning: At some point we expect to have multiple servers processing transastions in parallell. It would be good to know at what speeds & latency that communication needs to occur.

Requirement: Based on previous experience (BitShares, Steem) and based on your best guesses on how quick EOS.io will scale after launch… What bandwidth do you guesstimate will be needed for:
• EOS Storage (ipfs)
• Query Services

Additional Questions:

  1. Do we have a date when EOSIO Storage / IPFS functionality will be available? Also… Do we know anything on how it will work?
  2. What is the minimum amount of EOS to be staked for BP?
  3. Will block.one offer us help with planning and performing benchmark testing before the start?
  4. Would there ever be a reason why EOS could perhaps send certain tasks for processing by the GPU?
  5. Are there any other software development requirements as it relates to initial launch besides compiling and installing EOS?
  6. In regards to staking for ram. How do we calculate how much ram someone gets for 1 token? Will the calculations be based both on number_of_staked_tokens and on total_amount_of_ram, both of which can change?
  7. Are you planning on releasing something in depth in regards to Regional scaling?
  8. Do we have any information on how we will lock down bandwidth for the IPFS based on staked tokens?