# Public report

The public package is derived from canonical local aggregate results and the existing public-portfolio chart set. It is a packaging operation only; no benchmark, cloud access, account access, or provider API call was performed.

Published aggregates: VPC throughput approximately 1.98–2.03 Gbit/s; sequential block storage up to approximately 452 MiB/s; selected random read/write approximately 6994 / 3006 IOPS; PostgreSQL at 32 clients approximately 10272 TPS and 3.055 ms; PostgreSQL at 64 clients approximately 9937 TPS and 6.249 ms; failed transactions 0.

The package excludes the resource ledger, raw network/storage/database logs, SSH incident raw material, provider portal captures, original PDF/DOCX, and files carrying address, host, account, connection, or local-path identity. The architecture figure is redrawn with generic labels because the source diagram exposed those fields.

The points apply only to recorded test conditions. They do not represent every region or size, do not form a complete performance audit, and do not create a production SLA. Evidence archive gaps remain. The package does not claim a single proven causal explanation for the relationship among CPU, load, queueing, throughput, and latency.