mirror of
https://github.com/lux-apps/papers.git
synced 2026-10-30 18:15:21 +00:00
Restructured from flat .tex/.pdf files into per-paper subdirectories. Added Antje Worring as co-author on all papers. Synced PDFs to pdfs/ for download site.
51 lines
2.7 KiB
TeX
51 lines
2.7 KiB
TeX
\section{Introduction}
|
|
\label{sec:intro}
|
|
|
|
The cloud computing industry is built on a fundamental inversion of ownership.
|
|
When a developer deploys an application to AWS, GCP, or Azure, the provider
|
|
assumes custodial control over the application's data, encryption keys, network
|
|
identity, and operational continuity. The user pays for access to their own data
|
|
through the provider's proprietary APIs. Switching providers requires rewriting
|
|
application code, migrating data through provider-controlled export tools (when
|
|
they exist), and accepting downtime measured in days or weeks. This is not a
|
|
service market. It is a custody arrangement.
|
|
|
|
The consequences are well-documented. Amazon Web Services holds approximately 31\%
|
|
of global cloud infrastructure spend \cite{synergy2024cloud}. Google Cloud and
|
|
Microsoft Azure hold a combined 33\%. Together, three companies control the
|
|
operational infrastructure for the majority of the internet's applications. When
|
|
AWS experiences an outage, large fractions of the internet become unavailable
|
|
\cite{aws2021outage}. When a provider decides to change pricing, deprecate a
|
|
service, or restrict access, customers have no recourse except migration---which
|
|
the provider has made as expensive as possible.
|
|
|
|
This paper presents an alternative: the Lux provider model, in which cloud
|
|
infrastructure providers are service markets, not data owners. The key insight
|
|
is a separation of concerns:
|
|
|
|
\begin{enumerate}[label=(\roman*)]
|
|
\item \textbf{Users own their keys.} All data is encrypted client-side with
|
|
keys stored in capsule vaults. No provider ever holds plaintext data or
|
|
decryption keys.
|
|
\item \textbf{Providers compete on service quality.} Relay latency, storage
|
|
durability, compute throughput, and recovery speed are the competitive
|
|
dimensions---not data lock-in.
|
|
\item \textbf{Exit is a first-class right.} Users can switch providers at
|
|
any time without data loss, because providers never hold the only copy of
|
|
user data in a format only they can read.
|
|
\item \textbf{SLAs are chain-enforced.} Service level agreements are encoded
|
|
as smart contracts with stake-backed penalties for violations, verified by
|
|
on-chain receipts.
|
|
\end{enumerate}
|
|
|
|
The remainder of this paper formalizes this model. Section~\ref{sec:model}
|
|
defines the five provider roles. Section~\ref{sec:reputation} introduces
|
|
the reputation protocol. Section~\ref{sec:sla} describes SLA enforcement.
|
|
Section~\ref{sec:markets} defines the compute and resource markets.
|
|
Section~\ref{sec:portability} proves the portability guarantee.
|
|
Section~\ref{sec:comparison} compares against centralized clouds.
|
|
Section~\ref{sec:economics} presents the economic model.
|
|
Section~\ref{sec:related} discusses related work.
|
|
Section~\ref{sec:conclusion} concludes.
|
|
|