papers/lux-decentralized-cloud/sections/04-sla-enforcement-via-chain-receipts.tex

\section{SLA Enforcement via Chain Receipts}
\label{sec:sla}

Service level agreements in the centralized cloud are legal contracts enforced
through billing disputes and, occasionally, litigation. They are not technically
enforced. A provider that violates an SLA may issue a service credit---typically
a fraction of the monthly bill---at its own discretion.

In the Lux provider model, SLAs are smart contracts with stake-backed penalties.

\subsection{SLA Contract Structure}

An SLA between user $U$ and provider $\mathcal{P}$ for service $\mathcal{S}$
is encoded as a smart contract on the C-Chain:

\begin{equation}
    \mathsf{SLA}(U, \mathcal{P}, \mathcal{S}) = (\tau_{\max}, \alpha_{\min}, \delta_{\min}, p_{\text{epoch}}, \sigma_{\text{penalty}})
\end{equation}

where:
\begin{itemize}
    \item $\tau_{\max}$: maximum response latency
    \item $\alpha_{\min}$: minimum availability (fraction of time)
    \item $\delta_{\min}$: minimum durability class
    \item $p_{\text{epoch}}$: payment per epoch
    \item $\sigma_{\text{penalty}}$: penalty function for violations
\end{itemize}

\subsection{Violation Detection}

SLA violations are detected through three mechanisms:

\begin{enumerate}
    \item \textbf{Missing receipts.} If a provider commits to a service
    interaction but fails to produce a chain receipt within $\tau_{\max}$,
    the absence is detected by the SLA contract's monitoring logic.

    \item \textbf{Failed proofs.} If a storage provider fails a
    proof-of-spacetime challenge, the failure is recorded on-chain and
    triggers the penalty function.

    \item \textbf{Latency probes.} The network runs periodic latency probes
    from geographically distributed verifier nodes. Probe results are
    submitted on-chain and compared against the committed $\tau_{\max}$.
\end{enumerate}

\subsection{Penalty Execution}

When a violation is detected, the penalty function executes automatically:

\begin{equation}\label{eq:penalty}
    \sigma_{\text{penalty}}(v) = \min\left(\beta \cdot S_{\mathcal{P}}, \gamma \cdot p_{\text{epoch}} \cdot \mathsf{severity}(v)\right)
\end{equation}

where $\beta$ is the maximum slash fraction (governance-set, typically 5\%),
$S_{\mathcal{P}}$ is the provider's total stake, $\gamma$ is a multiplier
(typically $2\times$ to $10\times$ the epoch payment), and $\mathsf{severity}(v)$
maps violation types to severity scores.

Slashed funds are split: 50\% to the affected user as compensation, 50\% to
the protocol treasury. This creates a direct economic incentive for users to
report violations and for providers to maintain SLA compliance.

\begin{proposition}[Rational SLA Compliance]
A rational provider complies with SLA terms if and only if:
\begin{equation}
    \mathbb{E}[\text{revenue from compliance}] > \mathbb{E}[\text{savings from violation}] - \mathbb{E}[\sigma_{\text{penalty}}]
\end{equation}
Under the Lux penalty model with $\gamma \geq 2$, compliance is the dominant
strategy for any provider with stake exceeding the minimum threshold, provided
violation detection probability exceeds $1/\gamma$.
\end{proposition}