OSMay 20

ParaCell: Paravirtualized Secure Containers with Lightweight Intra-Container Isolation and Intent-Driven Memory Management

arXiv:2605.2090666.0
AI Analysis

ParaCell addresses the isolation-performance trade-off in secure containers for cloud and agent workloads, offering significant latency and memory improvements.

ParaCell introduces a paravirtualized secure container runtime that uses MPK-based intra-address-space isolation and intent-driven memory management to reduce latency by up to 57% and 79% over PVM, and by up to 33% and 88% over RunV, while saving up to 35.6% memory on agent workloads compared to HyperAlloc.

Secure containers isolate each container with its own kernel, mitigating shared-kernel attacks prevalent in traditional container systems. However, existing designs still face a fundamental isolation--performance trade-off. Nested-cloud deployments amplify the cost of VM exits and page-table management, while emerging agentic workloads expose bursty memory demand that requires fine-grained elasticity. We attribute this trade-off to two root causes. First, existing designs lack lightweight intra-container isolation primitives for frequent container user--kernel transitions. Second, the host treats container memory management as opaque, forcing reactive secondary faults and coarse-grained huge page mappings to amortize their cost. This paper presents ParaCell, a paravirtualized secure container runtime built on two insights. First, intra-address-space hardware protection primitives can provide lightweight intra-container isolation. ParaCell uses MPK-based XGates to isolate the container user and container kernel within a single address space, turning frequent user--kernel transitions into direct domain switches. Second, container kernel allocators already encode memory-management intent. ParaCell introduces Pager to interpose on allocation and free events, batch proactive GPA to HPA bindings and unbindings, and avoid reactive shadow page-table faults while preserving fine-grained memory elasticity. ParaCell is implemented as a drop-in replacement for RunV. Our experiments demonstrate that, across traditional cloud and emerging agent applications, ParaCell reduces latency by up to 57% and 79% over PVM, and by up to 33% and 88% over RunV, in bare-metal and nested setups, respectively. On agent workloads, ParaCell saves up to 35.6% memory compared with the state-of-the-art VM memory reclamation technique, HyperAlloc.

Foundations

The foundational work for this paper's niche, ranked by how specifically the neighbourhood builds on it — not by global fame.

Your Notes