Search papers, labs, and topics across Lattice.
This paper introduces Aquifer, a novel system that leverages a hierarchical memory pooling architecture combining CXL and RDMA to optimize MicroVM snapshot management in serverless computing. By characterizing snapshot images and employing a hotness-based snapshot format, Aquifer effectively reduces cold-start latency by eliminating zero pages and prioritizing hot working sets in low-latency CXL memory. The implementation demonstrates a significant performance improvement, achieving a 2.2x speedup in end-to-end invocation time compared to Firecracker and a 1.1x improvement over the next best alternative.
Aquifer cuts MicroVM cold-start latency in half by intelligently pooling memory across CXL and RDMA, transforming serverless computing efficiency.
Memory stranding wastes 25-35% of installed DRAM in production cloud clusters. Memory pooling over CXL and RDMA offers a remedy, but neither technology alone suffices: CXL provides low-latency, load/store-transparent access limited to a pod, while RDMA provides cluster-wide reach at higher latency with software overhead. A hierarchical architecture combining both tiers is the practical path forward, yet remains unexplored for MicroVM-based serverless computing, where snapshot restore latency is the dominant cold-start bottleneck. We present Aquifer, the first system to serve MicroVM snapshots from a hierarchical CXL+RDMA memory pool. A characterization of snapshot images reveals that the vast majority of pages are either zero or cold, enabling a hotness-based snapshot format that eliminates zero pages and places only the hot working set in the CXL pool while storing cold pages in the RDMA pool. Sharing these snapshots across hosts on CXL 2.0 multi-headed devices, which lack hardware cache coherence, requires Aquifer's ownership-based coherence protocol to ensure correctness. Finally, Aquifer uses a copy-based page serving mechanism pre-installs hot pages from CXL memory before MicroVM resume and demand-pages cold pages asynchronously from RDMA. On emulated CXL+RDMA hardware, Aquifer achieves a 2.2x geometric-mean speedup in end-to-end invocation time over Firecracker and 1.1x over the next best alternative.
