Programming distributed systems has never been easier, but making them perform well and efficiently exploit available resources remains difficult. Fast programs on a single host slow to a crawl when given more resources, while systems designed for scale squander resources on modest problems.

Compute Express Link (CXL) 3.0 can turn stranded DRAM into a composable datacenter memory fabric, but hardware is scarce and cycle-accurate simulation is prohibitively slow for full workloads. CXLMemSim is an open-source software CXL memory simulator that keeps computation at native speed and models only CXL-affected memory behavior. Across SPEC CPU2017, Memcached, wrf, vector search, and llama.cpp, CXLMemSim stays within 11–15% of hardware and gem5 while completing a Redis-scale run in 45 minutes instead of gem5’s 73 hours. The public release at SlugLab/CXLMemSim provides PEBS/LBR/eBPF tracing, JSON topology descriptions, and reusable migration/cache policy hooks for CXL design-space exploration.

In recent years, the microservice architecture has become the standard technique for building scalable applications, yet it still confronts developers with numerous challenges. Splitting an application into separate pieces can lead to unpredictable performance, and the process of doing so is complex and labor-intensive. Although a variety of solutions exist to simplify distributed application development, their programming abstractions offer poor support for existing applications in need of scaling. In this paper, we propose Monolift, a new technique for developing distributed applications that prioritizes incremental adoption and support for legacy code. Our proposal treats distribution as a compiler pass, allowing users to guide the distribution process with lightweight code-level annotations, and supports dynamic distribution strategies based on performance metrics rather than development-time decisions. We show that our Monolift prototype can automatically synthesize a variety of distributed architectures from the same source code, each having different performance tradeoffs, and that Monolift's dynamic distribution mechanisms outperform static microservice architectures under varying run-time conditions.