In datacenters, workload throughput is often constrained by the attachment bandwidth of proxy servers, despite the much higher aggregate bandwidth of backend servers. We introduce a novel architecture that addresses this problem by combining programmable network switches with a controller that together act as a network "Prism" that can transparently redirect individual client transactions to different backend servers. Unlike traditional proxy approaches, with Prism, transaction payload data is exchanged directly between clients and backend servers, which eliminates the proxy bottleneck. Because the controller only handles transactional metadata, it should scale to much higher transaction rates than traditional proxies. An experimental evaluation with a prototype implementation demonstrates correctness of operation, improved bandwidth utilization and low packet transformation overheads even in software.

This paper argues that the lack of explicit support for non-volatile main memory (NVMM) in network stacks fundamentally limits application performance. NVMM devices have been integrated into general-purpose OSes by providing familiar file-based interfaces and efficient byte-granularity access by bypassing page caches. However, this powerful property cannot be fully utilized unless network stacks also support it and applications exploit such support. This requires a thoroughly new network stack design, including low-level buffer management and APIs. We propose such a new network stack architecture to support NVMM and demonstrate its advantages for efficient write-ahead logging, a popular technique to implement transactions.