Presto

TLDR

Datacenter networks handle diverse workloads, but flow‑hashing load balancing causes congestion and existing solutions require centralized traffic engineering or specialized hardware. The authors propose Presto, a soft‑edge load‑balancing mechanism that offloads traffic management to virtual switches without altering transport layers or hardware. Presto is implemented as a soft‑edge load balancer, evaluated on a 10 Gbps testbed, and includes a TCP receive‑offload reordering mitigation strategy. Presto matches the performance of a single non‑blocking switch across workloads and adapts to failures and asymmetric topologies.

Abstract

Datacenter networks deal with a variety of workloads, ranging from latency-sensitive small flows to bandwidth-hungry large flows. Load balancing schemes based on flow hashing, e.g., ECMP, cause congestion when hash collisions occur and can perform poorly in asymmetric topologies. Recent proposals to load balance the network require centralized traffic engineering, multipath-aware transport, or expensive specialized hardware. We propose a mechanism that avoids these limitations by (i) pushing load-balancing functionality into the soft network edge (e.g., virtual switches) such that no changes are required in the transport layer, customer VMs, or networking hardware, and (ii) load balancing on fine-grained, near-uniform units of data (flowcells) that fit within end-host segment offload optimizations used to support fast networking speeds. We design and implement such a soft-edge load balancing scheme, called Presto, and evaluate it on a 10 Gbps physical testbed. We demonstrate the computational impact of packet reordering on receivers and propose a mechanism to handle reordering in the TCP receive offload functionality. Presto's performance closely tracks that of a single, non-blocking switch over many workloads and is adaptive to failures and topology asymmetry.