Related papers: Avoiding Cross-Datacenter Collective Congestion vi…
In today's data center, a diverse mix of throughput-sensitive long flows and delay-sensitive short flows are commonly presented in shallow-buffered switches. Long flows could potentially block the transmission of delay-sensitive short…
Today, considerable Internet traffic is sent from the datacenter and heads for users. The characteristics of connections served by servers in datacenters are usually diverse and varied over time, with continuous upgrades in network…
Disaggregation is an ongoing trend to increase flexibility in datacenters. With interconnect technologies like CXL, pools of CPUs, accelerators, and memory can be connected via a datacenter fabric. Applications can then pick from those…
Applications running in geographically distributed setting are becoming prevalent. Large-scale online services often share or replicate their data into multiple data centers (DCs) in different geographic regions. Driven by the data…
In data centers, the nature of the composite bursty traffic along with the small bandwidth-delay product and switch buffers lead to several congestion problems that are not handled well by traditional congestion control mechanisms such as…
Layerwise offloading reduces the GPU memory footprint of large diffusion transformer (DiT) inference by prefetching upcoming layers from host memory, but its effectiveness hinges on hiding prefetch latency behind per-layer computation. This…
As link speeds increase in datacenter networks, existing congestion control algorithms become less effective in providing fast convergence. TCP-based algorithms that probe for bandwidth take a long time to reach the fair-share and lead to…
The peculiar congestion patterns in data centers are caused by the bursty and composite nature of traffic, the small bandwidth-delay product, and the tiny switch buffers. It is not practical to modify TCP to adapt to data centers,…
We present MLTCP, a technique to augment today's congestion control algorithms to accelerate DNN training jobs in shared GPU clusters. MLTCP enables the communication phases of jobs that compete for network bandwidth to interleave with each…
Over the past decade, Supercomputers and Data centers have evolved dramatically to cope with the increasing performance requirements of applications and services, such as scientific computing, generative AI, social networks or cloud…
State-of-the-art congestion control algorithms for data centers alone do not cope well with transient congestion and high traffic bursts. To help with these, we revisit the concept of direct \emph{backward} feedback from switches and…
In this work, we provide the design and implementation of a switch-assisted congestion control algorithm for data center networks (DCNs). In particular, we provide a prototype of the switch-driven congestion control algorithm and deploy it…
Distributed deep learning (DDL) training systems are designed for cloud and data-center environments that assumes homogeneous compute resources, high network bandwidth, sufficient memory and storage, as well as independent and identically…
RDMA-empowered cloud services are gradually deployed across datacenters (DCs) with multiple paths, which exhibit new properties of path asymmetry, delayed congestion signals, and simultaneous flow routing collisions, and further fail…
The rise of LLMs has driven demand for private serverless deployments, characterized by moderate-sized models and infrequent requests. While existing serverless solutions follow exclusive GPU allocation, we take a step back to explore…
Urban traffic congestion is a chronic problem faced by many cities in the US and worldwide. It results in inefficient infrastructure use as well as increased vehicle fuel consumption and emission levels. Congestion is intertwined with…
We consider the problem of designing a packet-level congestion control and scheduling policy for datacenter networks. Current datacenter networks primarily inherit the principles that went into the design of Internet, where congestion…
Effective congestion control for data center networks is becoming increasingly challenging with a growing amount of latency sensitive traffic, much fatter links, and extremely bursty traffic. Widely deployed algorithms, such as DCTCP and…
Recently, high-speed and short-distance networks are widely deployed and their necessity is rapidly increasing everyday. This type of networks is used in several network applications; such as Local Area Networks (LAN) and Data Center…
As communication protocols evolve, datacenter network utilization increases. As a result, congestion is more frequent, causing higher latency and packet loss. Combined with the increasing complexity of workloads, manual design of congestion…