We present a decade of evolution and production experience with Jupiter datacenter
network fabrics. In this period Jupiter has delivered 5x higher speed and capacity, 30 …

Memory disaggregation has attracted great attention recently because of its benefits in
efficient memory utilization and ease of management. So far, memory disaggregation …

We present the design, implementation and engineering experience in building and
deploying MegaScale, a production system for training large language models (LLMs) at the …

We expose the state of the art in the topic of one-way delay measurement in both traditional
and software-defined networks. A representative range of standard mechanisms and recent …

Given the wide adoption of disaggregated storage in public clouds, networking is the key to
enabling high performance and high reliability in a cloud storage service. In Azure, we …

Programmable packet scheduling enables scheduling algorithms to be programmed into the
data plane without changing the hardware. Existing proposals either have no hardware …

For the widely deployed ECN-based congestion control schemes, the marking threshold is
the key to deliver high bandwidth and low latency. However, due to traffic dynamics in the …

Increasingly stringent throughput and latency requirements in datacenter networks demand
fast and accurate congestion control. We observe that the reaction time and accuracy of …

Emerging Zoned Namespace (ZNS) SSDs, providing the coarse-grained zone abstraction,
hold the potential to significantly enhance the cost-efficiency of future storage infrastructure …

Data center networks are inclined towards increasing line rates to 200Gbps and beyond to
satisfy the performance requirements of applications such as NVMe and distributed ML. With …