The Myth of Cut-Through Switching in Datacenter Networks

By Ori Aruj

February 10, 2009

Cut-through switching is mentioned as a requirement for datacenter switches. Originating from high performance computing (HPC) environments, cut-through switching was aimed to reduce network latencies to the minimum. In this article, we review the background behind cut-through switching, and examine the effectiveness of the cut-through scheme in typical datacenter networks.

What is Cut-through Switching?

Cut-through is a switching method for packet-switching systems, wherein the switch starts forwarding a packet before the entire packet has been received, normally, as soon as the destination address is processed. This technique reduces latency through the switch. Cut-through is compared to store-and-forward switching, that, as its name reflects, requires storing the entire packet before its transmission begins.

Cut-through switching became popular again in InfiniBand networks, since these are often deployed in symmetric environments such as supercomputer clusters, where latency may be a concern.

Latency gain for small-medium packets

The overall computing performance of an HPC system may be sensitive to the low latency of control packets. These packets are typically short (used for synchronization), and thus they are the focus of interest for cut-through switching. Long packets have less significance in this discussion.

When evaluating the performance of cut-through switching versus store-and-forward switching for small-medium packets (up to 256B/512B) both methods have about the same performance. This is because even a cut-through switch accumulates/stores 64B-512B chunks, depending on the micro-architecture, before it forwards it to the egress port.

For instance, note that Cisco’s Nexus 5000 series switches have cut-through switching and specify a minimum latency of 3.2us, a value that is equivalent to modern store-and-forward switches with packets up to 1KB.

Cut-through between ports of different speeds

Cut-through switching cannot operate when sending traffic from a slow port to a faster port. This, however, is a typical case as packets flow from servers through the datacenter access switches (top-of-rack or end-of-row).

Let us evaluate uplink traffic in a top-of-rack (TOR) switch attaching up to 40 servers each connected with a 10 Gigabit Ethernet (GbE) port. The uplink ports connecting the TOR switches to an aggregation/core switch are 40GbE, going to 100GbE.

Traffic from the slower 10GbE server connection to the faster 40GbE/100GbE uplink port cannot support cut-through. The uplink port sends the packet faster than the rate it is being received from the slower port. Transmission cannot start before the packet is fully received. Otherwise the uplink port TX FIFO might under-run. Hence, packets between a slower port and a faster port are store-and-forward even in a cut-through capable switch.

Cut-through between equal port speeds

Cut-through is not likely to occur between two ports of the same speed.

By definition, a packet can be cut-through only if the egress port is empty. As packet traffic is bursty, a few under-utilized ports sending cut-through traffic to the same egress port will experience intermittent egress port congestion. One can show that even networks with low utilization — as little as a few percentages — would not be able to use cut-through in most cases, and would not permit the establishing of deterministic low-latency bound that cut-through switching allegedly guarantees. Furthermore, with today’s server virtualization technologies the server port is unlikely to be under-utilized.

To conclude, under most utilization scenarios, packets between a few ports of the same speed are store-and-forward even in cut-through enabled switches.

Summary

The key value of cut-through is to enable low latency for traffic switched across the datacenter network. After examination, it seems that cut-through does not have a practical use in Ethernet-based datacenter switches. The new generation of store-and-forward switches present latencies that are equivalent and deterministic for the small-medium packets that matter. Latencies of new store-and-forward switches have no impact on the performance of real applications versus cut-through switches.

Cut-through switching does not come for free. Devices that support cut-through typically do not support: (a) large buffers for congested TCP traffic, (b) system capacity scalability, and (c) clear channel 40G and 100G ports across the fabric — or if they do then fallback to store-and-forward.

Typical datacenter networks are multi-tier, multiplying delay with the number of hops. This is why system vendors focus on building large, strictly non-blocking switches. If the switch is not strictly non-blocking (e.g., XAUI based switch), the probability for a packet to be unnecessarily delayed multiplies, not just due to intermittent oversubscription to one port, but also due to artificial (i.e., blocking) oversubscription. As a result, the original drive to provide cut-through switching actually produces increased delay and latency, contradicting the objective.

In conclusion, when building a datacenter switch, it is far more important to have an architecture that provides deterministic latency under all traffic conditions, strictly non-blocking configurations from 10’s to 1,000’s of ports, effective 40GbE and 100GbE port support, and buffers that ensure reliable network operation under real network conditions.

Subscribe to HPCwire's Weekly Update!

Be the most informed person in the room! Stay ahead of the tech trends with industry updates delivered to you every week!

MLPerf Inference 4.0 Results Showcase GenAI; Nvidia Still Dominates

March 28, 2024

There were no startling surprises in the latest MLPerf Inference benchmark (4.0) results released yesterday. Two new workloads — Llama 2 and Stable Diffusion XL — were added to the benchmark suite as MLPerf continues Read more…

Q&A with Nvidia’s Chief of DGX Systems on the DGX-GB200 Rack-scale System

March 27, 2024

Pictures of Nvidia's new flagship mega-server, the DGX GB200, on the GTC show floor got favorable reactions on social media for the sheer amount of computing power it brings to artificial intelligence.  Nvidia's DGX Read more…

Call for Participation in Workshop on Potential NSF CISE Quantum Initiative

March 26, 2024

Editor’s Note: Next month there will be a workshop to discuss what a quantum initiative led by NSF’s Computer, Information Science and Engineering (CISE) directorate could entail. The details are posted below in a Ca Read more…

Waseda U. Researchers Reports New Quantum Algorithm for Speeding Optimization

March 25, 2024

Optimization problems cover a wide range of applications and are often cited as good candidates for quantum computing. However, the execution time for constrained combinatorial optimization applications on quantum device Read more…

NVLink: Faster Interconnects and Switches to Help Relieve Data Bottlenecks

March 25, 2024

Nvidia’s new Blackwell architecture may have stolen the show this week at the GPU Technology Conference in San Jose, California. But an emerging bottleneck at the network layer threatens to make bigger and brawnier pro Read more…

Who is David Blackwell?

March 22, 2024

During GTC24, co-founder and president of NVIDIA Jensen Huang unveiled the Blackwell GPU. This GPU itself is heavily optimized for AI work, boasting 192GB of HBM3E memory as well as the the ability to train 1 trillion pa Read more…

MLPerf Inference 4.0 Results Showcase GenAI; Nvidia Still Dominates

March 28, 2024

There were no startling surprises in the latest MLPerf Inference benchmark (4.0) results released yesterday. Two new workloads — Llama 2 and Stable Diffusion Read more…

Q&A with Nvidia’s Chief of DGX Systems on the DGX-GB200 Rack-scale System

March 27, 2024

Pictures of Nvidia's new flagship mega-server, the DGX GB200, on the GTC show floor got favorable reactions on social media for the sheer amount of computing po Read more…

NVLink: Faster Interconnects and Switches to Help Relieve Data Bottlenecks

March 25, 2024

Nvidia’s new Blackwell architecture may have stolen the show this week at the GPU Technology Conference in San Jose, California. But an emerging bottleneck at Read more…

Who is David Blackwell?

March 22, 2024

During GTC24, co-founder and president of NVIDIA Jensen Huang unveiled the Blackwell GPU. This GPU itself is heavily optimized for AI work, boasting 192GB of HB Read more…

Nvidia Looks to Accelerate GenAI Adoption with NIM

March 19, 2024

Today at the GPU Technology Conference, Nvidia launched a new offering aimed at helping customers quickly deploy their generative AI applications in a secure, s Read more…

The Generative AI Future Is Now, Nvidia’s Huang Says

March 19, 2024

We are in the early days of a transformative shift in how business gets done thanks to the advent of generative AI, according to Nvidia CEO and cofounder Jensen Read more…

Nvidia’s New Blackwell GPU Can Train AI Models with Trillions of Parameters

March 18, 2024

Nvidia's latest and fastest GPU, codenamed Blackwell, is here and will underpin the company's AI plans this year. The chip offers performance improvements from Read more…

Nvidia Showcases Quantum Cloud, Expanding Quantum Portfolio at GTC24

March 18, 2024

Nvidia’s barrage of quantum news at GTC24 this week includes new products, signature collaborations, and a new Nvidia Quantum Cloud for quantum developers. Wh Read more…

Alibaba Shuts Down its Quantum Computing Effort

November 30, 2023

In case you missed it, China’s e-commerce giant Alibaba has shut down its quantum computing research effort. It’s not entirely clear what drove the change. Read more…

Nvidia H100: Are 550,000 GPUs Enough for This Year?

August 17, 2023

The GPU Squeeze continues to place a premium on Nvidia H100 GPUs. In a recent Financial Times article, Nvidia reports that it expects to ship 550,000 of its lat Read more…

Shutterstock 1285747942

AMD’s Horsepower-packed MI300X GPU Beats Nvidia’s Upcoming H200

December 7, 2023

AMD and Nvidia are locked in an AI performance battle – much like the gaming GPU performance clash the companies have waged for decades. AMD has claimed it Read more…

DoD Takes a Long View of Quantum Computing

December 19, 2023

Given the large sums tied to expensive weapon systems – think $100-million-plus per F-35 fighter – it’s easy to forget the U.S. Department of Defense is a Read more…

Synopsys Eats Ansys: Does HPC Get Indigestion?

February 8, 2024

Recently, it was announced that Synopsys is buying HPC tool developer Ansys. Started in Pittsburgh, Pa., in 1970 as Swanson Analysis Systems, Inc. (SASI) by John Swanson (and eventually renamed), Ansys serves the CAE (Computer Aided Engineering)/multiphysics engineering simulation market. Read more…

Choosing the Right GPU for LLM Inference and Training

December 11, 2023

Accelerating the training and inference processes of deep learning models is crucial for unleashing their true potential and NVIDIA GPUs have emerged as a game- Read more…

Intel’s Server and PC Chip Development Will Blur After 2025

January 15, 2024

Intel's dealing with much more than chip rivals breathing down its neck; it is simultaneously integrating a bevy of new technologies such as chiplets, artificia Read more…

Baidu Exits Quantum, Closely Following Alibaba’s Earlier Move

January 5, 2024

Reuters reported this week that Baidu, China’s giant e-commerce and services provider, is exiting the quantum computing development arena. Reuters reported � Read more…

Leading Solution Providers

Contributors

Comparing NVIDIA A100 and NVIDIA L40S: Which GPU is Ideal for AI and Graphics-Intensive Workloads?

October 30, 2023

With long lead times for the NVIDIA H100 and A100 GPUs, many organizations are looking at the new NVIDIA L40S GPU, which it’s a new GPU optimized for AI and g Read more…

Shutterstock 1179408610

Google Addresses the Mysteries of Its Hypercomputer 

December 28, 2023

When Google launched its Hypercomputer earlier this month (December 2023), the first reaction was, "Say what?" It turns out that the Hypercomputer is Google's t Read more…

AMD MI3000A

How AMD May Get Across the CUDA Moat

October 5, 2023

When discussing GenAI, the term "GPU" almost always enters the conversation and the topic often moves toward performance and access. Interestingly, the word "GPU" is assumed to mean "Nvidia" products. (As an aside, the popular Nvidia hardware used in GenAI are not technically... Read more…

Shutterstock 1606064203

Meta’s Zuckerberg Puts Its AI Future in the Hands of 600,000 GPUs

January 25, 2024

In under two minutes, Meta's CEO, Mark Zuckerberg, laid out the company's AI plans, which included a plan to build an artificial intelligence system with the eq Read more…

Google Introduces ‘Hypercomputer’ to Its AI Infrastructure

December 11, 2023

Google ran out of monikers to describe its new AI system released on December 7. Supercomputer perhaps wasn't an apt description, so it settled on Hypercomputer Read more…

China Is All In on a RISC-V Future

January 8, 2024

The state of RISC-V in China was discussed in a recent report released by the Jamestown Foundation, a Washington, D.C.-based think tank. The report, entitled "E Read more…

Intel Won’t Have a Xeon Max Chip with New Emerald Rapids CPU

December 14, 2023

As expected, Intel officially announced its 5th generation Xeon server chips codenamed Emerald Rapids at an event in New York City, where the focus was really o Read more…

IBM Quantum Summit: Two New QPUs, Upgraded Qiskit, 10-year Roadmap and More

December 4, 2023

IBM kicks off its annual Quantum Summit today and will announce a broad range of advances including its much-anticipated 1121-qubit Condor QPU, a smaller 133-qu Read more…

  • arrow
  • Click Here for More Headlines
  • arrow
HPCwire