Subscribe to Blog Notification Emails

Latest Blog Post

Does Latency Matter in Cloud Networking?

Jayshree Ullal
by Jayshree Ullal on May 17, 2009 11:49:51 PM

Cloud Networking has arrived upon us. Take server density on a scale never seen before plus compute demanding application workloads and you get a class of non-blocking cloud networking that is more aware of dynamic application flows than vanilla packet switching.

A key design criterion here is Latency. Of course there are differing views on latency. Some may say it does not matter since you always run into the low speed WAN/internet or speed-of-light limit. Others say that, with build-outs of private or public cloud islands as a complement to the Internet, new application intensive workloads traversing cloud environments are demanding closer attention to latency. Both are right actually. It is really about different strokes for different folks in different parts of the networking segments.

Where is latency introduced?

Let’s first review the culprits and contributors of latency. End-to-End Latency is typically contributed by three factors in total application level latency:

Server Latency: The delay due to server-based hosts adapters, operating system, stacks and user space to transfer data from application buffers to cloud network

Network Latency: The delay due to processing L2 or L3 packet headers and switching the network payload through the network mesh from source to destination.

Resource Latency: Typically a storage or database resource contributing to delays due to transfer of data from application to cloud network

I recall the raging arguments in the prior decade about implementation of Ethernet switches and the discussions on "cut through" versus "store and forward". The issues centered on the deploying older store & forward techniques for preservation of packet integrity versus latency improvements with cut-through. Latency was less of an issue back then since application flows were kilobyte class (with email and file transfers) and network speeds were only 100 megabits transitioning to 1 gigabit Ethernet. Typical network latencies tolerated here were 100 to 1000 microseconds and most classic enterprise networks are still built that way with many hops across three tiers of oversubscribed networks.

New Cloud Networking Architecture

Clearly with Cloud Networking, one must anticipate wild matrix flows with overlapping peaks and valleys and move these flows without dropping in uniform manner. To deliver uniformity, Cloud Networking must be non-blocking, with low and predictable latency for all traffic types. Typical cloud networking latency demands are 1-10 microseconds with some clusters even demanding nanoseconds.

The new generation of 1Gb/10Gb Ethernet switches from Arista is designed with these key goals in mind. They deploy the best cut-through techniques intelligently. The cut-through architecture delays the packet transfer only enough to look up the Layer 2/3 headers and determine destination port. Additionally, a key parameter is optimal buffer capacity that ensures predictable latencies.

Latency in Financial Applications

Perhaps one of the most compelling and proven cases for low latency is algorithmic trading. The market data industry is able to adjust and input data near instantaneously, vying each other for a competitive edge on order transactions measured in fractions of microseconds! The Financial Information Exchange (FIX) protocol is widely used for real time information transactions and requires profiling protocols (such as Precision or Network Time Protocol, PTP/NTP) for latency measurements. Measuring and sampling latency delays accurately using Arista’s Extensible Operating System (EOS) and third party tools for optimal synchronization in microseconds is also becoming key factor.

In addition to Financial services, there are a growing number of Cloud Networking use cases that need low latency architectures such as high performance computing, web and database clusters, storage access, seismic analysis, large scale data analytics and virtualization. At Arista Networks we have pioneered the worlds best combination of 1Gb/10Gb Ethernet port density, footprint and latency with our 7xxx series of switches.

Welcome to the new ways of Cloud Networking! For those of you who will be at Interop, do visit us at Arista Networks Booth #1651.

As always I invite your views at


Architecting Low Latency Cloud Networks Whitepaper

Opinions expressed here are the personal opinions of the original authors, not of Arista Networks. The content is provided for informational purposes only and is not meant to be an endorsement or representation by Arista Networks or any other party.
Jayshree Ullal
Written by Jayshree Ullal
As CEO and Chairperson of Arista, Jayshree Ullal is responsible for Arista's business and thought leadership in AI and cloud networking. She led the company to a historic and successful IPO in June 2014 from zero to a multibillion-dollar business. Formerly Jayshree was Senior Vice President at Cisco, responsible for a $10B business in datacenter, switching and services. With more than 40 years of networking experience, she is the recipient of numerous awards including E&Y's "Entrepreneur of the Year" in 2015, Barron's "World's Best CEOs" in 2018 and one of Fortune's "Top 20 Business persons" in 2019. Jayshree holds a B.S. in Engineering (Electrical) and an M.S. degree in engineering management. She is a recipient of the SFSU and SCU Distinguished Alumni Awards in 2013 and 2016.

Related posts

The New Era of AI Centers

In 1984, Sun was famous for declaring, “The Network is the Computer.” Forty years later we are seeing this cycle come true...

Jayshree Ullal
By Jayshree Ullal - May 29, 2024
The Era of Microperimeters

Paradigm Shift to Zero Trust Networking

The new age of edge, multi-cloud, multi-device collaboration for hybrid work has given...

Jayshree Ullal
By Jayshree Ullal - April 30, 2024
The New AI Era: Networking for AI and AI for Networking*

As we all recover from NVIDIA’s exhilarating GTC 2024 in San Jose last week, AI state-of-the-art news seems fast and furious....

Jayshree Ullal
By Jayshree Ullal - March 25, 2024