& cplSiteName &

Are You Really Getting 400 Gbit/s Performance?

Sterling Perrin
9/11/2013
0%
100%

Driven by continuing IP traffic growth, a new generation of 400 Gbit/s network processors is beginning to emerge. Such processors have been announced by a handful of vendors, with more to come.

However, scaling from 100 Gbit/s to 400 Gbit/s processors is not an easy feat. Recently, we witnessed some 400 Gbit/s network processor performance testing in a lab and were surprised by some of the results. Based on this data, we believe that there may be some wide variations between what is claimed on spec sheets and the real world performance that is achievable.

We believe that service providers evaluating new higher capacity systems must perform extensive testing on their own in order to ensure that the high-end systems perform as advertised under a variety of conditions that mirror real-world traffic mixes and traffic growth. Otherwise, service providers may be stuck with true performance that is far below the advertised 400G sticker. In fact, under certain conditions, true performance may not even be better than the legacy routers that are being replaced.

One important test is the packet sweep test. The packet sweep tests router performance over a spectrum of different packet sizes, from small (64 bytes) to large (1,500 bytes or greater), in order to ensure consistent performance across the full range of incoming packet variants. It is somewhat counterintuitive, but the greatest demands on the processor occur at the smallest packet sizes. The reason is that small packet sizes force greater amounts of table look-ups per second, and this burdens the processor.

The test we reviewed showed a 400 Gbit/s processor consistently dropping packets at all sizes, both small and large. As the packet size increased, the performance did not improve. On a few occasions, packets dropped by as much as 50 percent, and we never saw more than 80 percent throughput at any point in the test.

A second test, known as the Internet mix (or IMIX) test, was reviewed. Just as there is no typical packet size, there is no typical service provider IP traffic mix. It varies greatly from provider to provider. The test we reviewed used a couple of different real-world IMIX samples provided by service providers, based on their network scenarios. The results showed poor performance across a range of IMIX profiles. The processor appeared to suffer from problems related to storing packets for lookup, which resulted in severe performance impact on all traffic types tested. Significantly, when comparing two generations of processor performance, the new 400 Gbit/s processor demonstrated anywhere from 20 percent to 50 percent less performance based on serviceable bandwidth.

A third test measured the performance of a processor while enabling service level agreements (SLAs). Here, the results showed that enabling SLAs on connections reduced the actual throughput of the processor from 400 Gbit/s to 100 Gbit/s, meaning that only 50 percent of the advertised 100 GigE ports would be available with SLAs turned on.

The tests above are tied closely to the conventional views of scale, meaning the ability to handle more and more bits through the processor and on the network. As the Internet evolves from person-to-person communications to the machine-to-machine dominated Internet of things, another component of scale is becoming increasingly important: the ability to handle more and more flows through the network.

For next-generation routers, this means they must not only process massive amounts of bits, but they must also be able to process massive numbers of flows. Routers equipped with 400 Gbit/s processors must be tested in their abilities to handle tens of thousands of flows. As a rule of thumb, a 10 GigE port will typically serve 500 customers. If we assume just four classes of service per customer (a low assumption), this creates 2,000 different flows per 10 GigE port. With 20 x 10 GigE ports per line card (the state-of-the-art in existing designs), this creates 40,000 different flows per card. As we move to an Internet of things, it is very possible that these next-gen routers could hit performance limits on supported flows long before they hit their maximum capacities in bit/s. It is another dimension of scale that must be accounted for and tested in evaluating core routers.

The final point we will touch upon is efficiency in power consumption and footprint. It is well understood that these opex factors are of critical importance to service providers. In some regions, where power costs are well above global averages or where equipment is deployed in dense urban areas where space is severely limited, power and space requirements may make or break a buying decision. Even in the US, however, we have had discussions with large service providers who place power and space at the top of their lists.

Here again, service providers need to dig deeper than the power and space requirements that are advertised. To get a true understanding of space and power, these advertised specs must be placed in the context of the overall performance of the system -- as determined by packet sweep tests, IMIX tests, flow limitations, and any other tests that are performed on the system.

For example, if performance testing and evaluations show that a system will perform at 80 percent of its capacity limits under real-world network conditions, then this needs to be taken into account for space and power consumption. In this case, 20 percent more space and 20 percent greater power consumption would be needed to achieve 400 Gbit/s of capacity. Failing to account for any performance limitations sets service providers up for unwelcome opex surprises when these new systems start to fill up in the network.

In summary, vendors rolling out new 400 Gbit/s silicon promise routers that are smarter, faster, and greener than previous generations, but some in-depth test results we've seen indicate that this may not be the case. Service providers cannot afford to rely on vendor claims in this area. Rather, extensive testing -- of the kind we’ve described in this article -- is a must. The added time and costs of upfront testing will pay dividends for years to come.

This blog was commissioned by Cisco Systems. The blog was created independently of Cisco, and Heavy Reading is wholly responsible for its contents.

— Sterling Perrin, Senior Analyst, Heavy Reading


Interested in learning more on this topic? Then come to Ethernet & SDN Expo, a Light Reading Live event that takes place on October 2-3, 2013 at the Javits Center in New York City. Co-located with Interop, Light Reading's Ethernet & SDN Expo will focus on how the convergence of Carrier Ethernet 2.0 with emerging carrier software-defined networking (SDN) and network functions virtualization (NFV) technologies could change the whole telecom landscape for service providers. For more information, or to register,
click here
.


(15)  | 
Comment  | 
Print  | 
Newest First  |  Oldest First  |  Threaded View        ADD A COMMENT
Page 1 / 2   >   >>
mfaisalk
50%
50%
mfaisalk,
User Rank: Lightning
9/14/2013 | 11:05:32 PM
400G performance
We need to be careful with our judgements here. It is one vendor testing another vendor. What good you can expect here ?

 

Faisal
David Burns
50%
50%
David Burns,
User Rank: Light Beer
9/13/2013 | 2:05:16 PM
Check the math
Quote (snipped for conciseness):
if performance testing and evaluations show that a system will perform at 80 percent of its capacity limits under real-world network conditions ... 20 percent more space and 20 percent greater power consumption would be needed to achieve 400 Gbit/s of capacity.

Wrong. If actual performance is 20% lower, you need to provision 25% more equipment to achieve the targeted performance. Simple example: 8 x 400 Gbit/s should give you 3200 Gbit/s. If processors perform at 80%, actual performance of one processor is 320 GBit/s, so you need 10 to reach your target - 25% more than 8.
desiEngineer
50%
50%
desiEngineer,
User Rank: Light Beer
9/13/2013 | 1:17:08 PM
Re: Which network processor tests?
Sterling,

If you can guarantee that the tests were fair and impartial, then yes, enquiring minds want to know.

If not, don't attach your name (and LightReading's), and then hide behind some "this is newsworthy" BS.  And I don't care how sophisticated and expensive the lab is, I am not ingenuous enough to believe that the lab was created for the purpose of the general public.  Clue#1: I haven't seen one report come out of that lab that showed a single Cisco product in a negative light.

No vendor has a vested interest in impartiality.  Knowing how much marketing is attached to all claims, it behooves the operators to do their due diligence, but to pretend that Cisco can do that due diligence for all of us is naive at best.

-desi
Surfer Blue
50%
50%
Surfer Blue,
User Rank: Light Beer
9/13/2013 | 12:54:07 PM
Re: Which network processor tests?
Service providers do extensive testing before deploying any new hardware in their networks.  FP3 cards have been available for a couple of years and are widely deployed – clearly they have been thouroughly vetted in the field.  Personally I'd take real world testing and deployment experience over contrived competitor lab testing any day.  

 
sterlingperrin
50%
50%
sterlingperrin,
User Rank: Lightning
9/13/2013 | 9:36:34 AM
Re: Which network processor tests?
Yarn,

You are correct - I have no interest in standing in the middle. Here's the thing. Cisco has a test lab that would be the envy of all but the very largest tier 1 operators. They conduct very thorough and very expensive testing of their competitors' products, which they've been doing for 13 years (probably not something they want publicized). In this case, they came to us - and some other media as well - with information from their labs that was compelling and newsworthy.

Are we saying that Cisco's results are conclusive about ALU, or any particular competitor? Absolutely not!! But we do believe there was enough in there to raise broad questions and write this piece.

I have received calls from people at ALU who are understandably upset. They are getting calls from customers who are asking questions. It's not an easy thing to go through, I'm sure, but as media, it is our job to bring the right topics of debate to the forefront and spark the dialogue. We made a judgment that 400G performance testing is a legitimate topic and that questions should be asked of all 400G NP players - ALU, Cisco, Huawei, Juniper, and any other market entrants that emerge.

Sterling

 
mendyk
50%
50%
mendyk,
User Rank: Light Sabre
9/12/2013 | 7:28:34 PM
Re: Which network processor tests?
We note the sponsorship of the blog for full-disclosure purposes. We also note that Heavy Reading is wholly responsible for the content. The post by Sterling calls attention to the fact that performance claims often do not pass scrutiny, which means operators should do their own evaluations before making purchase decisions.
tb100
50%
50%
tb100,
User Rank: Moderator
9/12/2013 | 6:01:11 PM
Re: Which network processor tests?
Wait a minute. Did you just say, and I quote: "It is Cisco's testing of Alcatel-Lucent's 400G processor"? (and 'blog was commissioned by Cisco Systems')?

So Cisco paid you to write an article that shows the 'results' of their testing of a competitor's product? Aren't the results pretty much predetermined, no matter what the capabilities of the product?

 

I am at a loss for words....
yarn
100%
0%
yarn,
User Rank: Light Sabre
9/12/2013 | 2:06:22 PM
Re: Which network processor tests?
Sterling,

Why lend your name to help Cisco get credibility for this type of quasi-technical FUD bs articles they're so well known for? Now you're caught in the middle and have to stand up for their crap, and tarnish your own credibility in the process.  
mendyk
50%
50%
mendyk,
User Rank: Light Sabre
9/12/2013 | 11:53:53 AM
Re: Which network processor tests?
All product testing -- including tests from independent labs -- should come with a big "grain of salt" warning. And in the case of vendor-run tests, results are clearly subject to question, especially from competing suppliers who had no influence on how the testing was done. From the point of view of prospective customers, though, it's important to understand when performance claims come with an asterisk or two.
sterlingperrin
50%
50%
sterlingperrin,
User Rank: Lightning
9/12/2013 | 11:11:34 AM
Re: Which network processor tests?
Manish,

Correct, we thought there was value in the piece as a call for due diligence.

If service providers are testing these systems at their limits across scenarios applicable to their network (including growth and possible changes to traffic patterns, etc.) then they are covered.

400 Gbit/s is very new. I don't how much testing has been done by anyone yet.

Sterling

 
Page 1 / 2   >   >>
Educational Resources
sponsor supplied content
Educational Resources Archive
More Blogs from Heavy Lifting Analyst Notes
Cable operators have some key hurdles to overcome if they wish to fill the growing mobility needs of small and large businesses alike.
With more and more businesses seeking mobility solutions, cable operators have the chance to carve out an even larger share of the commercial market.
Cable is going big with data analytics, digging into its big data stores, with the hope of garnering information, simplifying operations and improving services and customer care.
A new Heavy Reading report examines why NFV, SDN and cloud are so important to CSPs, and explores their impact on data center infrastructure.
In this first part of a new sponsored blog series, AT&T's Bill Smith explains how AT&T hopes to complement its GPON-powered GigaPower rollout with additional technologies, such as G.fast.
From The Founder
Download our complete guide to de-risking NFV deployment in 2016, including:
  • An eight-step strategy to deploying NFV safely, based on input from the companies that have already started virtualizing their production networks.
  • Interviews with leading executives at Colt, AT&T, Deutsche Telekom, Cisco, Nokia, ZTE, Ericsson and Heavy Reading.
  • Flash Poll
    Live Streaming Video
    Prepping for the Future: Upskill U Explained
    During this short kick-off video, Doug Webster, Vice President of Service Provider Marketing, Cisco, and Light Reading’s CEO & Founder Steve Saunders give an overview of Upskill U.
    Telecom Innovators Video Showcase
    Atrinet's NetACE – Migration to NFV & SDN With NetOps-Driven LSO

    5|4|16   |     |   (0) comments


    At Atrinet's headquarters, Ray Le Maistre sits down with Roy Silon to get an in-depth look into the company's focus and the secret recipe for their success.
    LRTV Huawei Video Resource Center
    Amsterdam ArenA, Powered by Huawei

    5|4|16   |     |   (0) comments


    Huawei's ICT solutions power the state-of-the-art Amsterdam ArenA, turning it into a smart stadium.
    LRTV Interviews
    Testing When There's No 'There' There

    5|4|16   |     |   (0) comments


    The benefits of SDN/NFV are well known, but the transition comes with some challenges, prominent among them is: how do you test a network that has been abstracted and has the potential to be endlessly reconfigurable? Light Reading was at NFV World Congress in Santa Clara, Calif., where we bumped into Mats Nordlund, CEO and co-founder of Netrounds, a Swedish ...
    LRTV Interviews
    Ditching the Slash & the Orchestration Wars

    5|3|16   |     |   (2) comments


    SDN and NFV have been inextricably bound with each other for so long that on a conceptual level, smooshing them together into one catch-all phrase – SDNFV – is now justifiable, according to Dan Pitt, executive director of the Open Networking Foundation (ONF). Light Reading spoke to Pitt at the NFV World Congress, where he explained that the next ...
    LRTV Custom TV
    ZTE TV Connect Highlights

    5|3|16   |     |   (0) comments


    ZTE gives us a tour of its booth and new products at TV Connect in London.
    LRTV Custom TV
    Deluxe's Unified Delivery Solution

    5|3|16   |     |   (0) comments


    Join Alan Breznick of Light Reading and visit the Deluxe booth at NAB! Here you'll find Deluxe's Unified Delivery Solution, OTT video, virtual reality, HDR, 4K and much more!
    LRTV Interviews
    Verizon Puts Gray Boxes in the Shade

    5|2|16   |   04:33   |   (1) comment


    When it comes to the white box trend, "gray" boxes, which have a slight proprietary twist, don't give service providers and end users the advantages they're seeking, according to Verizon's Vice President of Product and New Business Innovation Shawn Hakl.
    LRTV Custom TV
    Dealing With a Disrupted Video Market

    5|2|16   |     |   (0) comments


    Ericsson's Simon Frost discusses how traditional pay-TV providers can cope with the big changes wrought by the rise of OTT video and IP technology.
    LRTV Custom TV
    The VNF Responsibility of Red Hat

    5|2|16   |     |   (0) comments


    At MWC, Caroline Chappell of Heavy Reading visits the Red Hat booth and sits down with Chris Wright to talk about the responsibility the VNF needs to take on in order to ensure the operators get the carrier-grade performance they expect for their network.
    LRTV Interviews
    AT&T Expert on the Key Pillars of UC

    4|29|16   |   03:58   |   (0) comments


    Vishy Gopalakrishnan, AVP of product marketing at AT&T, talks about the three developments that are making unified communications and collaboration secure and reliable for enterprise users.
    LRTV Documentaries
    LRTV Report: Mobile Core Innovation

    4|28|16   |   25:32   |   (0) comments


    Hear from multiple industry experts from Deutsche Telekom, SK Telecom, Heavy Reading, Huawei, Cisco, Ericsson, Nokia, NEC and many more about developments in the mobile core as operators virtualize their IMS and evolved packet core systems and prepare for a 5G world.
    LRTV Huawei Video Resource Center
    NFV World Congress Highlight

    4|26|16   |     |   (0) comments


    The highlight of the NFV World Congress contains exciting telecom news. Join us for an inside look at Huawei's ICT 2020 plan and its latest collaboration with industry leaders.
    Upcoming Live Events
    May 23, 2016, Austin, TX
    May 23, 2016, Austin Convention Center
    May 24-25, 2016, Austin Convention Center, Austin, TX
    September 13-14, 2016, The Curtis Hotel, Denver, CO
    December 6-8, 2016,
    June 16-18, 2017, Austin Convention Center, Austin, TX
    All Upcoming Live Events
    Infographics
    A new survey conducted by Heavy Reading and TM Forum shows that CSPs around the world see the move to digital operations as a necessary part of their overall virtualization strategies.
    Hot Topics
    Ultra-Broadband Summit, Hong Kong
    Iain Morris, News Editor, 4/27/2016
    Amazon AWS Reports $2.6B Quarterly Revenue, Up a Colossal 64%
    Mitch Wagner, West Coast Bureau Chief, Light Reading, 4/28/2016
    WiCipedia: Woman Cards & Bitch Switches
    Sarah Thomas, Director, Women in Comms, 4/29/2016
    Sprint CEO: Our Spectrum Is for 5G
    Dan Jones, Mobile Editor, 5/3/2016
    Amazon & Other 'Big 4' Cloud Providers Crushing Competitors
    Mitch Wagner, West Coast Bureau Chief, Light Reading, 4/29/2016
    Like Us on Facebook
    Twitter Feed
    BETWEEN THE CEOs - Executive Interviews
    In this latest installment of the CEO Chat series, Craig Labovitz, co-founder and CEO of Deepfield, sits down with Light Reading's Steve Saunders in Light Reading's New York City office to discuss how Deepfield fits in with the big data trend and more.
    Grant van Rooyen, president and CEO of Cologix, sits down with Steve Saunders, founder and CEO of Light Reading, in the vendor's New Jersey facility to offer an inside look at the company's success story and discuss the importance of security in the telecom industry.
    Animals with Phones
    Sloth Mail Click Here
    Sloth mail -- somehow even slower than snail mail.
    Live Digital Audio

    Of all the tech companies in the Valley, Intel has made the most aggressive commitment to building a diverse and inclusive workplace culture. It's doing so by taking concrete, measurable steps, making a large financial investment and through a commitment to complete transparency about its progress. In this radio show, WiC Director Sarah Thomas will be joined by Shlomit Weiss, Intel's Vice President, Data Center Group, and General Manager of Networking Engineering, who will share with us why Intel is tackling this huge challenge, how and to what effect. She will also discuss her unique experiences leading development of Client SOC development in the past and today leading development of all of the chipmaker's silicon hardware for networking IPs and discrete devices and managing a team of 600 engineers across Israel, Europe and the US.