& cplSiteName &

Validating Cisco's NFV Infrastructure Pt. 1

Light Reading
Prime Reading
Light Reading
10/12/2015
100%
0%

Virtual switch FIB scalability

    Evaluation overview: In a pure virtual switching scenario, VPP showed no throughput degradation when forwarding to 2,000 IPv4 or Ethernet MAC addresses at 20 Gbit/s, less than 1% reduced throughput towards 20,000 MAC addresses and 23% reduced throughput when forwarding to 20,000 IP addresses. OVS IPv4 and Ethernet forwarding was reduced by 81% when forwarding to 2,000 IPv4 addresses.

Both previous scenarios focused exclusively on Ethernet layer throughput with a small number of flows because they both involved virtual machines. In the third scenario, we evaluated pure virtual switching without any actual application. This is, of course not a realistic application setup: instead it is a reference test of vSwitch properties that need to be determined independently of VNF performance.

One of the most basic and important scalability figures of an Ethernet switch is its ability to handle many Ethernet flows between different endpoints (associated with MAC addresses) in parallel. In a data center, there is usually much more East-West traffic between servers and services directly connected on the Ethernet segment than there is North-South traffic. vSwitches need to enable virtual services to participate in data center communication and need to be able to connect to many Ethernet destinations in parallel.

Separately, vSwitches obviously need to support many IP addresses in their forwarding information base table (FIB) simultaneously, when configured for IP forwarding. If traffic is routed towards a virtual firewall or a virtualized packet filter, there are usually tens of thousands of flows involved from thousands of IP addresses.

We verified forwarding performance of the standalone virtual switch with multiple layer 2/layer 3 FIB table sizes.

VPP showed its strengths based on the optimized, vector-based handling of large tables. It achieved very consistent IPv4 forwarding: The throughput was not dropped at all when forwarding to 2,000 IPv4 addresses compared to a single address scenario; throughput dropped only by 23 % when forwarding to 20,000 IPv4 addresses. The average forwarding latency was largely unaffected by the larger tables, and the maximum IPv4 forwarding latency was still bearable at 400 microseconds for 2,000 address entries and 1,200 microseconds for 20,000 entries.

In contrast, Open vSwitch seems to use less optimized FIB lookups. Throughput dropped from 20 Gbit/s (IPv4) and 8.8 Gbit/s (Ethernet) for the single-entry case down to around 4 Gbit/s in both cases for 2,000 IPv4 and MAC addresses. For 20,000 FIB entries the OVS-DPDK implementation was not usable in its current version as the throughput dropped to almost zero and maximum latency skyrocketed to 37 milliseconds (not microseconds!). We are not complaining -- after all, it's free software -- and in fact we hope that Cisco will contribute its VPP improvements back to OVS to improve the open source software.

These results indicate much better VPP performance in higher scale network deployments.

vSwitch-only IP forwarding performance.
vSwitch-only IP forwarding performance.

vSwitch-only Ethernet forwarding performance.
vSwitch-only Ethernet forwarding performance.

vSwitch-only latency performance.
vSwitch-only latency performance.

Next page: Evaluating the Cisco IOS XRv 9000

 
(3)  | 
Comment  | 
Print  | 
Newest First  |  Oldest First  |  Threaded View        ADD A COMMENT
Ray@LR
50%
50%
Ray@LR,
User Rank: Blogger
10/22/2015 | 12:03:34 PM
Re: Where is Part2
Part 2 is imminent.... a minor clarification neede dto be verified.

Thanks!
NPUArchi20846
100%
0%
NPUArchi20846,
User Rank: Light Beer
10/20/2015 | 5:25:21 PM
Where is Part2
"Part 2, which will be published on Monday October 19"
ashvin213
100%
0%
ashvin213,
User Rank: Light Beer
10/13/2015 | 12:12:27 AM
Some inconsistencies
Hi there,

 

I found some issues that I did not follow:

1.      Throughput in PPS and Gbps don't seem to match. For example, I do not understand how 2.2e6 fps translates to 7Gbps for large frames (it should be around 26Gbps).

2.      Even for small packets, the latencies are in us. Shouldn't they be in ns? The authors claim 20us is in line with ASIC based switches (which I don't think it is. Most hardware switches take a few ns to switch packets. For example, Trident-1 took around 600ns for L3 forwarding and 400ns for L2 forwarding. This was way back in 2009).

3. Adding more MAC addresses must simply increase the memory size. Given that all of these are exact-match tables, I fail to understand

Ashvin
Featured Video
From The Founder
Light Reading is spending much of this year digging into the details of how automation technology will impact the comms market, but let's take a moment to also look at how automation is set to overturn the current world order by the middle of the century.
Flash Poll
Upcoming Live Events
November 1, 2017, The Royal Garden Hotel
November 1, 2017, The Montcalm Marble Arch
November 2, 2017, 8 Northumberland Avenue, London, UK
November 2, 2017, 8 Northumberland Avenue – London
November 10, 2017, The Westin Times Square, New York, NY
November 16, 2017, ExCel Centre, London
November 30, 2017, The Westin Times Square
May 14-17, 2018, Austin Convention Center
All Upcoming Live Events
Infographics
With the mobile ecosystem becoming increasingly vulnerable to security threats, AdaptiveMobile has laid out some of the key considerations for the wireless community.
Hot Topics
Muni Policies Stymie Edge Computing
Carol Wilson, Editor-at-large, 10/17/2017
'Brutal' Automation & the Looming Workforce Cull
Iain Morris, News Editor, 10/18/2017
Worried About Bandwidth for 4K? Here Comes 8K!
Aditya Kishore, Practice Leader, Video Transformation, Telco Transformation, 10/17/2017
What Does Automation Mean to You?
Ray Le Maistre, International Group Editor, 10/17/2017
Animals with Phones
Selfie Game Strong Click Here
Latest Comment
Live Digital Audio

Understanding the full experience of women in technology requires starting at the collegiate level (or sooner) and studying the technologies women are involved with, company cultures they're part of and personal experiences of individuals.

During this WiC radio show, we will talk with Nicole Engelbert, the director of Research & Analysis for Ovum Technology and a 23-year telecom industry veteran, about her experiences and perspectives on women in tech. Engelbert covers infrastructure, applications and industries for Ovum, but she is also involved in the research firm's higher education team and has helped colleges and universities globally leverage technology as a strategy for improving recruitment, retention and graduation performance.

She will share her unique insight into the collegiate level, where women pursuing engineering and STEM-related degrees is dwindling. Engelbert will also reveal new, original Ovum research on the topics of artificial intelligence, the Internet of Things, security and augmented reality, as well as discuss what each of those technologies might mean for women in our field. As always, we'll also leave plenty of time to answer all your questions live on the air and chat board.

Like Us on Facebook
Twitter Feed
Partner Perspectives - content from our sponsors
The Mobile Broadband Road Ahead
By Kevin Taylor, for Huawei
All Partner Perspectives