06-29-2012 01:35 PM - edited 06-29-2012 03:04 PM
Today I am putting the issue of Cisco’s superior AP technology (ClientLink 2.0, 4x4 MIM) and video density performance vs. Aruba to rest once and for all. Unlike Cisco, we have not paid for an “independent 3rd party” to write up a report in which no configuration or setup information is shared. Instead, we have provided all test setup, methodologies, and configurations for both vendors. We have also made every attempt to optimize Cisco per their latest documentation and best practices and provide an apples-to-apples comparison.
Aruba’s Technical Marketing team has completed a set of repeatable tests comparing Aruba’s AP-135 AP running 220.127.116.11 software to Cisco’s 3600 AP running 7.2-103 software. We have taken each of the Miercom tests in sequence and have refuted each one in turn.
First we take a close look at “rate vs. range” performance using Ixia Chariot and comparing Aruba to Cisco at several locations, and we demonstrate that ClientLink 2.0 and 4x4 MIMO are still inferior to Aruba’s built-in performance and ARM benefits:
- Aruba outperforms Cisco at three non-line of sight (NLOS) locations for both TCP download and TCP upload performance by 11-13% at 30’
- At 120’, a MacBook Pro cannot connect to the Cisco network but is able to maintain a decent throughput with the Aruba AP
- Aruba performs 50% better than Cisco for an iPad tablet that is rotated to a vertical orientation
- At some key testing locations, Cisco performs better with ClientLink 2.0 off than on
Next we tackle video density, and there are essentially three main tests: HD multicast streaming using VLC media server and a mix of laptops running 2 or 5 Mbps streams, a high density of iPads running TCP unicast Air Video, and a high-performance density mix of clients and applications.
For the multicast streaming test, we leveraged Aruba’s distributed dynamic multicast optimization (D-DMO) feature to provide the highest video scale in the industry today. Aruba accommodated 16% more 2 Mbps videos than Cisco (51 vs. 44), and 25% more 5 Mbps videos than Cisco (39 vs. 31) with good quality. The iPad test included one iPad which was mirroring to an Apple TV, and other iPads were added until poor quality (buffering and freezes) was observed. Aruba accommodated 31% more iPads streaming 1 Mbps video than Cisco (21 vs. 16).
The high density mixed application test was a bit more complex, as there were different client types (3x3, 2x2, 1x1, Intel, Broadcom, Apple TV) and applications (multicast video, unicast video, AirPlay, large data file transfers) all operating simultaneously in a highly dense scenario. Aruba was able to accommodate 37 clients in this heterogeneous mix, and the following comparison was made with Cisco:
In the Aruba case, high quality video (both 2 and 5 Mbps video) was observed without any artifacts, including pixilation or video freezes. The iPad videos did not exhibit any buffering or video quality issues. The Apple TV stream was low-latency, reflecting what was seen on both the iPad and the projected screen simultaneously. The 11 GB FTP file download never timed out.
For Cisco, the HD video (both 2 and 5 Mbps) experienced a significant amount of pixilation artifacts and video freezes across all laptops. There was noticeable buffering on the iPads playing TCP video. The Apple TV had difficulty mirroring, and there was significant latency between the iPad and the projected monitor. Finally, the 11 GB FTP file download timed out in the middle of the test.
The key here is that Cisco is not optimized for BYOD high-performance, application delivery, does not have an integrated stateful firewall, and does not outperform Aruba in any test.
Finally, we ran a large file transfer to a tablet and showed comparative battery drain performance. We did not see any significant differences in battery life at the end of the test (6% drains for both Aruba and Cisco).
Did we do this in an Aruba location? Yes. The difference is that Aruba stands behind the test data we have provided. A link to the test report with all configuration changes for both vendors is provided here for you to test yourself.
And, as we like to do at Aruba, we provided a video to keep you all entertained during the test!
Check it out at http://youtu.be/cWm3_y1TPKM
06-29-2012 02:23 PM - edited 06-29-2012 02:24 PM
Thanks for the great level of detail you've shared with us. Nevertheless, I think Cisco's WLAN Controller configuration would be better understood with a "show run-config commands". If you have it I think some of us would really appreciate if you could share it with us.
ACMP, ACCP, ACDX#100
If I answerd your question, please click on "Accept as Solution".
If you find this post useful, give me kudos for it ;)
06-29-2012 03:13 PM
Sure Samuel, I've asked one of our TMEs to post this too. Please standby.
06-29-2012 03:28 PM
Cisco controller configuration is attached. Feel free to send an email to email@example.com in case you have any questions.
06-29-2012 05:45 PM - edited 06-29-2012 05:47 PM
Sorry - looks like we are having problems posting this .txt file. Please email me directly and I will send you the output of the show running config command at firstname.lastname@example.org
06-30-2012 11:30 AM
Looks like there is a problem with attachments. Just to give you a brief overview of the configurations used on the Cisco controller:
- Tests were run on a clean 5 GHz channel for both Aruba and Cisco. The same channel was used - verified using a spectrum analyzer.
- Both the vendor's AP's were cieling mounted at the same location for the respective test cases.
- The AP's were operating at similar power levels.
- The same client types, client location was used.
- The same video stream with a bitrate of 5 Mbps was used.
- On the Cisco controller: WMM was enabled, QoS profile "Gold" was configured for prioritizing video traffic, the appropriate dot1p value was configured on the QoS profle, a medial stream with the right multicast group address was configured with the recommended RRC values, multicast was enabled globally, IGMP snooping was enabled, video CAC was set to allow max RF bandwidth for video traffic, multicast direct and unicast video redirect was enabled for the multicast to unicast conversion to happen, AP multicast mode was configured.
- While the test was running - we verified if the packets are getting prioritized, if they are getting converted to unicast and are sent out using high unicast rates using ominpeek. We also were monitoring the client stats on the Cisco controller to see how things looked.
Let us know if you have further questions.