Tuesday, March 22, 2011

Wireless Tech Field Day - HP Review

The Wireless Tech Dield Day delegates working in the HP EBC conference room
Day 2 started out at the HP Executive Briefing Center (EBC) for an overview by Rich, Chris, and Andres of their wireless architecture from the 2008 acquisition of Colubris. HP has been playing the shell game over the last few years trying to figure out their wireless networking strategy to compete in this market, as they previously did not offer an in-house solution but opted to OEM Symbol gear until their acquisition by Motorola.

HP Wireless Architecture
The HP / Colubris architecture is built around the now-common wireless LAN controller and "smart" APs, calling the collective solution an "optimized wireless LAN". I find this to be a case of marketing hubris, as the pushing control and data planes to the controller can hardly be called optimized. This architecture was pioneered 5-7 years ago by Airespace to solve serious challenges around system-wide wireless network coordination (RRM, QoS, security, roaming, etc.) which was not prohibitively expensive with AP hardware and silicon of the time.

Unfortunately, HP is late to the game and trying to play catch-up. They are just now offerring solutions that  competitors have had for at-least 4 years or more. What's worse is the Colubris acquisition appears to making their job harder because they are trying to re-design an existing controller architecture into a modern distributed architecture which is proving difficult (as it is for other vendors as well). HP may have been better served to design a solution internally from the ground up, which was fully distributed from the beginning. As it stands, the HP solution is typical of most controller solutions with central traffic forwarding, optional distributed traffic forwarding directly from the AP, and serious feature limitations when APs process traffic locally.

In addition, I have serious concerns of the solution's high availability capabilities. A license upgrade is required to move from an "Access" controller to a "Mobility" controller, which enables layer 3 roaming and controller teaming for redundancy. Isn't that the point of a wireless network, to be mobile and highly available?! Additionally, the management platform seems to be an afterthought. Up to 5 controllers teamed together can be managed using the RF Manager software on the controllers themselves. For greater scalability the HP ProCurve Manager must be used, but it doesn't support all system features (such as WIPS/WIDS reporting). The graphical interface also leaves much to be desired, looking like it was designed in the mid-90's and does not appear to have been updated using modern HTML standards. I would expect frustration and a maddening workflow for system configuration.

Access Point Announcement
HP's recent announcement of the 3-stream MSM-460 and 466 APs looks promising, but the devil is in the details. The APs use a 3x3:3 configuration, and the 3rd spatial stream will likely be limited in range, especially given the loss of extra radio chain which prevents Maximal Ratio Combining (MRC). MRC is quite possibly the second most important advancement in 802.11n behind MIMO. Also, the MSM-466 is the only model which supports dual-concurrent 5GHz operation on both radios. Running both radios of an AP in the same band is tricky because of severe adjacent channel interference caused by RF side-lobes of each radio when in such close proximity. The radio antennas either need to be shielded properly from one another (which they are not in the MSM-460), or spaced a minimum of 4 feet apart (which is possible with external antennas on the MSM-466).

The OFDM spectral mask shows the relative power reduction at distances from the center channel frequency (in dB).
AP radios located too close will cause severe interference with each other without sufficient Free Space Path Loss.
Performance Demo
Andres, the self-proclaimed "lab rat", provided a live iperf throughput demonstration. The performance with distributed traffic forwarding and 3 spatial streams was comparatively high, at 231 Mbps downstream without competing traffic and clients at close range. I would have liked to see performance at greater ranges and under varying loads, but the time was not available for an in-depth demonstration. A second test resulted in 191 Mbps downstream with a simultaneous single video stream running to the client (I can't remember the video resolution).

The Future of Wireless
The HP visit wrapped up with a lengthy, detailed discussion on the future of wireless technologies. The delegates focused conversation around the changing requirements for mobile device security of both corporate-liable and personal-liable devices. Emphasis was placed on the need for greater context awareness of both users and devices not possible today, and dynamic policy enforcement at the edge of the network (not upstream, disconnected from device network access). Additional talking points included  a discussion of the need for layer 1 physical RF assurance and stability. This is a pain point for most wireless admins today, as the variance of device performance is directly attributable to vendor implementation variances. Greg (@etherealmind) mentioned the poor transparency in the IEEE standards process, causing vendor greed in delaying or stalling standards process.

Revolution or Evolution? - Andrew's Take
HP's wireless portfolio appears to have been a quick fix for a missing product in their solution catalog. They are just now releasing products that can compete with competitor's solutions from 3 years ago, and are struggling to maintain feature parity. This will continue as competitors migrate to distributed architectures. The product mainly appeals to existing HP customers who prefer a solution from a current vendor, and this product fills that gap. Unfortunately, I doubt HP is winning many customer evaluations based on the larger market.


Other posts you might like:


  1. Hi Andrew,

    I was looking for reviews of the HP wireless systems and found your homepage.
    Within our company HP one a public submission and now we are looking for the right wireless components.

    As you mentioned, HP bought Colubris some time ago, but last year HP also Bought H3C/3Com.
    So we now have two different wireless series within HP.

    In your article you only mentioned the Colubris ones. So, does this mean, that the H3C series will be dropped? Was H3C/3Com ever mentioned during your discussion?

    For me it´s a question about what do buy:
    - MSM460/MSM466 plus E-MSM765zl controllers
    - WA2620E plus A-WX5004 controllers


  2. Hi Michael,
    During Tech Field Day only the Colubris line was discussed and demonstrated. I am not positive what will become of the 3Com line of gear.

    I recommend getting clarification from your HP account team or SE.