Aruba wants you to stop buying the AP134-135. Round 2.
July 17, 2012 10 Comments
Aruba recently posted a rather snarky post about the technological shortsightedness and irrelevance of 802.11ac upgradability of todays wireless infrastructures. This original post (mirrored here) admittedly ruffled my feathers on several fronts so I wrote this response. If you haven’t read these, I encourage you to go do that now.
Aruba product marketing manager, Ozer (@ozwifi) replied to my reply. Before we get to the meat of this post, in the interest of full-disclosure, this post has no direct ties to the Wireless Tech Field day events hosted by Gestalt IT. I have been selected as a delegate for the upcoming Wireless Tech Field Day event that Aruba (among others) has sponsored in the past. As a Tech Field Day delegate I have been given access to hardware and solutions from the event sponsors to utilize as I see fit. At the time of this writing, Aruba is not currently listed as a sponsor of the WFD3 event, but we certainly welcome them and look forward to their involvement!
Ding Ding!
Hey Sam,
It is @ozwifi here. It is not uncommon that we get on each other’s nerves in the Wi-Fi industry and by the tone of your reply I am guessing that’s exactly what we did. But you gotta admit, there are no personal attacks in the blog entry since it is delivering an educated technical opinion.
Oz! Good to hear from you. I apologize for the rather public response to your post, but this seemed the fairest way to address this in its entirety. To the audience at large, I apologize for the broken up, threaded reply and will do my best to make it as cohesive as possible. You are indeed correct that it’s not uncommon to get on each others nerves and you are spot on that this one hit home for me. Perhaps I shouldn’t be so personally vested in industry vision, but I’m sure it’s one of many faults that I have. 🙂 You are correct that there are no personal attacks in the Aruba post and I hope that no one believes that my reply was somehow a personal attack on the Aruba team – infact the only team I mentioned explicitly was the executive team and I certainly don’t hope they *actually* jump off the top of the tallest building in San Jose. That would not be pretty or professional and was merely a ‘leaping’ analogy. Regarding the blog post being an ‘educated technical opinion’, I do take exception to this being an educated technical opinion. It doesn’t sound educated whatsoever and I think that Aruba’s shortsightedness regarding 802.11ac is rampant in the article. Also, I’m still interested in just what the heck a 2400 is…
Poking fun at Aruba’s #1 competitor in the WLAN space with a bit of humour. You have to meet with the author, Scott, during the next WFD – he is not that bad of a person as you might think. So there is really not much to be ashamed of since we are not proposing the kidnapping of new born puppies.
Indeed I look forward to meeting him in person and we look forward to Aruba participating in another lively discussion this year! Also for the record, I wholeheartedly disagree with kidnapping new born puppies.
Before we talk tech – please leave your comments on our website.
I did indeed leave exactly my reply on the Aruba website and as of now, the post has not been approved and is not present in your comments section. To contrast, your post to my replies section was almost immediately approved. I welcome the conversation and look forward to Aruba being more transparent about their comments in the future.
First we do not have many people leaving comments, so we can use some. Second we are not that evil – look at our YouTube channel… anyone can say whatever they want. Unless it is personal attacks of course, cause that’s just not cool.
Alright, let’s talk tech.
Here is where Aruba stands:
1. We believe that dedicated AP hardware is going to provide the best coverage & capacity. Best antenna choices, speeds & feeds optimized for 11ac. If it was such a great thing to install modules on an AP in terms of either of these two, many WLAN vendors including us would have jumped on the bandwagon.
There will always be advances in technology and I believe that most any new solution will ultimately outperform legacy solutions. We see this time and again in the industry and this is a byproduct of Moore’s law. The 802.11ac module is about investment protection. The message from Aruba is clear: either a) don’t buy a 3SS AP today and wait till the 802.11ac AP comes out in the future or b) buy two Access Points (3SS today and 802.11ac tomorrow). Cisco has an option that addresses this concern head on. Aruba does not.
2. Since we are a WLAN company, you will not be too far off in assuming that we will an 11ac AP available down the road. That’s a given. I cannot tell you when, what, how since the info is still very much confidential and shared under NDA.
Of course! This adherence to an NDA is critical in our industry and competitive speculation beyond NDA is what Aruba is good at. This is FUD until you can empirically prove otherwise (more on this later).
3. We are obviously not going to stop promoting AP-130 series product line. We educate our customers regarding the benefits of first gen 11ac and second gen 11ac all day everyday. We do not hide information or try to corner them into buying 130 series. That will be very wrong. Upgrading to dedicated 11ac AP from Aruba 11n will require same process that folks are used to performing during the last 10 years – climb the ladder, plug out AP, plug in AP. As opposed to Cisco, we are not proposing a change in this process. There are no hidden costs here.
I have every expectation that Cisco will not only have a dedicated 1-st gen 802.11ac Access Point in the future, but will also have a 2nd gen and whatever comes after that. The market is always evolving. Cisco’s message today is that the price of two Access Points from Aruba is more than the 3600 + a 1st gen 802.11ac module. Again, investment protection. The costs that Cisco is addressing with this module are not hidden. They are outright and Cisco is head-on tackling this proactively. Aruba is behind the 8-ball and does not offer investment protection. If I were an Aruba customer, I’d not buy new Access Points today because there is no low-cost upgrade path to 802.11ac in the future. Either that or write your check out to ‘Aruba Catalog of Compromise’. ‘Aruba Catalog of Shortsightedness’? ‘Aruba Catalog of Technical Irrelevance’? ‘Aruba Catalog of FUD’? I don’t know – pick one, they all work for me.
Here are my comments on your responses for what they are worth. I am guessing that we will agree to disagree at the end of it… although I hope I can provide more color commentary and that you will find them useful. Again, I am trying to talk tech here not disagreeing with the fact that 3600 11ac module is good marketing.
Oz, I 100% agree with everything you said here and am speechless that we’re so in sync! 🙂
1250 series: Folks invested in the platform found out later that there was no need for this modular AP since moving from draft 2.0 of the standard to the ratified version did not require an hardware upgrade.
We see this time and again with the Cisco product lineup. The radio modularity in the 1220s was upgrade investment protection for 802.11G. The radio modularity in the 1252s was upgrade investment protection for 802.11n. The radio modularity in the 3600 is upgrade investment protection for 802.11ac. There is a trend here.
Cisco’s predictions were wrong.
No, infact Cisco’s predictions were right! They took a ‘best guess’ at the hardware that it would take to support the finally ratified specification and there was never a module released because it was never needed. No hardware changes required was a win-win for Cisco customers.
It was a 5-pound AP
Auxiliary boat anchor, yes. It was heavy. Don’t beat up on it because it was big-boned. It needed that modularity. It’s mommy told it so.
with no dual-radio support 802.3af (if you rememeber, Cisco was claiming at the time that 11n APs will not be able to support 802.3af).
Unfortunately, you’re wrong here. The 1252 does indeed support 802.11n on both radios utilizing 802.3af. Quit spreading flat out lies.
I believe that 1250 series was mostly about marketing, capturing attention and not so much about delivering best of breed Wi-Fi technology. Given that the product line lived only about a year, on this side of the fence we think that our predictions about those first generation of 11n APs were the right ones.
1 year, huh? I show final date of support for the 1252 as early 2017. My memory isn’t all that clear on the 1252 launch date, but it was first supported in WLC code 4.2.61.0 which has a release date of March 21, 2011. My math is a bit fuzzy on this one, but 2011 to 2017 seems a much larger window than 1 year.
Difficult to deploy: Here is the Cisco process… Install 3600 today. Wait 8 months. Buy 11ac modules. Climb up the ladder. Unscrew the mounting bracket. Take the AP down. Install module. Climb up the ladder. Screw back the mounting bracket.
The vast majority of the installations I see are ‘snap in’ mount. I don’t recall how the Aruba 130 mount bracket works, but palming the butt of an AP to snap it out of place and snapping a module in seems pretty straightforward to me.
Cisco *will* come up with their dedicated 11ac AP hardware that’s based on Marvell chipset, as opposed Broadcom running inside the 11ac module for the 3600.
I do not have technical documentation about the chipset in the 802.11ac module from Cisco. This would be the first time Cisco has used Broadcom in an infrastructure device and would certainly be a departure from their M.O. Having said that, if you have NDA insight into the hardware diagram and working structure of the AP, I believe this would be covered by NDA and subject to change. Either way, you’re speculating or sharing data that is NDA and is subject to change. We’ll have to agree to disagree until the module comes out and we can take it apart and do performance testing with it.
With that upgrade, that’s three trips to the ceiling. And when the 2nd gen 11ac AP comes out, you do it again. That’s four. We cannot call this simple as opposed to difficult.
I still have 1252s in place today. They service a need for many of my customers that simply need to support 802.11n. I foresee that the 802.11ac module will support 1st gen 802.11ac needs for a long time. Aruba has no products today that can be purchased and upgraded later. Again, upgrade investment protection.
CPU speeds: Here is the thought process. Aruba AP-135 beats Cisco 3600 in peak performance. Whether it is pure 3×3:3 MIMO laptops, UDP or TCP traffic flows, or a mix of smartphones, tablets, laptops… that’s what we see using Cisco release 7.2 and Aruba release 6.1.3.2. Aruba product managers prefer not to use AP-135 CPU and memory subsystems for an 11ac AP per our interviews in order to be able to deliver the best peak 11ac performance. This tells me that Cisco product managers have to think the same way since AP-135 outperforms Cisco 3600. Using your argument, although looking at it from a different angle, how can we be sure that Cisco 3600 plus an 11ac module will deliver greater performance than a dedicated 11ac AP hardware?
We can’t until it’s out and available. Regarding your other performance claims, I welcome those head-on and would encourage readers to visit ciscobeatsarubayetagain.com. Aruba has addressed these performance tests inconclusively (performing iPad throughput tests with Android devices, transcoding their video down to lower bit rates, and disabling recommended enterprise feature sets such as spectrum analysis and IDS). When will we see Aruba engage a 3rd party like Miercom to do independently validated performance tests instead of continuing to poke and prod at Cisco? Let’s back your claims up independently. As an aside, I welcome the performance claims of existing hardware but it’s off-topic for this thread.
Inconsistent RF and feature set: 3600 will run two separate Wi-Fi chipsets from two different vendors: Broadcom and Marvell. Why on earth would I want to do this if I want uniform features and functionality across my 2.4GHz and 5GHz radios? No AP that was built for enterprise WLANs ever had this design. I am sure there was a good reason behind it.
Adressed above.
Upgrades: Cisco 3600 requires 7.2 release, which requires latest generation of Cisco controllers and NCS management instead of WCS management. We are just making it more apparent for those who care, although Cisco release notes clearly state these facts as well. The tradition of having to upgrade something in your network whenever there is a new WLAN product or solution from Cisco is really what gets on our nerves. For instance ISE… BYOD solution that requires me to upgrade from ISR to ISR G2… why would I want to touch my branch router if there is an employee owned iPad connecting to my network? Some of this stuff just does not make sense to us and we have just watched this episode way too many times … hence it is a reflex motion… we do not miss an opportunity to remind folks of what they need to be careful about.
I’d like to hear more about your ISR concerns. I’m not sure where the mindset of routers being upgraded to support your iPad comes from. The iPad is not a wired device. Are you referring to the AP801/802 module? Both of those are integrated into the ISR and fully supported in 7.2 code. If you have a switch that supports ISE, there is no need to replace the router between the switch and the Access Point. Although, I always liked the idea of cabling my iPad to my ISR router…
Alright my apologies for the long comment post, tried to do my best to keep it short. I hope you can give me a chance to respond by accepting my comments.
Your comments are always welcome (despite being shunned on the Aruba post comments) and I apologize again for the threaded response. If you’ve read this far, I formally invite Oz (and Scott for that matter) to come onto the No Strings Attached Show and discuss Arubas stance on 802.11ac. I look forward with taking more about this in a forum more conducive to back and forth dialogue.
See you at WFD3.
I as well as the entire WFD3 delegate team most certainly look forward to Arubas participation. I recall last year being lively and look forward to it!
-Sam
SC-WiFi 
I have a blog post up that explains how well tr Aruba mounts work from #wfd2.
http://BlakeKrone.com/2012/02/17/hanging-an-ap-the-good-the-bad-the-ugly
You might want to check your dates on the 1252. I don’t know the exact dates, but Amazon shows the 1252 was first available there December 2007, which is far closer than the 2011 date you listed. Also, technically the last date of sale was 1/30/2012 http://www.cisco.com/en/US/prod/collateral/wireless/ps5678/ps6973/end_of_life_notice_c51-681596.html, but it has been replaced with another 1200 (1262) since the modular design no longer was needed after the standard was finalized. Clearly the Aruba 1 year statement is a flat out lie.
I double checked my dates on the 1250 as requested. The current release notes for WLC code at: http://www.cisco.com/en/US/docs/wireless/controller/release/notes/crn7_0_235_0.html
show that the 1250 was first supported in software release 4.2.61.0. The ‘big list of code’ at:
http://www.cisco.com/en/US/products/ps6366/prod_release_notes_list.html
shows that the release date for 4.2.61.0 as being 21/Mar/2011. The last date of Hardware support at the link you posted is Jan 31, 2017. I suppose it depends on where you draw your lines at and what you declare as ‘launch date’ but I admitted that my math may be slightly off. These two links and the EOS notice you posted are where I got my dates. Regardless, no matter which way you slice it, 1 year is not accurate.
Thanks! -Sam
If you read inside those release notes, it shows a date of Oct 26, 2007, which is consistent with the fact that I personally worked with some in 2008. By the way, this is a very interesting dialog. I look forward to hearing more responses from Aruba to see if they can substantiate any of their claims, and how they explain the incorrect facts you have called them out on.
Sure enough! Thanks for the reiteration. I thought they were much older than that, but didn’t dig enough. 🙂 Now the stretch is 2007 to 2017. About 10 years seems like a pretty good run for a ‘failed Access Point’…
-Sam
FYI – The 1252 was not dual-radio-N PoE capable until subsequent code releases, and only supported up to MCS 7 data rates. But it did support 17dB across all channels, which was arguably better than the 1140; although the 1140 supports up to MCS 15, the transmit power is reduced as the MCS rates go up.
One thing for sure, Cisco is great at pumping out hardware. The 1140 and 3500 are already in their rear-view mirror. They needed to offer investment protection, as customers are probably suffering from product fatigue.
Thanks for the so ewhat heated, but still enlightening discussion. Everyone benefits from factual and philosophical dialogue.
Alright I am back for round 2… I hope this does not last until round 15 🙂 I gotta tell you I love the “ding-ding” opening! I am glad that we can keep the discussion fun, engaging instead of using anger and personal attacks… Thanks again for accepting my reply, glad to have the discussion going. BTW, you type fast!
Your comment to Aruba blog…
I am assuming it is a side effect of web changes yesterday (new navigation and converging 3 blog pages into 1) but I will check shortly.
Regarding 2400…
small typo as you can guess: meant to refer to 2500 series controllers.
Alright back to tech…
Regarding 1250 series AP (since many commented on it)…
Almost a year after 1250 series, 1140 series was announced. I am not claiming that the AP actually physically failed (it obviously worked just fine after you managed to install it) – it was no longer the right AP to install for many, unless you are installing APs in a warehouse or similar challenging environments. Cisco’s promise of “modular AP is the way to go” was no longer. 1140 had better form factor, better price, did not need external antennas, better PoE efficiency. There was almost no reason to install 1250 series in a classroom or a carpeted office space after 1140 series was released. During that timeframe Aruba’s AP-124/125 series won many deals against Cisco 1250 series (support for PoE and better form-factor were big technical reasons) when we get the chance to sit at the table. Market demanded something better than 1250 series.
Of course, there is a trend with Cisco’s modular APs – great marketing for Cisco, brings in more dollars. I am just not convinced that it is the right thing for the customer. My humble opinion…
Let’s double click on Cisco’s investment protection….
Note that 1st gen 11ac AP does not go above 3 spatial streams (instead of up to 8 defined per 11ac standard) and does not support multi-user MIMO (which is really beneficial for the upcoming 11ac capable smartphones and tablets as you know). My guess is 2nd gen 11ac APs will have up to max of 5 spatial stream support… since putting 8 antennas in an AP may not be that great of an idea since folks want APs that can be carried by hand… alright let’s go through couple of investment scenarios.
Case#1: I am a Cisco 11n customer who purchased 1140 series as a first move to 11n – $1000. I do not have awareness against noise sources yet (I think this changed for 1140 series in the latest Cisco release 7.2 but not 100% sure) and no spectrum analysis (CleanAir). Then I move to Cisco 3500 series to enable CleanAir and noise awareness in RRM. Upgrade to 3500 series after 2-3 years on 1140 series costs $1300. Now I have a choice… do I go to first gen or second gen 11ac. Assume customer says 2nd gen 11ac and assume that the AP price is $1000…. Total is $3300 while I go from 1140 series to 2nd gen 11ac. During that same period, an Aruba customer invests in the equivalent AP-105 series – $700. Noise awareness is available but no spectrum analysis. With a software upgrade on my controllers and $80 per AP license (unless I already have the WIP license) I enable spectrum analysis. Assume this customer follows the same track and spends the same amount for a 2nd gen Aruba 11ac AP. Total is $1780. That’s a big difference.
Case#2: Cisco customer, following same path but does not move from 1140 series to 3500 series but to Cisco 3600 series at $1500. Then to first gen 11ac with the $500 module. So the total here is $3000. Aruba customer follows the similar path – moves from AP-105 to AP-135. That’s $2000 at that point. If Aruba 1st gen 11ac AP price is less than $1000, Aruba pricing is better. If not Cisco pricing is better. Will Aruba 1st gen 11ac AP price will be too far off from $1000? sorry can’t tell 🙂
Case#3: I am a new Cisco customer (no existing Cisco WLAN) who wants to purchase 3 spatial stream 11n AP. Cisco 3600 is available at $1500. After I review what’s available with first gen 11ac and 2nd gen 11ac, I decide to wait another 2-3 years before I move to 2nd gen 11ac. I upgrade to dedicated 2nd gen 11ac AP, say at $1000. Total is $2500. On the other side, assume I am a new Aruba customer who follow the same path, assuming same 2nd gen 11ac AP pricing. This new Aruba customer invests in AP-135 at $1300 at first and then to 2nd gen 11ac AP. Total is $2300. Let’s call it even.
Case#4: A new Cisco customer invests in 3600 series and first gen 11ac module sounds like a great idea. And he does not believe all the Aruba noise about what a bad idea modular AP design is 😉 … and ready to climb back up to the ceiling a second time in the same year… total to move to 1st gen 11ac AP is $2000. Similarly, a new Aruba customer is thinking the same way… ready for 1st gen 11ac AP, does not want to wait until 2nd gen 11ac AP, does not mind climbing back up to the ceiling. Aruba AP-135 and then to first gen 11ac AP investment is $1300 + price of 1st gen 11ac AP. Cisco wins the list price argument here since we can assume that Aruba’s 1st gen 11ac AP will have a greater price tag than it’s 2×2:2 11n AP-105 (which is $700).
Ok I think I just got the cross-eye that Scott was talking about in his blog… without having to use the OptiGrab! So investment protection argument by Cisco applies to the last case listed above. My educated guess is we will see more of #1, #2, #3 than #4. Again that’s my opinion… agree to disagree.
BTW, Cisco’s 2nd 4×4:3 11n AP (this time non-modular) is about to be released it looks like. Our periodic FCC searches returned the following: Cisco 2600 series (http://fcc.us/OJHJxr). Of course don’t know the pricing for it… but… if it is around $1000, then I am actually paying $1000 extra (not $500) to be able to use Cisco’s 1st gen 11ac module: Extra $500 for Cisco 3600 instead of buying Cisco 2600 series and another $500 for the module. It will be interesting to see what the list price will be for the 2600 series.
Aruba performance tests…
We do not have Android tablets to replace iPads – no reason to – we have 100+ iPads in the TME labs. No change in video resolution for Aruba WLAN compared to Cisco WLAN – it is the same exact infrastructure, testbed. Again no reason to. Enabling and disabling RF scanning, IDS, spectrum/CleanAir does not make any difference for either vendors. Aruba TMEs ran those tests for weeks. We should talk about “maximizing airtime” in another opportunity – Aruba’s RF engineering focuses on this topic nowadays than ever. For instance, a test for you to consider running on Cisco WLAN… start with 5 smartphones on 11n 2.4GHz radio. Record TCP download throughput. Repeat with 10, 15, 20 smartphones. Then add TCP upload traffic into the mix and record total throughput. Results are interesting.
Miercom = independent… really? Cisco TMEs run these tests in their labs, publish it on the website URL that you shared and it just happens that a separate set of engineers who work for Miercom happened to run the same set of tests – not less or more – and come up with exactly the same set of test results. Independently. Without being paid any consulting fees by Cisco. Really? 🙂 I firmly believe that something like Network World Clear Choice test reports are independent – and I cannot see how Miercom follows the same model.
NSA podcast show is a great idea! Let’s do it. I will email Blake.
ps. Happy to chat about ISRs and ISE more down the road!
If the FCC link I shared does not work… try this:
1. Go To https://fjallfoss.fcc.gov/oetcf/eas/reports/GenericSearch.cfm
2. Add 01/01/2012 to 07/19/2012 for date range.
3. Type Cisco for applicant name.
Sam, I’d like to say that I have thoroughly enjoyed reading your blog and especially this article in regards to the low marketing tactics being used with what would seem absolutely no factual information.
It’s interesting that if you look at that the FCC link Ozer Dondurmacioglu posted and dig down and download the manual for the 2600 it would seem that this is going to be a lower price AP (maybe something aimed at the lower end 600/1040 market)
Looking at the specs: (I have summarised the main features to keep it short-er)
The 2600 series supports high-performing Spectrum Intelligence which sustains three spatial stream rates over a deployable distance with high reliability when serving clients. The 2600 series provides high reliability and overall wireless performance.
The features of the 2600 series are:
– CleanAir—Automatic detection, classification, location and mitigation of RF interference
– ClientLink+—BeamForming to 802.11n clients as well as legacy 802.11a/g OFDM clients
– Throughput, forwarding, and filtering performance scaled to meet 3 spatial stream 450 Mbps data-rates
• 2.4 GHz and 5 GHz 802.11n radios with the following features:
– 4TX x 4RX
– 3-spatial streams, 450 Mbps PHY rate
– Spectrum intelligence
– DPD (Digital Pre-Distortion) technology
– Cisco Vector Beamforming—Implicit Co-phase beamforming for .11ag clients and 1×1 11n
So it would seem that this is going to be a low-end/cheaper version of the 3600 with no upgrade path to 802.11ac (for those that do not have a need for it or can’t justify the cost for a large scale deployment) and not a replacement to the 3600 as Ozer speculates.
It’s also interesting that if you do the same search on the FCC for Aruba it shows no new products submitted for certification, this would seem to indicate that Aruba may be some time away from releasing anything new and hence the original blog post (sounds like Aruba can’t keep up so are trying to slow down the market instead, but I’m just speculating:-))
The realistic fact is that most enterprise customers are at least 1 – 2 years before having a large enough deployments of 802.11ac compatible client devices, that the customers with 1140 series AP’s in place won’t warrant upgraded them for another 1-2 years, and if you have just deployed the 3600 then you still have at least 4-5 years before need to upgraded especially if you have added the upgraded module.
I would also add to this that there will probably be 1st gen and 2nd gen of 802.11ac clients so this again prolongs the upgrade needs. Not to mention the lacking number of smartphones and tablet devices that aren’t using 802.11a chipsets in their devices, therefore not support 802.1ac at all.
There are so many other comments I could add particularly around the 1250 references and the “Difficult to deploy” statement but this is long enough already.
While entertaining, I hope this back and forth finds its climax soon. What I find interesting are the assumptions (on both sides of the argument) that customers will immediately upgrade to the latest and greatest across their whole deployment. Unless they are relatively small, customers (arguably the ones to be most affected by large scale investment protection) are highly unlikely to flip their entire AP deployment. What is much more likely is that once the customer decides to go towards the next technology (11ac in this case), purchases from there will be the new model. Existing “old” models are still functional and likely will continue in place for several years to come.
Unless they’re bleeding cash, I don’t think you’ll find folks making an 11n investment and then tearing that out for an 11ac investment once available. I also agree with Troy in that it’ll take some time for 11ac clients to gain momentum in heterogenous device populations…. all the more reason that current (non-11ac) AP investments will inherently be protected.