Tag Archives: AI

The “Not to Be a Jerk, But…” Mist #MDF5 Blog

I write this piece fresh on the heels of Day 1 of 2020’s Mobility Field Day 5. Mist (Now a Juniper company) talked for about six years yesterday, and it was fairly riveting, end to end. It was one of those marathon sessions that needs a little time to settle in your brain before you can resolve it, figuring out what you actually liked to hear and what maybe raised some red flags. Let’s talk about those red (well, maybe a pale red, sorta orangey-pink) flags that sprouted in my mind as I slumbered on the whole thing.jerk

Mist Systems has had a fantastic run as a late-comer to a competitive industry filled with incumbents. That’s not easy, and their AI-inspired story has served them well. Now, we see the company moving i’s own cheese, and I can’t help but think about maybe  a few areas of concern.

  • Mist is no longer its own little WLAN product line bubble. Mist started off as a wireless-only product line. That let it focus on one discreet area with all of its development and quest for excellence. Sure, AI has been a key ingredient. But AI is not a magic wand. Just because you use it, doesn’t mean you have the Golden Ticket forever. NOW, Mist is spreading its methodology into the Juniper LAN side of its new house… the bigger you go, the more places there are for things to go wrong. The more opportunities there are for code bugs…
  • Mist has finally introduced a respectable AP product line. Again, Mist has had the luxury of not offering many APs to date. Life has got to be easier on the development side when your product set is smaller, I’m guessing. We see it frequently from other vendors- certain model APs are prone to issues and bugs. Will Mist bump into the same sort of customer-facing shame now that they have some diversity of AP lineup? Or will their promised self-debugging whizz-banginess eliminate that as a potential? Time will tell.
  • The um… well… uncomfortable thing to mention. I have the utmost respect for Mist’s senior leadership. Their results to date with injecting the AI/higher reliability story into an industry often fraught with overpriced buggy code suck speak for themselves. However- some of Mist’s senior folks come from that world of buggy code suck. They helped to author the very realm they now take potshots at. I mention this only to make the point that nobody is  perfect with a perfect past, and that history sometimes repeats itself. In the buggy code suck world, complexity only exasperates the buggy stuff, and Mist, as an overall operational paradigm, is very much becoming more complex as it matures. Ergo… more opportunities to stumble? Hopefully, they can keep it on the rails and not fall victim to the past woes that some of their own Bigs have at least partial ownership of elsewhere in the industry.

That little burst of sunshine aside, it really was a thought-provoking session. See it for yourself here, and feel free to leave me a comment below.

 

Extreme Networks Has Good Footing to Lead Network Fabric Evolution from Hype to Reality

If you manage a  network today, you are likely getting peppered by the drumbeat of  ideas for new ways of doing networking. Concepts like SDN, automation, AI, machine learning and fabric are becoming the next-generation lexicon of connectivity. Sure, us long-timers have heard it all before in different incarnations- but this is a pot that is really beginning to simmer while the industry tries to collectively move the way enterprise networks are done forward.

Meanwhile, those of us in the trenches have production environments to run. It’s not particularly comfortable to contemplate moving our own cheese in response to abstract promises of better ways and sunnier days, but Extreme Networks,Inc. may just be the company to break down the wall of hype and deliver the industry to the actual realization of the promise of network fabric architectures.

Before I get into why I think Extreme is the most likely company to show that the new network magic can actually be delivered in a way that leads to wide-scale adoption, let me share one of the best whitepapers I’ve read yet on what vendors are actually trying to do with the latest fabric initiatives. All the expected promises of simplification and reduced OpEx are in the Extreme Automated Campus document, but so is an excellent summation on some of the not-so-obvious advantages and evolutions that come with a properly implemented automated network. Among them:

  • The use of 802.1aq Shortest Path Bridging (SPB) as essentially a single-protocol replacement for traditional building blocks like MPLS, BGP, multicast PIM, OSPF, VLANs, and others. That’s huge, and reduces complexity by several orders of magnitude in large environments.
  • The notion that hop-by-hop network provisioning is a thing of the past. The network core is essentially unseen to most network admins, and all changes are done on the edge (live and without outages/maintenance windows).
  • User and device policies are the basis for automated network changes, and constant analytics provide feedback used to tune performance and anticipate issues.
  • By employing hyper-segmentation, a security breach in one part of the network is contained like never before, as the rest of the network is invisible to the bad guys because the old protocols leveraged for nefarious purposes are no longer present.
  • The use of APIs mean that third-party network components can interoperate with Extreme’s Automated Campus.

Extreme 3

There’s a lot more to the whitepaper, and I encourage anyone who’s been underwhelmed by other explanations of what network fabrics/automation are supposed to deliver read it as an excellent primer.

As I digested insights from Extreme’s whitepaper, I also found myself reminded that obsolescence can be insidious with the legacy methods we do networking with now. Dated designs can underperform today and fail tomorrow while we miss subtle signs of trouble because of disparate logs and dashboards. This isn’t news to anyone running large business networks, and is why automated analytics has a fairly strong appeal. This brings me back to Extreme and what puts them at the head of the pack within the networking space.

Extreme pioneered and set the bar high for network analytics with its ExtremeAnalytics platform. The value proposition has been proven in many cases, via a range of customer relationships. Where other networking companies are relying on third -parties or are just getting around to developing analytics solutions, Extreme has been optimizing networks based on machine-learning analytics for years.

Extreme 1

Then there is Extreme’s purchase of Avaya earlier this year. By my estimation, Avaya was the absolute creator of SDN-enabled network fabric environments. I visited the company’s Silicon Valley facilities in 2014 during Tech Field Day, and got a first-hand look at the impressive technology that  has become part of Extreme’s fabric offerings. Extreme now has real-world fabric customers and a mature offering among newcomers to the game.

Extreme 2

The fabric/SDN thing is here to stay as evidenced by the market leaders all talking it up as “what comes next” in unified networking. But how to get there – and whether you want to stay with your incumbent networking vendor for the leap – is a more complicated discussion. Some of the new initiatives feel cobbled-together, i.e. placing  frameworks of APIs into legacy hardware that may not have the best track-records for reliability. I’m of the opinion that some vendors are trying to figure out how to proceed with network-wide fabric methods,  while painting beta-grade efforts up with glitz and catchy slogans (though lacking depth and a track-record). This just isn’t the case for Extreme.

Extreme has done a great job in integrating their acquired Avaya fabric assets with their established portfolio and consolidating it all (along with their excellent technical support) into the Extreme Automated Campus. It’s new, on paper, but made up of mature industry-leading building-blocks. This is why I see Extreme as the one to beat in this space.

Learn more about the Automated Campus solution here.

Register for Extreme’s upcoming Automated Campus webinar here.

 

FTC-required disclosure: I was compensated to comment on the Extreme Networks Automated Campus referenced in this blog, by PR company Racepoint Global. I have no direct business relationship with Extreme Networks, and in no way claim to be an Extreme Networks customer or representative of Extreme Networks. The opinions expressed here are my own, and absolutely true at the time of publication.

Mist Systems Polishes Their Message at Mobility Field Day 2

It’s always refreshing when a truly original story comes along in the WLAN world. Mist Systems isn’t quite brand new (I wrote about them for Network Computing back in 2016) but their approach is fresh enough to cause some good energy in the room when you do get the chance to hear a briefing from the company’s top dogs, and I recently got that chance again at Mobility Field Day 2.

Here’s the thing about Mist, now- today: if you’re not careful, their story can sound like another one of the many from network vendors where terms like AI and Machine Learning are bandied about like the Buzzword Flavors of the Month. But Mist was talking this language well ahead of the current curve. Where established vendors are largely painting their long-running gear up in a coating of hype and APIs, Mist is actually new magic built by data scientists and proven network visionaries. It’s heady, exciting stuff.

But can Mist make a legitimate go of it as new player in the Big Customer Kickball Game where most current potential customers have already chosen a side? Here, only time will tell as Mist’s marketing paradigm is put to the test. I can share my own opinions and gut reactions from the Mobility Field Day for you to consider, and welcome any dissenting opinions or comments:

In Mist’s Favor

  • When Bob Friday and team tell the tech story behind Mist, it’s impressive and believable
  • Mist is the real deal when it comes to Machine Learning, etc- where the message feels forced with other vendors
  • Mist seems to have mastered the UI challenge (put lots of important stuff in front of the WLAN admin without making it feel like overload) with their cloud dashboard
  • Mist uses no controllers (bug hotels) or user-upkept bloated NMS system
  • They tell a great story on bug management and code quality
  • The virtual BLE beacon thing is huge. As in freakin huge. And it can stand alone even if you don’t need Mist’s WiFi solution
  • Nyansa-like analytics are compelling
  • Long-time users of established systems are getting burned out in spots on license overkill, huge costs- creating potential openings for a WLAN vendor change

Of Concern

  • Mist is late to the overall WLAN party, so is up against established players
  • The lack of switches and security appliances can be problematic in some RFPs, and when looked at through bullshit lenses lenses like Gartners Magic Quadrant
  • We’re still not hearing enough about “unnamed Fortune Blah Blah Blah customers” to really do our own independent verification of how Mist is working out in the real world
  • Mist is just getting ready to ship outside APs later in 2017, and how that impacts their analytics (especially when outside/inside WLAN are managed in same pain of glass) remains to be seen

I really enjoyed what I saw and heard from Mist, and it’s obvious that the company’s leaders truly believe in their baby’s potential. And- you don’t just have to hear my opinion… form your own after watching the Mist MFD session here.