Particle Mesh update — a note from the CEO

This is a very important point. I have found that on occasion keys got corrupted or something else happened which required some kind of auth with the upstream Particle cloud. It would be good to get guarantees about this serviceability into the future.

I had similar thoughts. Thanks for articulating this so clearly. The mesh is very important to a core group of us, and while technically BLE might be able to satisfy our need we really don’t want to reinvent the wheel. If Particle can commit to minimum support to keep Xenons alive and kicking in their current form, and perhaps have a small lib so that <=v1.6 can still effortlessly communicate via BLE with >v1.6 that might provide 99% of the functionality for 1% of the effort.

2 Likes

I have been an engineer and developer for more than 30 years in many industries and many embedded devices. I purchased my first Spark Core in 2013 when when I discovered they were onto something special. Building in all the pieces hardware, firmware, tools, services and a dynamic and responsive support group for one powerful little device. So too were the Photon and Electron milestone devices. It was with great expectations when the 3rd Gen with Mesh was introduced in the fall of 2018. The Mesh capability was the shining star that set Particle apart from all the others. During previous year I and many others were experimenting with nRF24L01 to build out our own low powered networks. Particle took the bold step and built an amazing Mesh family with the means to reach the last 100 meters of our distributed networks. As a young company there were teething pains and I’m sure they have grown a lot in the past couple of years. I am constantly impressed by the dedication of all the support people that respond almost immediately to the broad range of questions. It is incredibly difficult to handle questions from a broad range of developers from newbie hacker to experienced professional. No product can be all things to all customers. I know Particle realizes that there should have a great deal more testing with a range of customers both large and small. For the first six months of 2019, there was a lot of catch up to fix a lot of problems and stability issues. Then somewhere around June or July the stability improved, releases came more regularly, the bugs were getting squashed. I found my application uptime was consistency high. Everything was great. Then came the news from Zach and the Xenon deprecation. It hit me like a ton of bricks. What was wrong? They had surmounted the great obstacles. The finish line was in sight. I would really like to know why the Xenon was considered technically unsupportable. Was it because they didn’t think OpenSource was going to be a major standard? Were there complaints from customers that were expecting kilometer transmission ranges? My suggestion for what it is worth is that they should have defined as clear a specification as possible what Xenon will and will not do. Is it possible to use it in a noisy industrial application? Perhaps not. It is what it is for now. It would be better to iterate and come out with a version 2.0 in 2 years that addresses the problems that are discovered in the field. I too hope that Particle will do what is in the best interests of the many Mesh developers and maintain as much of the Xenon support as possible.

4 Likes

My understanding is that yes, removing the mesh functionality will free up some user space. But I don’t currently know how much that is.

Your situation is similar to mine. My conclusion is identical to yours…

This was asked during the livestream but not answered.

“Xenon Mesh has a been totally successful technical solution for my application. Who are these customers that are having problems? Perhaps they have unrealistic expectations? Did Particle’s clearly specify what those capabilities and limitations are? Did they do the field testing to characterize the devices and ranges and environments? “It just works” is not a specification, it’s marketing speak. That’s Particle’s fault for not being clear. BTW - Expecting Mesh to work through a refrigerator wall (assuming it’s metal) is unrealistic and Particle needed to make it clear that the mesh does not support ALL use cases. That’s the problem with trying please everyone. You end up pleasing no one.”

1 Like

@zach attended the live Q&A. Thank you for hosting it.

There were a bunch of questions posted but it was mentioned that all questions were anwsered. I am assuming they were not seen and we also sort of ran out of time. Here are the questions I posted on the live Q&A stream.

  • With regards to “some of us may not be networking experts”, that is true. However, those who need mesh badly enough will put the time and effort to figure this out or companies may even hire specifically to implement this internally. This will definitely help those who want mesh and will probably help mesh mature in the event Particle revisits mesh in the future.
  • In addition, those who bought Gen 3 units are now paying for hardware that now cannot be used. To me this is a problem. Any considerations of a price adjustment with this feature cut? Now some may regret their decision and wished they bought an electron or photon instead.
  • With regards to it not being the correct technology, I honestly do not buy that. The wireless technology may have limitations and It may not fit some use cases but having the ability an option to do so is much better than having no option. Especially coming for a product that pitches itself as a platform. Not all users will use all features on a platform, only those that are needed. In addition, the mesh management techniques, user flow will probably carry forward regardless of what wireless technology is used. You brought up using BLE for small micro networks, this is fine but what if we now have applications that need both BLE and micro networks simultaneously? With the deprecation of an option you now hinder the flexibility of the platform in general.
  • I am an enterprise customer and I signed numbers for all 3 Argon, Boron and Xenon units. Now that mesh is gone, what will happen to my commitment numbers. Deprecation of mesh will now affect all numbers for all Gen3 skews.
  • Any comments on bundling mesh into a 3rd party particle library? I brought this up a few posts ago:

One more point I would like to bring up in the live stream. You also mentioned the main focus of mesh was creating large networks over large range and then short 10m range of 2.4Ghz was not sufficient. This is the “ideal” and perfect scenario I agree. However, I see mesh at the moment as more of a cost saving solution. In the event I want a small number of devices (e.g. 5-10) connected out doors in close proximity, mesh solved the problem. With out it now I would have to purchase 5-10 Argons and Borons which are almost double or quadruple the price in terms of hardware. Our goal was to perhaps sell bundled pre configured sets of meshed 2-5 units to customers as a cheaper alternative then buying each gate way skew of the product.

3 Likes

This was kind of my thought of the Mesh implementation all along. For indoor use, if the area isn’t already covered by WiFI, it’s just not that hard or expensive to install a router. For outdoor use, if you’re talking about a small (up to 10) device setup with 5-10m range between devices, a single WiFi router will easily give you more range than the particle mesh would have.

So the only difference is cost as you say. But if we’re talking about a max $10 difference between a Xenon and an Argon and $100 for the router (if needed), it’s a $200 difference in hardware. That’s not even a rounding error on the installation cost for such a setup, let alone the R&D for the project. So IMO, the cost savings using Mesh is meaningless.

I also don’t understand why, if their goal was large networks over large range they would use devices with such radio limited range. Not to mention if they were talking about 100 node networks from day one of their pre-order announcement, they chose a chip that their earliest proof of concept test would have shown them didn’t realistically have the memory to go much beyond a 10 node network. If this was their end goal, they must have known they’d chosen the wrong technology before they ever announced the product.

1 Like

Yup this is the ideal and perfect scenario. For areas with bad connectivity. We have customers like this as well and we agree.

The cost between Argon and Xenons are the best case scenario. Yes its $10 USD, for Canadians its x1.3 so its about 13~14 and for other parts of the world it maybe more. In addition if you buy the trays, they do not come with an antenna which is needed for WiFi and Cellular. WiFi antenna being an additional $5.90USD and and cellular antennas around $9.80 USD. Installing a router is by no means expensive, and like I said is the best case scenario. Even looking at the base case scenario, certain routers have bandwidth capacity and can also only take up so many concurrent connections. If I am installing 100 of these devices with packages of 5 meshes, this gives me 25 simutanous connections verses 100. Now scale this up even more. 5 was an example but it could be batches of 10. In addition, if the customer is a corporate, business or company, additional WiFi router installations may also need to involve their IT services. Now if we take a look at the cost of cellular, where is close to $55-$60 USD per cellular device (excluding antenna), this becomes more significant when scaling over large numbers.

I agree cost savings may not be the main and absolute goal but to us it is another reason for mesh and another way to look into how it can be utalized.

And this is now another part of the huge problem people will have using particle at all in the future. Using Particle infrastructure for products that should be in use for many years means any company that relies on you is literally betting their entire existence on you. And you’ve now shown that you’re quite willing to pull the rug out from under people who have invest huge amounts of money and effort in your system.

You’re going to lose a lot of customers and potential customers over this, and the fallout will be huge for you. So even for a customer who is willing to give you the benefit of the doubt and figure you’ve learned a lesson and won’t let it happen again, you also have to reassure them that particle will survive the fallout from this. Will there even be a Particle cloud at all in 3 years after the hit you’re going to take over this? You’re going to have to do a lot to convince people there will be.

And it’s a different market than it was 2 years ago when you launched mesh. There’s a lot more good alternatives than there were 2 years ago, and many have a lot more flexibility in terms of not being so locked into an ecosystem. All of which just raises more doubt in a company that has to be their survival on you.

1 Like

We can help you out with this, OJ. If you submit a ticket to support.particle.io our team can look into your specific order and find a fix.

It is not the case that customers expected those kinds of transmission ranges, simply that the customers who we were working with the deploy Mesh were doing so mostly in commercial and industrial environments, versus in the smart home.

As such, extended distances across a factory floor, transmission through concrete floors of an apartment building, and penetration through metal containers were all challenges that 2.4GHz is almost suited to satisfy, but not quite. These are transmission challenges that sub-GHz technologies like LoRA are better equipped to overcome.

As a result, the majority of our Mesh customers were attempting to implement a communications technology that partially worked for their use case, but couldn’t deliver the strong reliability across a range of environments that those applications demanded. And, as Zach mentioned on the livestream, Particle is only successful when our customers are, and we weren’t confident that the majority of our interested customers could be successful with Particle Mesh.

2 Likes

Did you undertake to find out how many customers were able to implement Mesh successfully? You cannot make everyone happy. Sometimes its hard to say no to a customer with the promise of a large order. Could a scaled down product be supported within a viable market niche? Carefully defined specifications, field tests and Beta test programs help make better products, clearer expectations and happier customers. A small success is often preferable to a large and painful failure. Gain the experience with lowered expectations and iterate with your customer’s successes and failures. I feel for the individuals and companies that have invested a great deal of time, money and resources. Mesh technology (with its imperfections) made Particle stand out in the crowd, but now - not so much.

2 Likes

… Seriously? So, these were not aspects that you considered and tested thoroughly before you make us jump onto the bandwagon?

2 Likes

so you are launching the product first and then figure out who the target customer group is?

I’ve also supported Particle since the Spark Core and appreciate all the effort that went in to making the first accessible IoT Arduino type device.

Whilst we are using Mesh now, for a very specific case in Mexico, I always thought it seemed quite niche.

For our other projects, the ESP32 has been perfect. It is a fraction of the cost of the Argon or even Photon, and has a huge community supporting it. Getting cheap ESP32 boards made in china at low volumes is easy. I always thought Particle should have embraced this chip (not just as a radio for Argon), as well as other 3rd party devices. The Particle dashboard, OTA and console would have been a great contribution. As it is, Particle seems too closed an ecosystem and doesn’t embrace popular technologies.

Having said all of this, I agree that the cellular products are fantastic.

2 Likes

Hmmm, maybe a fancy way to scratch our heads or backs? Didn’t anyone come come up with that use case yet?

1 Like

Who knows when this will be discontinued…
I was also a Particle fanboy, all the greater is now the disappointment.

1 Like

True but in this case I choose to have a bit more faith in Particle here. In the Q&A live stream they ensured us that their flag ship products are not going any away anytime soon. I know it’s currently words but they have a good track record with the electron and photons so far. This might be a personal judgement call but for me, everyone gets more than 1 strike with time allowing wounds to heal (i.e. frequency) and how damage can be mitigated. No one will have a perfect track record forever and I rather channel the energy into damage reduction and future mitigation at this stage.

Mesh is hard and was hyped. But do take a step back and look at alternatives, and then reevaluate particle. Particle has still done a lot of good in my opinion and provides huge gains.

5 Likes

Slick marketing strategy: lure the customer with cheap devices (Xenon) to invest in and commit to your ecosystem. Then, when they are hooked, pull the rug and make them buy the more expensive chips that cost four, five times as much… and maybe lure them again with throwing them a little bone (a.k.a. “store credit”).

1 Like

I didn’t think mesh was going away either. I’m not quite ready to forgive and forget. I’ve got way too much time and money invested in this. My pockets aren’t as deep as Particle. I can’t afford to start over.

One thing is being overlooked IMHO. It was sold as something that was ready to go. It was not sold as an experiment that did t work. They as a company can make business decisions and terminate products, I get that. But they should not be selling it as a functional product. That is why they have changed their policy now. Further, I believe others have been successful. I don’t know much about it but I think pycom.io offers mesh, cellular, WiFi, sigfox, etc. I will be checking into it for sure.

2 Likes