Particle Mesh update — a note from the CEO

If I’m not mistaken you should still be able to add a device via CLI - just not via mobile app or cloud - but maybe @zach can comment on that.

If I’m wrong about the above that would be something I’d suggest as a way to mitigate the sores that will come with the deprecation: Allow for new nodes to be added to an existing mesh without the cloud ever needing to know about it.

He did already in the announcement. It says in the PARTICLE MESH DEPRECATION TIMELINE:
Dec 31, 2020 : ability to set up new mesh networks, or add devices to existing mesh networks will be removed from the mobile app and cloud APIs
Jan 31, 2021 : Network management features (the “networks” page) in the Console will be removed

So, no network management anymore. Any existing Mesh setup will have to remain as they are.

This is the part that I understand as a limiting clause to the preceding statement.
CLI is neither touched by the mobile app, nor does it have to use the cloud API - nor console.

Still, I might misunderstand that, but as said above

There are still two main releases (1.5.0 and 1.6.0) expected to come and in one of these any device OS limitations that would prevent adding nodes to a mesh locally could - and should IMO - be lifted.
I’d be fine with the Xenons becoming invisible for the cloud but still locally connected.

2 Likes

Let me address one of the more common concerns being raised: “how can I ever trust Particle again?”

First, let me start by saying: I don’t think it’s unreasonable for any of you to have lost trust in us. Now it’s our job to earn your trust back, and I hope you’ll give us the opportunity.

One of the things that I personally learned from the Mesh launch is that we need to think differently about product launches for technologies that we are still exploring. We launched Mesh the same way we launched Wi-Fi and Cellular – with a bang. We did a big pre-order campaign, generated a lot of excitement around it, and made it a major element of our product portfolio.

In hindsight, this was foolish. At the time of the launch, we were absolutely excited about the potential for Particle Mesh, but we didn’t yet know how customers were going to use it in practice. The product needed some market discovery before we could really build confidence that the underlying technology (802.15.4 Thread-based mesh networking) was going to work for our customers.

What we should have done – and have now built into our product development process – is launch the product in beta. We should have been more transparent that there was still work to be done to explore how and when Particle Mesh would (or wouldn’t) be a good fit for customers.

The “Product Lifecycle Policy” that I’ve linked to in our documentation is somewhat new; we defined this policy after the Mesh launch was behind us. But going forward, it is something we intend to stick to. If we launch a product in GA (“generally available”), it’s fully supported and production ready. If we launch a product in beta, we’re still testing it.

If you’re going to trust us, it shouldn’t be blind trust. It should be because we’ve learned from the experience of this product launch and have baked those learnings into how we will do future product launches so that you can feel confident about the future of products in GA while, if you’re interested in doing so, exploring new technologies with us in beta.

9 Likes

Oh, this makes me feel so much better now.

1 Like

that will be extremely cool!

1 Like

I know this sucks for many. I’ve been there myself with other products. The worst hit was using the Intel Edison module for me. My spidey senses told me don’t use it. The forum is riddled with issues, and it’s too new of a product. Not to mention Intel never made something like this before. But it was cool, and I could not resist. I paid a big price for not listening to myself.

The lesson i would take away from this is to look to the forums to see if you should invest in a new product just yet. Reading the forums will give a glimpse into how they handle things, and if its really ready.

NOTE: I say the rest not to dig at particle, but to offer a few suggestions since you said:

I have started with particle when the Photon first came out, and loved it. Then I moved onto the P1, and made a few commercial products that run 24/7. Not had one fail yet in over 4 years.

These are the things I see that you need to do / address.

#1) Get rid of that dam CLI interface. It’s absurd to say the least, and should have been a GUI interface long ago. All of my industrial customers will never allow a internal device to be connected “to the cloud” to get its updates through the Particle servers. If there is a update, the customer needs to do it, and there is no way they are going to understand CLI.

#2) This one may hurt. Stop releasing products before they are ready for prime time. This has been the theme of Particle from day one. Every new product release you give a big blurb of info showing it off, saying the product is coming out in x months from now. Everyone gets so excited, and when it hits the streets it has many (software) issues that clearly shows you did not spend enough time proving it all out. The hardware has always been is solid IMHO. It takes a special kind of engineer to final test a product. The kind that can switch gears in their head and think like a new customer which typically does not read the docs and does silly things that cause it to have problems.

I would have agreed with you whole hardheartedly months ago, and I tried hard to tell Particle just that. Sadly, they had other plans. That awesome window of opportunity for them is now gone. Others have filled the gap like VisualGDB & PlatformIO, and even now espressif has release a plugin for Visual Studio Code. IMHO, Particle could have made more money than they ever did, and became the new “Arduino” for ESP32. I started using the ESP32 about 8 months ago, and bang for buck you just cant touch it with anything else that is out there.

4 Likes

I want to make sure it is clear that we are not bricking any Xenon hardware, and previously activated Xenons will not lose connectivity to the Particle Cloud.

Rather, we are ceasing new feature development for Particle Mesh, will not be actively manufacturing new Xenon devices, and will be allowing new Xenons to be activated and added to Mesh networks through the end of 2020.

We have no plans to discontinue cloud connectivity, OTA, or cloud API functionality for Xenons.

What are we deprecating?

  • In March we will release the Device OS v1.6.X line which will be the last with new feature development for Particle Mesh (and Xenons)
  • We will no longer be manufacturing new Xenon hardware, and will be selling through our remaining inventory at discounted rates
  • At the end of 2020 it will no longer be possible to activate new Xenon devices with Particle
1 Like

I really appreciate the kind words in your post and am glad that you have felt that Particle has demonstrated genuine care and support in your interactions with us. That is always our goal, and while it is clear we have not succeeded with every customer I’m glad to know you have felt that way :slight_smile:

1 Like

And this is because I totally lost interest in mesh last year:

One more thing:

why are you going to cease Xenon production if BLE is here to stay and an ETH featherwing is available ? Am I missing something ?

Thank you.

2 Likes

I am currently working on a project that relies on the Particle Mesh. You(all guys from Particle) said that we can use BLE - which is fine for my use case. But my question is: If I stick to Particle, would you develop and provide us with an easy pairing method as you did with Mesh(just scanning the DataMatix on the device) and also an easy way to publish messages from the BLE nodes to the Particle cloud as it was with Particle.publish with mesh?

2 Likes

Zach, thank you for your honesty.
Having worked in the professional electronics industry I know how easy it is to get all too excited about new technology (or market opportunities) and not taking the time to do in-depth exploration. “Fake it until you make it” is unfortunately also too common. Having that good Product Management Process in place forces you to ask questions at the end of each phase so you are not caught by unpleasant surprises downstream. Really important if you, like in our case, make products with an extended life span (> 25 years). I hope this event was the wake-up call Particle needed to get to a more mature organization.

I want to make sure I understand your question – do you mean:

  • Easily pairing a Particle device to your phone using the data matrix code -OR-
  • Easily pairing two Particle devices to one another using the phone as the interface (like it is used during Mesh network setup to “introduce” the joining device to the commissioning device)?
1 Like

This is the most concerning part to me. The hardware appears to be done and Particle provides a great quality and price on nRF52 boards. I can understand dropping backend support for the hardware. I haven’t yet seen a good argument for stopping manufacturing.

As a company that creates hardware I think it’s important that people know that hardware will be available when replacements are needed. I appreciate that you’ve open-sourced the design files and am hopeful someone creates a xenon-clone. I’d much rather buy from Particle though.

If I was looking for hardware to run a new project on I’d have to take in to consideration that Particle boards may not be available when I need them.

2 Likes

With the removal of Mesh, will more program space be made available?

As an aside, I was in your presentation at Embedded World 2019 and got to talk to you at your booth, and I don’t doubt the effort that was put into developing the mesh api’s, and it was exciting to see the jump into trying something new. I think the mesh capabilities that were hoped for initially are met better with LORA, and the writing was on the walls with the amount of limitations of the interactions and setups of the mesh networks.

1 Like

Originally missed this in the backscroll, so just wanted to provide a definitive statement that the P1 is a generally available Particle device with full enterprise support. Its current inventory status at any moment is not an indication of future plans for the SKU.

If its lifecycle status ever changes (we presently have no plans to do so), we will communicate those plans clearly in accordance with our Product Lifecycle policy.

Thanks for your feedback, and I really appreciate that you have been satisfied by the quality of our hardware.

Ultimately, though, our goal is to assist customers in building products that connect to the Internet. The Particle development model, which includes key tools like OTA updates, device vitals, and device messaging, relies on a managed connection back to the Particle Cloud.

Without a persistent Cloud connection to support these features, much of the value of the Particle platform is eroded for that device. In other words, Xenons without a Particle Mesh backhaul to our Cloud no longer meet the bar that we’ve set for “first class” Particle devices.

We feel there are many companies that can build high quality development kits for bluetooth chips like the nRF52840, and want instead to use our expertise to focus on making the deployment and management of connected products as easy as possible for our customers.

1 Like

@will From memory (so I might have some dates wrong) the P1 became unavailable somewhere last year. The statement on your site was availability back in may (or so). May came and went and the statement was updated to August (or so). At some point the P1 was indeed available for a brief period (days). It became unavailable again and the dates shifted some more etc. Your site currently states that they are out of stock till Feb 29. For a while, the P1 was not listed as a product on your site. All that makes one wonder.

I asked for clarification on your site/sales/tech team. I asked for clarification on the boards. Crickets.

I understand part outages and will not hold that against Particle. I know you don’t make that part and are reliant on other partners/market. It is the lack of open communication that does me in.

But don’t worry, I have moved on. The photons and p1s I still have will not go into production with us.

Hello @zach

Hope you (or someone from particle) find sometime to read my post in your busy day. I’ll TRY to keep this short, but some of my honest feedback here so it might be a bit of a read. I’ll also be attending the Thursday live Q&A and will ask questions there, so I guess just some headups.

I’ve read through your post multiple times as well as a good chunk of the replies. Before I start:

  • Particle as a platform has been great for us so far. There are still many features that we are happy with and we will continue using Particle as an enterprise customer.
  • Particle has done a lot of things right and somethings wrong. Mesh is one of the wrong things and @zach I appreciate the honest and transparent post. The launch of mesh was also a mistaken as addressed in Particle Mesh update — a note from the CEO . There are consequencies to this which affects my company greatly, I will elaborate this in details below.

When investigating IOT platforms, mesh was one of the factors why we considered Particle. No other platform did this at all. We were right on board with the Gen3 launched and purchased multiple developer kits of all Argon, Borons (Both 3G and LTE-M) and Xenons. Although mesh “did not just work” functionality it was there and I had no problem setting up small networks of 5-10 devices. High availbility was a must for us in the beginning so we had hope that this feature will make it in eventually similar to the eventual deployment of NFC and BLE APIs which were not availble at launch or had some bugs. After 6-8 months of testing and prototyping with Particles and 2-3 other platforms we settled on Particle mid 2019.

The reason we chose particles over the other platforms were the following:

  • Particle dashboard - Your quote “We’re doubling down on our most powerful features – over-the-air firmware updates and device diagnostics – and we’re working on new solutions specifically for mobility and asset tracking, which have become some of the most common use cases for Particle”. This is one of the TOP reasons. Good diagnostics, fleet management and reliable FOTA which accelerated product development by months.
  • Support - We started off as a small quantity developer under the radar and support so far through the forms of via hello@particle.com were great. This was without paying anything aside from the developer kits.
  • Particle Enterprise Benefits - Creating orgs, enterprise dashboard etc… are great along with commited, guarentee’s volumes and 24h turn around support. Enterprise support has been exellent so far.
  • Architecture - As mentioned before, we compared particle to 2-3 other platforms and particle’s architecture stood out a lot. Your gen3 products are really something.
  • Tools - Particle developer tools are surprisingly good. Visual code add on, particle-cli etc… are great and a actually miles ahead of other platform tools.
  • Particle’s track record - Going with any 3rd party vendor comes depreciation risk. We considered particle discontinuing their products or support for “older” units. The electron and photon units have been supported for a few years and the updates still seem to be going strong for them. Given that Gen3 units just launched, we were expecting a long life cycle for these units. So this depreciation is really concerning for us. One saving grace is that , everything from your deviceOS, hardware schematics are all opened source which was very unique and cannot be said for all the other platforms.
  • Mesh - You guys were the only one I can find support this feature and actually set the benchmark. This made you very unique. No matter how “bad” your mesh implementation was, there was actually no other “pre-made” alternative.

By “settling” on Particle, we’ve signed an enterprise contract for all 3 Gen3 skews and commited for quantities in the thousands. This followed by spending the last 6-8 months after signing designing products entirely AROUND the Gen3 particle platform. This includes custom PCBs, plastic covers/molds etc… which comes with significant costs easily 6 digits in dollars of R&D.

So the questions I have in mind and maybe anwsered on Thursday are as follows:

  • Long term support of Gen3 products - Now that mesh is gone, there is less difference between the older gen particle models verses the new Gen3 units. A boron is simply a beef up electron and an Argon is a beef up Photon. From a business prespective now these 2 products seem to cannibalize each other more if it were me, eventually I assume one of the skews will have to go. What is Particle’s LONG term support plan for these product skews. I’m not only talking about the next 1-2 years but 4-5 years. The electron was launch in 2015, and given that the Gen3 units launched early 2019, I hope the Gen3’s supported life time is at LEAST just as long as it now. This is the “can we trust” part people are talking about. You mentioned it will take time to gain back trust but for people like me who are far down enough into a rabbit hole with particle, a longer term road map is much appreciated.
  • Mesh as a 3rd party particle library - I know Particle is trying to do a “it just works” approach. But I also assume the large majority of Particle users are somewhat technical. While we would REALLY appreciate a “it just works” approach, some of us can do without this “Apple” approach and quality. Some of us do not mind configuring mesh and setting it up. If Particle decides that it’s resources is better put not supporting and developing mesh (+ all the reasons that mesh is NOT the right technology) thats fine, its a company decision with resource/financial balacing that was done. Nothing I say here will probably change this decision. But instead of completely killing the feature, some of us would appreciate it if mesh was perhaps bundled into a 3rd party particle library (+ documentation) and posted up somewhere before completely dusting this mesh feature completely off your table. The mesh feature was developed based on OpenThread which is opened source and great. A Particle user could probably do what you did, pull the SDKs, develope on it and get something working. But since you’ve already done all the heavy lifting (since 2017!), if you don’t want to deal with it anymore, at least just give it back to the open source community instead of burning it to the ground. I would (and maybe some others) would really appreciate this as a good gesture and a last gesture for a product that (as you’ve also mentioned) launched “with a bang” along with Cellular and WiFi making it SEEMED like a ready to go complete feature. You’ve already addressed as part of “how can I ever trust Particle again?”, mistakes happens and I am happy that Particle has admited to this fault and we now have a clear line between “beta” and general availbility features. But admiting a mistake is one thing and can be done with just words but action to try to reduce the damaged caused is more valuable. We esentially bough a product (and paid more for a Boron/Argon vs an Electron/Photon) with one of its core features now being removed from support. Just refunding the Xenon purchases only scratches the surface of the problem for those who bought Particles for mesh and greatly affects those who developed products around Particle for mesh. In summary, if you commited to supporting it and now you don’t want to, that sucks for all of us. It will take effort to decouple mesh from the device OS + cloud, but you guys already plan to do this after OS 1.6X. At least bundle this somewhat working feature with the documents you already have and hand it over. If Xenons are not being manfactured, Mesh can still happen between Argon’s and Borons that are.
8 Likes

Thank you for your reply.

Maybe I’m not understanding your announcement. I’m under the impression that as of February 2021, all Xenons will cease to be particle cloud devices. If my Xenons would have the exact same functionality the Xenons currently have including OTA firmware updates, and API functionality including both Mesh and Particle cloud calls, and I could expect that functionality to continue indefinitely, as long as you operate your cloud, then I don’t think I’d have any issue at all with your decision. But my understanding is that’s not the case. Am I wrong about that?

Of the dozen Xenons I bought in the preorder, 11 have never been opened, the 12th I played with for a while with an Argon and then decided to wait for you to get the technology more mature. If I build a few Argon-based Mesh networks now and activate all my Xenons on one those Mesh networks, then I will be able to continue using them beyond 2021, including writing firmware for them beyond that date?

FWIW, I don’t really see myself ever needing much mesh functionality. The only important thing Mesh does for me is allow me to wake up from sleep, take some sensor readings, and publish them to my gateway to republish to the cloud in under 5 seconds using less than 1mAh of power. It lets me run sensor nodes off batteries in a way that would be impossible over straight WiFi.

So I should be able to do everything with my Xenon in 2022 that I can do today? What happens if I knock the Xenon off my mesh network? A couple of times I managed to get my Xenon into a non-bootable state through a bug in my firmware and in recovering the device I fully lost my Mesh credentials and had to reconfigure it from scratch.

What do I do with the 11 never-opened Xenons I have? Assigning them all to Mesh networks now is the right choice?