Primarily a Losant user here. At RWB’s suggestion I gave Losant a try last year and found I was quickly productive in getting prototype functioning. Very generous trial terms and great support forum.
I’m considering transition from trial to paid account. I am finding this a large leap in cost versus other alternatives.
My question to this group … for first step into paid IoT cloud platform (say for < 50 devices), what have you found to be cost effective?
Can you speak to what kind of cost your seeing with Losant since its usually not listed on their site?
ThingSpeak is free for 3 million messages per year (but there is a 15 second interval limit).
You can purchase 33 million/year with a 1 second limit for $650. That's a message every second for the year, unlimited # of Channels [edit] now it's 250 Channels max.
I don't know how this compares to the other Cloud Services. Several others don't make the pricing easily available.
Pricing is really all going to depend on how much data and how frequent you sent data for any service
Google Cloud IoT Core pricing can be found here (I use IoT Core)
https://cloud.google.com/iot/pricing
Azure pricing seems a little more expensive depending on your needs
I’m a fan of Tinamous, it’s free, up to 25 devices. After linking your particle account everything shows up automatically and you can send detailed data via senml format via particle.publish, no webhook or other integration necessary. I’m not familiar with the alternatives listed so no idea how it compares feature wise.
ubidots has paid options for $20, $100, $500, $2500 per month supporting 10, 60, 400, 5000 devices (respectively).
Losant starts at $100/month and climbs substantially after year 1.
What does Losant give you for $100 per month?
For starters … multi-user/tenant experiences (web sites) to support user specific dashboards, user specific access control to those web pages, etc.
How about device quantity and payloads per month?
I also received a quote of $100/mo from Losant. This was their cheapest plan, though they didn’t really specify how many devices/data this plan would be able to support. Business licenses start at $30K/yr.
I guess the exact price would be a negotiation and depend on device/data quantity.
Hi WoobaGooba, I am doing similar research now for use for a commercial use. What was your decision about IoT Platform? Have you moved forward with Losant or found anything better for your needs? Please feel free to PM me if needed.
All,
I have been very happy with Ubidots pricing and a steady stream of new features. Worth a look.
Chip
azure has IoT hub, IoT central, IoT edge, etc.
IoT hub: the price is nice, you can store the data in tables for a reasonable price as well. you can also trigger node.js functions stored on azure. be careful with their billing. it can escalate rapidly and no one seems to know what’s going on. i’ve gotten refunds but no real explanation as to why other than my node.js apps have 1GB of files and access to those cost a lot – this was NOT clear at the outset. it’s been over a month and i haven’t gotten a good response. therefore, i’m storing data in their flat tables but i’ve migrated all functions to a VM. essentially, with azure, you replace system admin hell with arbitrary billing insight hell. their customer service people are super nice, though. their system is also far superior to both ibm bluemix and google gcp. it’s far easier to use and far better documented. and usually cheaper. i hope they get their act together.
initialstate.com: nice graphs, very reasonably priced. $10/month for 3 million messages. you can organize the devices however you want. graphs seem nice enough. they were acquired by tektronix, so they should be around for a while.
blynk is quite expensive for commercial use ($200/mo or so) until you hit a certain number of devices.
essentially, i’m looking for a dashboard that works on laptops and mobile where the per-device cost is about $1/YEAR. most are $1/month, which is simply useless for our billing rate to our customers. there’s probably a market gap here that we should fill with a new product :-). it wouldn’t even be that hard.
I guess you got lucky with Azure.
I have been working with Azure and Google Cloud (GCP) in recent years and every time I want to implement something in Azure is nightmare-ish because I cannot find anything in their docs.
On the other hand, using Firebase on GCP is been almost a dream where the only limitation I find is my lack of knowledge and not their docs.
Perhaps, I did not get as lucky as you with Azure and that's it...
For my data logging application I built my own with Azure Functions (free) and a basic SQL server. Then you can read the SQL server from excel or whatever.
You could also build additional functions to trigger alerts and respond back to the devices.
what is firebase? i spent over a month deploying a basic test case on ibm bluemix (took 7 days to implement), gcp (10 days and it still wouldn’t work without a 3rd party tool). i wasn’t even considering azure because i figured microsoft doesn’t know much about server side. and i was wrong. their UI is easily the best. their examples are easy to follow, many of them are for both C# and node.js – nice, because i thought they’d try to push their environment down my throat. i don’t want to install a giant eclipse-like IDE, no thanks. notepad++ and node.js is plenty fine.
while azure’s pricing tool is pretty slick, turns out that their billing is very fine grained and not clear at all. and that’s why i left.
be careful with azure functions. it’s what i used. it was triggered by IoTHub_EventHub and saved data to storage tables. all was going well until i started testing ONE single IoT device in earnest. bang, the bill went up to $20 in a week. why? accessing node.js modules files, apparently. been over a month and they still haven’t figured it out. imagine if i had 1,000 devices. that’s why i don’t use azure functions. and because i used node.js, i was able to port over the code to my gateway in about 2 hours. i’ve spent, i’d say, over 7 full days dealing with microsoft’s billing & tech support (the former needs the latter in these complicated situations). i am not going to waste any more time on it. tech problems on a VPS are easier to handle than an accounting issue which is based on microbilling resulting from the number of transactions and god knows how many log files and where.
from a cost perspective, accountants aren’t any cheaper than techies. tech problems, once solved, tend to not recur. accounting problems do. and there is no stackexchange for billing that everyone can consult.
and, once again, the microsoft people are super nice and try to be helpful. google? good luck trying to get to a real person. from a business perspective:
microsoft: has always been focused on creating value for the small-to-mid sized business customer. i think, as a startup, we’re too small. real people available, paid support starts at about $25 and you don’t even really need that. forums are good. decent api.
ibm: only useful for the big clients. you can get to a real person. api is harder than azure’s. they lost out on the PC revolution, they’re going to lose out on the IoT one, too. and they still keep going and going and going… pretty amazing.
google. their focus hasn’t been business, it’s been about selling ads to consumers and on the side make geeky, fun, cool things. and it shows. you can’t get to a real person easily. their api and doc is godawful and out-of-date.
aws. didn’t try it. several colleagues advised me to stay away because it’s highly complicated and can get expensive very, very easily. that’s it’s more fun to get a tooth pulled without anaesthesia.
digital ocean. haven’t tried but have heard good things about them and their virtual servers seem to be very nicely priced.
mongo. i can’t really find a use for it because azure tables works well enough. besides, mongo, in the past, seem to have had problems with losing data.
if anything else comes to mind, i’ll post.
does anyone have any recommendations for gateway/linux server management? i’m playing with balena and it’s quite good. dataplicity is unreliable, i think it only works well when you have a good router – something beyond my control. remote.it is not much more reliable either. both dataplicity and remote.it have an intersection of unreliability and then other situations where one works and the other doesn’t. balena is pretty rock-solid thus far. and they’re designed, it seems, from the ground up to manage server groups. i’m still playing with it…
hey,
I do not have any recommendations but I played with Balena in the past and looked pretty cool.
I think I know what you mean here. GCP, in general, may not be as well documented as Firebase.
Firebase is a (small?) product in the whole GCP that may have seen great success as an app backend and they made the docs and site very cute and flashy. You can also use Firebase Firestore and its cloud functions for plenty of other things, like what you describe.
This is what one of my customers is paying for a product that I just finished implementing:
Those 52 cents (on the firebase blaze plan) include around 5 Particle devices reporting data to the cloud every minute. The data is stored in Firestore Firebase. This data is occasionally queried by iOS, Android and web apps.
They provide Cloud functions which are similar to Azure functions.
So far so good but I just started using it 6 months ago.
Free plan vs Blaze plan: If your cloud functions or infrastructure make calls to external sites (like Particle's cloud for authentication purposes) you need to pay for the Blaze plan since the free plan does not allow external calls.
How does this all look?
Gustavo.
EDIT: this is using the Google Cloud Platform from Particle
It certainly looks interesting. Do you know if the functions can be node.js? Is there a charge for file writes & reads? This was the gotcha on azure. And, truth be told, as we only need about a day’s worth of data to be stored, I’m just tired of researching, evaluating and testing their products for them. For the amount of time it takes to make a build vs buy decision, I might as well build, but then we’ll have a custom, and let’s face it, poorly documented system to maintain that’ll bite us in the ass 2y from now when we need more features. But I still don’t want to do their work for free.
Anyone had a good experience with couchdb? Also, I looked at the gcp/grafana/prometheus integration example. It’s like a horror movie without the entertainment factor. So, the what I get from a cloud provider are backup, failover, clear documentation and keeping up with the times (software infrastructure).
I’m still thinking the core (IoT architecture, provisioning, analysis, routing, twinning and remote management mqtt/https data feed, functions and analysis) should be custom. Storage on someone’s cloud. 3rd parties for dashboard and alerts. Other items, too, can be partitioned and moved to the cloud, if appropriate.
Mainly, the cloud has gotten more in the way of getting things done than being useful. It HAS been useful in helping us understand how we should architect the solution, the ux and user functionality. But I think we’ll take it over from here.
