Frustrations with Bluemix. What alternatives for Node-Red Cloud Service?

[Alan UK]

Hi



I was attracted to Bluemix as it allows integrated Node-Red flows both locally and in the cloud, but after spending a day getting a simple flow running across the ether, I wondering if there are other Node-Red cloud alternatives. A web search indicates only AWS but I've read that requires Node-Red to be installed and then maintained.

My frustrations with Bluemix are:

1. fragmented and complex menus and dashboards/consoles and need to define and keep track of organisations/regions/apps/services/devices/users etc and link them all together across the various dashboards. Plus the web interface is chunky, inconsistent and throws up errors seemingly because my region is UK and not US South! It took me a hour to find out how to launch the Node-Red web page :(

2. In the IBM Watson dashboard building a Card to display my app's log data but not being able to display as a list of records more than <timestamp> and one other field <status>. My Cloudant database contains <timestamp><status><status description> and other fields may be added. Plus the Card only showed 5 records and no scroll facility to see older records.

3. I looked at the Cloudant control panel and defined a view but could see no easy way to query and display the data. On-line tutorials were suggesting I need to transfer and convert the data to a dashDB database and then use Embeddable Reporting Service - 2 more services to install, learn and probably pay for, plus more fragmentation.

4. I looked at the API (URL) but that needs query strings added to the URL and the result is a JSON datastream, still needing lots of work to display as formatted table.

5. It equally looked cumbersome to simply delete old records (e.g. needing a query to extract keys older than a specified date and then an app to issue delete commands). Oh for something like MySQL's phpAdmin!


Maybe something like ThingSpeak is better to store sensor data and status logs, run some analytics, have some alert rules and send some action commands to my Raspberry pi Node_Red app.

PS I was going to add a couple of images but they always got insert at the start of this post text and I couldn't drag then to the right place in the text.

[Julian Knight]

The docs now include several I think. I believe some people are successfully running both on AWS and Azure. However, I'm not clear on what you really want to achieve so it is hard to comment.

[Alan UK]

On Thursday, 26 January 2017 21:02:04 UTC, Julian Knight wrote:

The docs now include several I think. I believe some people are successfully running both on AWS and Azure. However, I'm not clear on what you really want to achieve so it is hard to comment.

Thanks Julian for reading and replying and for mentioning Azure. I will look at it. What I'm evaluating is how my Node-Red app is distributed between client and cloud. This is in terms of rules and actions, display of data and actions taken, data logging and data analysis.

I've decided for now to concentrate on my app and put all the rules and actions and dashboard output in the app (i.e. client). When it's written I will look again to see if I can find any cloud based solution and if need be go for just storing log data and device data in the cloud using an MQTT broker.

PS I can remember the debate many years ago as to what processing should be off-loaded from mainframe to mini-computers and PCs. It seems like we now have the reverse decision: how much client processing and storage to off-load to the cloud. Obviously, organisations with 1,000 client devices, then centralisation allows overall analysis and control as well as data resilience.

From a security point of view I think it is better to remotely interrogate a cloud repository and set actions there, than for me to open up my router for inbound interrogation, analysis and control. This is what I have with my solar monitoring system (not Node-Red and supplied and run by the manufacturer and no APIs :(  ).

[Julian Knight]

Hi Alan, yes I'm also from the mainframe generation :-)

I think you can also do MQTT in Azure. I think that MQTT might form a core part of what you are trying to do since it is designed to allow consistent transmission of data and allows message sync so that you could, if you wanted to, have local msg hubs that could be used locally but also pass messages on to a more central hub - maybe in the cloud. So you could potentially have the best of both worlds.

Of course, one of the variables is the cost. You need to balance the cost of cloud vs local as well as the security.

[Lawrence Griffiths]

Quite a few of us FRED fred.sensetecnic.com 

   

On Monday, 23 January 2017 21:37:20 UTC, Alan UK wrote:

[Dave C-J]

As good a time as any to plug this... Upcoming talk on Node-RED... And FRED.

https://developer.ibm.com/open/events/dw-open-tech-talk-node-red-update/

[Alan UK]

Thanks guys

On Friday, 27 January 2017 23:15:46 UTC, Julian Knight wrote:

... that you could, if you wanted to, have local msg hubs that could be used locally but also pass messages on to a more central hub - maybe in the cloud.

Good idea. I could have a cron job that FTPs my data every day to my ISP, so at worst I only lose a day's data if my Raspberry PI went belly up.

Lawrence Griffiths   wrote on 28 Jan

Quite a few of us FRED fred.sensetecnic.com

I will look again at the Free Plan. The 72 hours / month = 6 mins per hour, if that is processing time message handling then that should be OK given the Rasberry Pi is doing all the heavy work, but may not be enough if it includes (say) time logged onto a dashboard. Worth another look.

Dave C-J     wrote on 28 Jan

As good a time as any to plug this... Upcoming talk on Node-RED... And FRED.
https://developer.ibm.com/open/events/dw-open-tech-talk-node-red-update/

Great, It's in the diary

Alan

[Mike Blackstock]

Thanks for the FRED plug guys.

Just to be clear, the free plan runs your flows for 72 hours continuously.  If you log in during that time, the 72 hour clock restarts.  If you don't, FRED stops your instances to free up resources.  You can restart it next time you login.  This is our current approach to balancing our need to 'keep the lights on' while we manage the cost of running Node-RED for many users on our servers for free as a community service.

If you sign up for a paid plan, we run your flows continuously, and you can make more complex flows with more nodes.

Like Node-RED, FRED is evolving, and we're looking for feedback on features, supporting infrastructure, nodes, and even our business model, so feel free to message me directly.

Mike

[Julian Knight]

Of course, Node-RED is "just" a Node.JS application. You can run that on pretty much anything. So if you want, you could run it on a cheap VPS. I can recommend Bluehost in the UK as one of the cheapest but who still give very good support. I have 2 VPS's with them and I've rarely had any issues at all.

[Peter Scargill]

Fred is very good - nice to see it now has BigTimer as  plug-in option :-) 

[Mike Blackstock]

Thank you for creating that node Peter and for your helpful blogs.  We had requests to make your node available :-) Mike

--
http://nodered.org
 
Join us on Slack to continue the conversation: http://nodered.org/slack
---
You received this message because you are subscribed to a topic in the Google Groups "Node-RED" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/node-red/tJqDlMjb7vI/unsubscribe.
To unsubscribe from this group and all its topics, send an email to node-red+unsubscribe@googlegroups.com.
To post to this group, send email to node...@googlegroups.com.
Visit this group at https://groups.google.com/group/node-red.

To view this discussion on the web, visit https://groups.google.com/d/msgid/node-red/7780fdb1-baef-4a80-b7a9-ba2a6853b2f4%40googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

[Alan UK]

On Tuesday, 31 January 2017 08:55:04 UTC, Julian Knight wrote:

Of course, Node-RED is "just" a Node.JS application. You can run that on pretty much anything. So if you want, you could run it on a cheap VPS. I can recommend Bluehost in the UK as one of the cheapest but who still give very good support. I have 2 VPS's with them and I've rarely had any issues at all.

Thanks Julian. I can install and run software with my existing ISP plan but a self install means self maintenance. Until recently I was running a CMS for a community group and it was a pain to monitor all the software announcements and work out whether I needed to update the software to plug a security hole. So a hosted service seems better provided the cost is reasonable for a home application.

Now I've "downsized" my use of ISP services I might be looking to change ISP so will bear Bluehost in mind.



Alan

[Julian Knight]

I get that Alan! It's a pain right now having to update a families worth of laptops, router, NAS, 2 VPS's with websites, etc. Seems like a full time job on its own sometimes! Certainly not trying to put you off the excellent FRED, just another option.

[Aidas Janulis]

But still missing your Timeout @Peter :) I found very useful. Of course, I can substitute it using standard node-red nodes and functions. But, Fred also miss IFTTT node and again - I can substitute it, but should I? If guys are trying to make money on us, maybe they can do a bit more than it is now.

[Peter Scargill]

Hi Aidas

When you say "still missing your timeout pete"...

Erm  - it's there.

https://flows.nodered.org/node/node-red-contrib-timeout

[Mike Blackstock]

Please let us know if there is a node missing that you would like to see in FRED.  Our team can be reached at in...@sensetecnic.com.  Sometimes we can install a node that we're unsure about on your instance to test it out before making it generally available.

Mike

On Mon, Oct 30, 2017 at 10:29 AM, Aidas Janulis <januli...@gmail.com> wrote:

But still missing your Timeout @Peter :) I found very useful. Of course, I can substitute it using standard node-red nodes and functions. But, Fred also miss IFTTT node and again - I can substitute it, but should I? If guys are trying to make money on us, maybe they can do a bit more than it is now.

--

http://nodered.org
 
Join us on Slack to continue the conversation: http://nodered.org/slack
---
You received this message because you are subscribed to a topic in the Google Groups "Node-RED" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/node-red/tJqDlMjb7vI/unsubscribe.
To unsubscribe from this group and all its topics, send an email to node-red+unsubscribe@googlegroups.com.
To post to this group, send email to node...@googlegroups.com.
Visit this group at https://groups.google.com/group/node-red.

To view this discussion on the web, visit https://groups.google.com/d/msgid/node-red/74b842cc-9c66-40c5-ad5f-318c6712e794%40googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

--

Mike Blackstock
Chief Technical Officer
Sense Tecnic Systems, Inc.

308 East 5th Avenue, Vancouver, BC, V5T1H4
www.sensetecnic.com

phone: 604-760-9035 email: mblac...@sensetecnic.com

[Peter Scargill]

Mike

I just sent an email - and received this

5.1.2 The recipient address <in...@sensetecnic.com.> is not a valid RFC-5321

So - here's the content of the email...

AH I realise what he meant about missing timeout - he meant - missing in FRED.....   suggest you add this to FRED  - you already have my BIGTIMER node...

npm install node-red-contrib-timeout

Also

npm install node-red-contrib-diode

Regards

Pete.

On Monday, 30 October 2017 18:44:40 UTC, Mike Blackstock wrote:

Please let us know if there is a node missing that you would like to see in FRED.  Our team can be reached at in...@sensetecnic.com.  Sometimes we can install a node that we're unsure about on your instance to test it out before making it generally available.

Mike

On Mon, Oct 30, 2017 at 10:29 AM, Aidas Janulis <januli...@gmail.com> wrote:

But still missing your Timeout @Peter :) I found very useful. Of course, I can substitute it using standard node-red nodes and functions. But, Fred also miss IFTTT node and again - I can substitute it, but should I? If guys are trying to make money on us, maybe they can do a bit more than it is now.

--
http://nodered.org
 
Join us on Slack to continue the conversation: http://nodered.org/slack
---
You received this message because you are subscribed to a topic in the Google Groups "Node-RED" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/node-red/tJqDlMjb7vI/unsubscribe.

To unsubscribe from this group and all its topics, send an email to node-red+u...@googlegroups.com.

To post to this group, send email to node...@googlegroups.com.
Visit this group at https://groups.google.com/group/node-red.

To view this discussion on the web, visit https://groups.google.com/d/msgid/node-red/74b842cc-9c66-40c5-ad5f-318c6712e794%40googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

[Dave C-J]

Though can't the default trigger node be configured to do the same ?

--

Sent from phone.

[Mike Blackstock]

Ah... - thanks Peter - we'll look into installing those, and check our email server!

Mike

[Peter Scargill]

If you're referring to node-red-contrib-timeout (sorry if you're talking about something else and I'm completely off-base) - I don't think so as timeout can trigger an alert warning at some arbitrary count toward the end - so three messages - and I have some ideas for a lot more in the future - major changes along those lines I would imagine would not be par of the roadmap for the default trigger mode. Plus timeout is prettier :-)

[Dave C-J]

True,  default trigger doesn't have warning - but for a simple watchdog it's not needed.

Timeout can also only send simple string messages and not messages of any type, and I'm not sure I'd call shocking pink prettier than the pastel shades... but hey. Of course it has it's place but may not be needed unless you need that extra warning.

[Peter Scargill]

Ok, I'll hand you the colour argument - it wasn't QUITE the colour I had in mind but I've just never gotten around to making it more subtle.

So - I'm not sure who suggested that timeout was supposed to be a simple watchdog. There are many examples of needing a warning in which case indeed the injector would do - let me give an example... A raspberry Pi is controlling a device that is supposed to be sending out a watchdog signal.  If it does not, eventually the Pi will perhaps reset it by removing it's power temporarily and then restoring it. Well, that could be done with the warning - it doesn't have to be a warning - maybe I should have called them action 1 and action 2.  Indeed now you mention it, I could add another... 

Alternatively there may be an external function which cycles the power on receipt of the timeout - in which case the warning message may be used to send a signal to tell the other pi to start a shutdown, giving it ample time to do so before removing the power.  Indeed anyone following my UPS project right now will know that I'm actively into this - having lost data more than once due to abruptly turning a Pi off without shutting down first. There are many examples of needing this - and that is why I made this node in the first place.

Now I know you guys put the trigger node together and it has a special place in your hearts  - so please know that I also use trigger nodes ALL OVER THE PLACE.. utterly essential - but I can't change that, it is probably more or less set in stone whereas I can improve timeout in response to user requests just as I keep improving BigTimer in response to user requests - always retaining backward compatibility.   As for only sending text, that is not a limit - because in many cases people are sending text to MQTT brokers or similar and you can do a lot of things with that. 

:-)

Pete.