This has be a real blind spot in the node community since inception. It's great to see a company focused on delivering a node runtime that is instrumented in a way that provides large node deployments with the necessary metrics/insights needed for node at scale.
Would be great if they posted some more info on whether or not their instrumentation comes with a cost to performance. Looks like you can enable/disable profiling at runtime without restarting too, which is huge!
> Node.js applications can be debugged and profiled live, without the need to restart, and therefore potentially miss critical state that led to the fault.
This looks beautifully designed, and I would love to test out how it profiles an existing app or two. Beyond the landing page, are there any types of docs that are accessible for leveraging your software? Hopefully it should be pretty self explanatory, but I'd love to see a set of docs as well.
We agree. We've focused heavily on the User Experience from day zero. We see no reason that consumer software should be a more pleasant experience than enterprise/business software. It should just be "software" and it should be well designed!
Here's the scenario: I want to instrument my production system using free and/or self-built tools. So I take a crack at it, and discover it's poorly documented and otherwise hard to find good information.
Next step: I start researching and asking the community. I'm quickly directed to "N|Solid," but it's too expensive. So I keep working and eventually cobble together enough knowledge to be passably competent in the area.
I'd like to contribute my expertise back to Node.js in the form of documentation and code improvements, so other people can easily do it for free, but I seem to be hitting a brick wall. Oh, it turns out lots of core Node contributors work for NodeSource / N|Solid, and have zero interest in making this aspect of Node.js easier for people to do for free.
I suppose it could just be coincidence that the io.js fork didn't just disable, but did non-trivial work to remove facilities that allowed for competing (and totally open source) profiling & debugging tooling.
Certainly casts a "community governance" spat in a bit of a new light when suddenly the major movers behind the fork turn out to have a significant financial interest in it.
Heroku also provides a bunch of monitoring "add-ons" compatible with Node.js.
I guess this solution is different because you don't have to modify your code to enable monitoring.
I'm curious how they intend to crack the Java stranglehold. I guess they intend to build on existing developer mindshare, and make sure ops can support the latest flavour of the month.
21 comments
[ 43.1 ms ] story [ 599 ms ] threadWould be great if they posted some more info on whether or not their instrumentation comes with a cost to performance. Looks like you can enable/disable profiling at runtime without restarting too, which is huge!
> Node.js applications can be debugged and profiled live, without the need to restart, and therefore potentially miss critical state that led to the fault.
You can quite literally click a button to profile you app in realtime...no code modification required.
https://cloudup.com/cNdDqLq_1sM
https://cloudup.com/cNdDqLq_1sM
http://hub.docker.com/r/nodesource/jessie
These guys appear to be on the ball, usually a cursory investigation would lead me nowhere but they've got a healthy presence in all the right places.
The linked DockerHub repository does point to a page with an example Dockerfile using node-0.12, but they do have tags for 4.0 and 4.1. Fascinating.
node app.js
nsolid app.js
same thing
I agree, this is something that the node ecosystem really needed. Strongloop was working on something similar; we'll see how that goes with IBM.
I'm really excited to see something that looks viable actually shipping.
Also, no Windows support on an "Enterprise" product?
Here's the scenario: I want to instrument my production system using free and/or self-built tools. So I take a crack at it, and discover it's poorly documented and otherwise hard to find good information.
Next step: I start researching and asking the community. I'm quickly directed to "N|Solid," but it's too expensive. So I keep working and eventually cobble together enough knowledge to be passably competent in the area.
I'd like to contribute my expertise back to Node.js in the form of documentation and code improvements, so other people can easily do it for free, but I seem to be hitting a brick wall. Oh, it turns out lots of core Node contributors work for NodeSource / N|Solid, and have zero interest in making this aspect of Node.js easier for people to do for free.
Maybe I'm just being paranoid?
Certainly casts a "community governance" spat in a bit of a new light when suddenly the major movers behind the fork turn out to have a significant financial interest in it.
- New Relic (http://newrelic.com): a well established player.
- Keymetrics I/O (https://keymetrics.io/): makes use of PM2, which is OSS. Seems to have decent plans for small dev shops.
- Nodetime (http://www.appdynamics.com/nodejs/): another option...
Heroku also provides a bunch of monitoring "add-ons" compatible with Node.js.
I guess this solution is different because you don't have to modify your code to enable monitoring.
I'm curious how they intend to crack the Java stranglehold. I guess they intend to build on existing developer mindshare, and make sure ops can support the latest flavour of the month.