Nathan,
I certainly don't mind using a web server (e.g., IIS or Apache) as a reverse proxy for routing to node apps on the IBM i. I actually see a lot of advantages to this approach.
Yet, three things keep sticking in my mind with a reverse proxy approach. These things probably represent my newbie understanding more than they represent real problems. But, hey, this is how I learn...
1. What happens when the reverse proxy goes down (either because of an error or for maintenance)? If users access all of my node apps through a reverse proxy, then all of my node apps become unavailable when the reverse proxy goes down. I have never set up multiple web servers for fail-over or load balancing. Would setting up web servers for fail-over and load balancing become any more difficult if the web server is serving primarily as a reverse proxy for a lot of node apps?
2. What happens when I have a web server that functions as a reverse proxy for, say, 150 node apps--and now I want to add or modify a single reverse proxy route in the web server configuration file? Is my only option to recycle the entire web server to start using the modified configuration file? Would such a recycle be graceful for users? Or would I have to tell all users to get out of the node apps so I can recycle the web server?
3. What about all of those port numbers? If I use a reverse proxy to route requests to each of my node apps on the IBM i, then each of my node apps on the IBM i requires its own port number. If I have 500 node apps, I'll have to have them listen on 500 different port numbers. As the number of node apps grows, so does the number of port numbers used. Do I really want to have hundreds, or perhaps even thousands, of port numbers being listened to? Does this create system performance, security, or maintenance issues?
Thanks,
Kelly Cookson
IT Project Leader
Dot Foods, Inc.
1.217.773.4486 ext. 12676
kcookson@xxxxxxxxxxxx
-----Original Message-----
From: WEB400 [mailto:web400-bounces@xxxxxxxxxxxx] On Behalf Of Nathan Andelin
Sent: Tuesday, October 13, 2015 3:25 PM
To: Web Enabling the IBM i (AS/400 and iSeries)
Subject: Re: [WEB400] Hosting a Large Number of Node Apps on the IBM i
Kelly,
It's nice to see questions which are relevant to the deployment of broadly scoped systems. And I agree that the answer always seems to come back to "routing".
I'm not sure how to answer your questions about options 1 - 3, so I'll ask more questions which might provoke thought.
What would be the best tool for handling:
1. Request logging?
2. Response logging?
3. Gzip compression?
4. TLS encryption?
5. Serving static content?
One benchmark I came across showed IIS performing 2.5 times faster than Node.js while serving static content, presumably due to caching. Apache offers similar caching. Wait for a Node.js caching module? Build one yourself?
Node.js experts seem to advise the avoidance of any computationally intensive work within the main event loop. Would "computationally intensive" include the evaluation of "routing" expressions?
On Tue, Oct 13, 2015 at 12:32 PM, Kelly Cookson <KCookson@xxxxxxxxxxxx>
wrote:
If our shop adopts node, we would probably use it to develop web and
mobile apps instead of 5250 green screens (for future development).
Over the years, we have accumulated a large number of 5250 green
screen interactive programs. If we adopt node, I can imagine we will
accumulate a similarly large number of node apps.
The way to host a large number of node apps is, according to my Google
searches, routing. But there are different strategies for routing.
1. Use a reverse proxy to route requests to node apps, where each node
app has its own URI and port number. The reverse proxy could be a web
server or something like node-http-proxy. Redbird also appears to be a
package for setting up a reverse proxy for node apps (
https://github.com/OptimalBits/redbird?utm_source=nodeweekly&utm_mediu
m=email
).
2. Use ExpressJS vhost to route requests to node apps, where each node
app has its own URI and port number.
3. Use ExpressJS routers and module.exports to route requests to node
sub-apps or modules (something like this: https://vimeo.com/56166857).
So, I am left with a few questions. What strategy or combination of
routing strategies would be best for ensuring high availability of
node apps? What strategy or combination of routing strategies would be
best for scaling node apps (e.g., using the cluster module)? Is
reducing the number of unique ports being listened to by node apps a
good reason for using the sub-apps strategy (option #3 above)?
Thanks,
Kelly Cookson
--
This is the Web Enabling the IBM i (AS/400 and iSeries) (WEB400) mailing list To post a message email: WEB400@xxxxxxxxxxxx To subscribe, unsubscribe, or change list options,
visit:
http://lists.midrange.com/mailman/listinfo/web400
or email: WEB400-request@xxxxxxxxxxxx
Before posting, please take a moment to review the archives at
http://archive.midrange.com/web400.
midrange.com
