Aloe Discussion

[scott_boss]

[quote=371614:@Dirk Cleenwerck]Would like to know if this is something we need to consider.
If there are no real benefits, I would prefer not to reprogram over 90 API calls.[/quote]

when i first read your message I read 9 API calls. and was like sure, migrate that over… until I re-read the message on the full-sized screen (laptop) and was like err… wait a second… Migrating 90 API calls is a massive chore. Not even in the same ballpark for level of effort.

@371627:@Tim Dietrich

In general, I think you’ll find that AE is simply easier to use and much more flexible than Luna (and the original version of Aloe as well). You get convenient access to all aspects of a request, as well as complete control over the response.

this and

@371627:@Tim Dietrich

The biggest technical benefit of Aloe Express (AE) over Luna is that AE uses a multi-threaded model for processing requests. As a result, you get non-blocking I/O, which can be extremely important depending on the nature of your API and the volume and frequency of the requests that it is handling.

this…

are my two biggest reasons to go with Aloe.Express. IF I had my API fully fleshed out and running in production today under Luna or Aloe, would I change it? nope. But I would look at migrating to Aloe.Express for the next generation of the API.

the REST API I was writing under WE+Aloe, I could overrun with a single Postman[1] instance running through the API. And my issue was the startup cost of the session and processing it just took long enough I was dropping ~5-7% of my calls when I had low latency between Postman and the API server. When I was at places like coffee shops, cafes, airports, hotels, I was running ~10-23% drop rate (same laptop running Postman). I mitigated the drop rates by running 3+ instances of the WE+Aloe app on the host with HAProxy in front of it load balancing across all the instances. That got my drop rate down to (near) 0%/1% respectfully. I say near zero as occasionally I will drop a single API call. Now I am going to deploy Aloe.Express the same way I am doing Aloe (HAProxy for load balancing & SSL termination, 3+ Aloe.Express nodes on the backend - all on a single host)

Now the performance increase, latency in the API server, etc is reduced with Aloe.Express. I don’t have hard numbers for you as I have not ported the API calls over to Aloe.Express yet. I have been playing with some mock up test API to do some basic testing. Now my work with Aloe.Express for non-API server work has proven to me the speed and flexibility of it.