Basic Assignments
 
Options & Settings
 
Main Time Information
Color Code: Yellow
Assigned To: Brandon Moore
Created By: Brandon Moore
Created Date/Time: 9/24/2019 10:57 am
 
Action Status: Blank (new)
Show On The Web: Yes - (public)
Priority: 0
 
Time Id: 4988
Template/Type: Brandon Time
Title/Caption: Server talks with Wayne
Start Date/Time: 10/8/2019 10:00 am
End Date/Time: 10/8/2019 11:00 am
Main Status: Active

Sorry, no photos available for this element of time.


Notes:

As we were getting into the meeting, we were just talking about management styles and what not. I pushed over a link to a small article from the Harvard Business Review from back in May 2010. The article is called "Bringing Out the Best in Your People". Here is the link: https://hbr.org/2010/05/managing-yourself-bringing-out-the-best-in-your-people

With AWS, Wayne is having the biggest struggle accessing images and copying session variables into local variables. Tons of different includes, paths, crazy look and feel values, etc. Any time we link to the images differently, it becomes a problem. We are still having problems with logos, alternate logos, watermarks, etc. Wayne is looking at creating a corporation object and then trying to map to that vs all of the normal session stuff.

Small talk about the progression of the system from early 2000's to 2013, 2015 (big influx of new developers), 2016-2017 snow owl theme settings, and so on. We also talked about our custom code options and how the black boxes played into the mix.

Wayne is leaning towards a single corp object. Alan is thinking along those same lines. They were talking about removing the page logic and making it exist much higher on the page. In a way, we have a ton of code that exists and then it gets used differently on the pages. We are needing a way to standardizing that flow and process. As these guys are searching and replacing certain variables... it isn't uncommon to get 1,600 different pages that need to be changed and what not.

Question from Steve - If we stay with Newtek, do we need these same changes or are these new changes just for moving things out to AWS? Lots of talk about localizing the variables and helping to standardize things. Being consistent and making sure that those pieces exists, no mater what.

One of the problems is switching between reports that are based on the secure login and other reports that don't need to be secured, but we still want the whole look and feel to play across. That gets challenging.

Talking about built-in clustering for ColdFusion and what offerings can we still do and achieve at Newtek? Steve had some questions there. We have three different environments... we have older physical dedicated boxes, we have new VPS (virtual private servers), and AWS clustering. We are thinking that we need focus more on the new possibilities with the VPS servers and possible clustering at Newtek. We put some rough percentages out there... this just a starting place. We are thinking 75% of our server efforts on the new VPS and ColdFusion Enterprise clustering. We are then thinking about a 20% effort out in the AWS land (that has been pretty challenging), and then maybe doing 5% efforts on the older physical dedicated servers.

Steve is seeing a new need for dedicated aggregated servers and aggregated systems. That will change our model and may alter some of our focus. Wayne was talking about some load balancing and even doing load testing. Using a scripting program and software to add some load and pound on our servers a little bit more. He provided a link to the Locust software package.

https://locust.io/ - program to automate some server beat-up drills (simulate loads)

Wayne wanted to know, what is our kryptonite (what makes us go slow) - We said, looking up items, tons and tons of invoices, pulling big reports like the P&L, balance sheet, sub inventory, history reports, and tons and tons of transactional data reporting. Steve was saying that we could copy over some data from some of our bigger clients and then use that background data to virtually beat-up the system using real data. Steve is going to check with a couple of clients to get permission to use their data to do the server beat-ups. The real data won't ever be exposed, just used to simulate server loads.

Alan and Wayne were talking about using the load testing options and then looking at other factors such as splitting up the databases, moving the actual databases to their own servers, keeping them on the same servers on different drives, etc. Lots of testing and putting equal automated loads to see what the best configuration might be. Good discussion.