Wednesday, November 20, 2013

Part 1: Writing Websites

This is the first in a series (I imagine) about issues in creating websites today.
Just random observations or brain dumps of what I is going on.

And of course the big news at the moment is...
It's been all of the news for a while now, since its launch at the beginning of last month.
So I've been laughing about it over all of that time, on news especially the Daily Show.

In summary, I don't think it's a big deal and I don't care a lot about the issues with the website. Well, other than entertainment.
Sure, I can see how these websites problems can be annoying and look bad, but to me it just looks like the news networks are bored and found some fresh meat to pounce on.
And with the latest comparisons of this website rollout to Katrina?! Stop. It's a website. Not a failed mission to start a space colony on Mars.

But in reality, my understanding is that 80% of the country doesn't need the website. I don't. My employer healthcare is good enough. Do I expect that the remainder of the country without health insurance (actual citizens looking for help) jumped on the website to sign up? Or do I expect that news network employees hammered the site?
Apparently there were long wait times on the rollout, it couldn't handle the load, slow at even 1K users but was expected to work for 50K and had up to 250K.

Sadly, I can tell you what I expect those numbers are not. They are not young people 20-30, especially those without a dependable job or one providing health insurance. Basically those the mandate is trying to push into having insurance. Those struggling with college debt and confused about this healthcare law. Sitting on the sidelines, waiting to be pushed or forced into action, waiting out the website storm in the news. The ones that really need a single payer health care system, because it means having to do nothing and higher taxes sound more positive than a tax penalty.
I'm just glad the mandate deadline to have health insurance has been extended farther into 2014.

Back to the site, I had a glance at it before, but tonight I was curious to see what I could gather at a longer look.
Well the page itself uses a bunch of fairly recent UI JS libraries like Bootstrap, JQuery, Backbone, Underscore. But it also loads almost 60 JS files including tons of JQuery plugins. Not that these are the cause of specific issues, I have no problem with them.
More interesting was the CreateSaml2 POST with a 200 Success but response text of "success: no". Awesome.
It gives two different feelings for the front end and back end, which makes sense given the history. Apparently these were contracted out to two different companies and then subcontracted.
I can imagine the mess involved with this project, and I'm sure it's worse than even that.
But the only lesson I get is that the government can't build websites. Should I be surprised?

I'm a dev, I know that bugs happen, so I am not at all surprised when they do.
I also have been staring at different HTTP error codes for years, since I began using the internet.
So websites are not a mystical beast.
But they can be complicated to develop, especially large scale.
Sounds like this project had changing requirements and no clear direction or project management.
And of course not enough testing, there can never be enough.

Demo Bug
With not enough testing, you can guarantee issues.
I sat through a horrible meeting today to demo work I've been loosely involved in.
And I had the pleasure of sitting through this uncomfortable and embrarrasing meeting,
seeing an error message in red up on the screen saying 500 NullPointerException.
Which just makes everyone involved look bad.
But that's how it goes sometimes. It shouldn't happen but it does.
Thankfully this isn't rolled out anywhere yet, and of course it wouldn't be like it is.

Signup Nightmare
To the extreme of website issues,
What if the individual mandate for health insurance was only a day, or even less than a day?
What if the country only had a limit of 100K slots for health insurance on Oct 1 to signup and get it super cheap?
What would that have looked like? A complete mess, yes.
It would be all over the news, and on that day the website would be swamped and would go down.
This is basically what the PAX 2014 signup was like.
Registration, especially getting hotel rooms was an online battle, not just between people but also servers. It was insane. It was sold out in less than an hour. Systems were down, and of course everyone just kept trying F5 with errors and partial page loads and trying to call in but systems were not available there either.

So building dependable websites to handle large load can be hard. Bottlenecks happen.
Not that I know any of the intricate details of these systems (I'd love to),
But I would imagine that knowing an advertised live date and number of users, with (god I so hate the word) "cloud" servers and stateless designs the flexibility exists to handle a lot of load. With Amazon, Google, whatever, scale up a ton on that first day, monitor it and adjust. At least be prepared for the onslaught that you welcomed with open arms.

No comments:

Post a Comment