Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

Disney+ is arguably the largest software release of all time in terms of scale. My estimate is that they will be handling 1-2% of all internet traffic by the end of the week.

I can't think of a launch of this scale that has gone smoothly. Modern console releases, AAA games with online multiplayer, Pokemon Go, healthcare.gov... all cases where organizations literally have billions in cash. It does not matter how much money you have or how great your engineering team is - you can't go scale 0 to 100 on launch day.



OK - so if it is so obvious that you can't scale from 0 to 100 on launch day, why not anticipate that?

Ticket sales platforms for major sports events and concerts don't try to do it - they plan for it by having customers queue.

Did Disney not attempt this?


I've never handled a project like this, but I feel like the goal should be to meet what average demand will be once the high demand of the first few days goes down. No one is going to remember that they couldn't login to Disney+ on the first day as long as the experience on the second and going forward is strong.


Plus, its free press. "Disney+ so popular at launch it was hard to get into for a couple hours." Meanwhile, as people are getting home from work and reading that headline, and Disney+ now works.


"You loved us so much you broke us."


OK - so if it is so obvious that you can't scale from 0 to 100 on launch day, why not anticipate that?

I imagine the gigantic Arctic cold front passing over the East Coast right now created a much larger demand than they were expecting on a non-holiday Tuesday morning while school is (supposed to be) in session.


To be exact they didn't go from 0 to 100 though as it was already live in the Netherlands for a month.


So 5 to 100, instead of adding on one or two small countries a day as their infrastructure allowed.


The olympics and world cups do that every year just fine.


But their usage pattern and infra can build on past experience, whereas D+ has none of that. Black swans galore.


BamTech has the experience, and they built the platform that Disney bought. It’s not a new team


Their past experience of running Hulu and ESPN?


> I can't think of a launch of this scale

Well that's the problem isn't it. They chose a scale they couldn't handle ergo, they fucked up.

A more sane approach would be to start by launching in smaller countries and then slowly scale up.


Agreed. One of the reasons why maybe they didn’t that is because someone thought “we gotta get out those big series everything is expecting like Mandalorian out, and we can’t release it first on Denmark and not the US”, which someone probably replied “why don’t we launch in Denmark without any big names” which got laughed at because then none would want to sign up for it.

The tension between Marketing and Engineering makes another victim.

I believe Google specifically fight against this by mandating every project to have an insane scalable architecture from the get go. You can’t be Google and fail at that when you sell your infra to corporations promising that won’t be a problem. Someone at Disney+ probably made a calculated bet that if they can solve the problem soon, none will care about this in a few weeks.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: