Suggestion for HackerNoon story for those involved in the migration

Yesterday our VP of tech did a vlog with Bret Fisher, and the interview was fairly wide ranging, but at this point: https://youtu.be/_i4gZDJS_9g?t=2652 he gets asked the question of how to migrate a system with 50+ microservices to Docker. He talks about dividing and conquering, not doing a big bang migration, and limiting the blast radius.

I think given our experience with the HackerNoon migration there are some good lessons here. There may have been a way to start the migration with just new articles going onto the new system and old links still working, or even just a fraction of new articles going onto the new live system until kinks were worked out. The big bang migration attempt has probably caused people to have some late nights and based on threads in this community has caused some writers a bit of consternation.

This is exactly the type of lesson that could be useful for the readership, and I’d encourage someone inside the migration to write this story and talk about how they’d do it differently if they could do it again. I personally can’t write it because I don’t have enough details on the decision process and the available migration options.

Food for thought!
Looking forward to smooth seas ahead.
–Jon

2 Likes

I told David a similar thing

Maybe this should be a dedicated thread where eveyone with some skills can advice - because being outside the box is easier to catch mistakes

Hey @jonxtensen & @arthur.tkachenko we started a longer form post on this :slight_smile: There is a lot to learn from our experience, but all migrations will have different motivations, pressures and time constraints. You may be able to guess ours. The big bang approach certainly caused some things to break, some late nights for the team, and worst of all, some stories not to load and others to not load quite right, but in our case I still think it was the right approach to get our independent infrastructure in good standing with the web.

Would also add this from an earlier post: Even though we’ve published far less stories than a normal week - as the team as been deep in QA, support , The Noonies, and the next random problem that surfaces from a migration - traffic is up. Here’s our GA daily traffic since May 1:

More encouraging are the early returns on why traffic is up. Page load times have been cut in half, page views per session have more than doubled, and our bounce rate has dropped by over 50% (the effect of no account creation or paywall pop up ad). We’re headed in the right direction, but there are gaps in our software we have to keep improving on.

2 Likes

There is two main things and one smaller one

  1. Startup approach - do quick, we’ll handle that. Or “Think where hockey puck will be, not where it’s right now”. I totally adore.

  2. Stress for team - not adore(Don’t worry I did the same mistake few times before - I don’t like the consequences)

  3. Coding parts (solvable - great team now onboard) / Project Management also can be solved. If there a “Wish” to do it.

Spike in traffic shows that Medium started to cut HN down. It maybe looks like conspiracy, but… :blush:

Again - all is cool. waiting when biggest wave will gone and will start to publish my articles that is all over this forum

2 Likes