Soliciting downtime patching architecture

yes, but not on todays maintenance.

[quote=ā€œCowvenent-area-52, post:14, topic:1704104ā€]
A replay log would make that fairly trivial (again, depending on backwards compatibility), but I’m going to bet they’re not doing anything append-only like Cassandra.[/quote]

That’s what I am thinking. As I wrote above, remember the big Cata and SL prepatch issues, I am pretty sure they were introduced by switching to a more modern DB infrastructure. People ā€œrewoundā€ days of effort, people lost gear, etc…

sad cat noises

Hrm… I’m trying to think back that far and failing lol. But if that did happen, that could easily have been snapshots though. Not to say it couldn’t have been replay logs or anything, just that seems more likely to me for that time.

Cataclysm was in 2010, which was pretty much infancy (or before) for NoSQL, if you mean that. I don’t know if I could see a good use case for it, based on what I’d guess would be in their model, but I’ve mostly focused on the types of models that fit nicely into denormalized data – logs, time-based series, etc. I’m sure these days there’d be one that’d fit, especially because you want the results to be consistent across reads, whereas you’re not always guaranteed that with most partition-tolerant types I’ve worked with.

I know some of the words being said here.

Before NoSQL, if I am not mistaken. I think Aerospike was late 2010. I wanted to (badly, it’s midnight here) simply point out, that it seems Blizzard was updating its data store technologies during major patches. It would make sense. Why remain in the world of 2004’s DB nightmares?

That said, DB isn’t the only reason a maintenance takes long. Swap out hardware, run other maintenance tasks, recheck, etc. If it takes a day, it takes a day. I find the weekly downtimes much more annoying and much less explicable.

Relational database?? You mean one big Excel spreadsheet or maybe its an Access database?

(All my DBA brothers out there just got REAL mad)

1 Like

Yes, super mad. Because the only database that matters is the JETSAM Key Store System in Mallard Basic on any CP/M model. If it was good enough for NASA in the 80s it should be good enough for Blizzard in 2023.

It’s not a 20+ year old system. The engine is upgraded every single expansion. And it’s modular in design, using something they refer to as ā€œWorld of Warcraft stack.ā€

I’m not saying the maintenance times need changed, but this fallacy of the engine being old and outdated needs addressed.

Most of what is being discussed in this thread has nothing to do with the game engine.

At one point in time Blizzard was hiring Oracle 10 admins. Oracle 10 came out 20 years ago.

Duct tape.

I quoted your OP concerning the ā€œ20+ year old system,ā€ I didn’t quote anything else in the thread, nor address ā€œmost of what is being discussed.ā€ Just your OP.

That changes nothing. We are talking about maintenance which encompasses more than just the game engine.

… you do realize that the maintenance being done is using the engine, right? It’s not some separate entity where you ignore the engine’s existence and update the game. That’s where the modular design comes into play.

I have no clue what you’re trying to accomplish with this dismissive conversation. I gave you factual information that has to do with the engine, thus the maintenance. That’s it, that’s all.

You misspelled ā€˜speculative misinformation’

Please go read and learn before you open your mouth again. Blizz has been telling us for ages that the engine is updated every single expansion. And yes, the modular design was mentioned in an interview as being called ā€œWorld of Warcraft stack.ā€

Nothing about that is misinformation. Not even responding to your blatant desire to act obtuse again.

Yea, I think Cassandra was still in dev at that time, might have been alpha, even. But yea, probably upgrading Oracle 8 to 9i or something lol.

Yea, it could definitely be other stuff, as you mention. I’ve not worked with the actual physical server-side items before, only logical/network up to about the load balancer (and even then, barely there). Though, I never recall things like that needing to be done that often or taking that long, but I’ve definitely dealt with things that had multiple redundant servers when it was on prem like that. Or maybe our server techs were just that good that I never noticed haha.

So… shall we pick this thread back up and try to help Blizz out now that the maintenance has been extended three times? :laughing:

1 Like

Looks like Sendryn solved it all with snarky replies about the engine. So we can’t talk about any other part of the maintenance, sorry.

1 Like

It turns out this entire time they were just waiting for Warcraft Rumble to load on their phones

1 Like
1 Like