Twitter | Search | |
Ben Johnson
Every time I hear someone say they need five 9's of uptime, I think of GitHub. Everybody relies on them, they have maybe 99.9% uptime, and they got bought for $7.5B by Microsoft. Unless you're making pacemakers, ease up on your crazy uptime requirements.
Reply Retweet Like More
Ben Johnson Feb 25
Replying to @benbjohnson
P.S. I don't mean that as any slight to GitHub. I think they do a great job. Services go down sometimes—life goes on.
Reply Retweet Like
Jan Decavele Feb 25
Replying to @benbjohnson @tvlooy
I'm pretty sure them being bought by Microsoft hasn't got anything to do with uptime and all the more with them being a central hub in the FOSS community Microsoft desperately wants to get a foothold in...
Reply Retweet Like
Ben Johnson Feb 25
Replying to @EpochDC @tvlooy
Yeah, I agree. My point is that high uptime is not necessarily a hard business goal (assuming the goal was acquisition).
Reply Retweet Like
Richard Tibbetts Feb 25
Replying to @benbjohnson
My favorite fault tolerance scheme ever was a day trading app where on failure of the primary system the backup system just closed all open positions. “If something bad happened the last thing I want is a computer trying to recover automatically.”
Reply Retweet Like
Ben Johnson Feb 25
Replying to @tibbetts
I agree. Even just switching to a read-only mode that runs off a replica is probably a decent strategy since most apps are read heavy.
Reply Retweet Like
Jimmy Zelinskie Feb 25
Replying to @benbjohnson
The vast majority of programming is spent working on code locally before it ever needs to interact with a git server, thus the server doesn't need to be available most of the time. Requirements are always based on the domain, not perceived importance or popularity.
Reply Retweet Like
Ben Johnson Feb 25
Replying to @jimmyzelinskie
I find that I use GitHub mainly as a project management tool & CI system more than a git server. It's definitely a domain specific concern—some areas have hard uptime requirements for sure.
Reply Retweet Like
Pinboard Feb 25
Replying to @benbjohnson @arclight
Five nines of uptime is easy, if there can be other numbers between the nines
Reply Retweet Like
Nate Vack Feb 25
Even easier if they can go in the front
Reply Retweet Like
on error resume next Feb 25
Replying to @benbjohnson
Think a bit more about how/who a GitHub outage affects and what they do. It's a rare case.
Reply Retweet Like
Ben Johnson Feb 25
Replying to @FlorianHeigl1
They provide a project management & CI platform. That seems pretty critical to most development groups. The git hosting seems like a minor part of their offering.
Reply Retweet Like