After architecting a distributed system from the beginning to be testable in a network simulator in accelerated time, you will feel completely exposed doing anything less. This approach can be millions of times faster than Jepsen, achieving a super tight development feedback loophttps://twitter.com/cmeik/status/1120499798258475009 …
-
Show this thread
-
Replying to @sadisticsystems
making time just another event in our
@ApacheSpark event-sourcing system was the smartest thing i've ever done.1 reply 0 retweets 6 likes -
Could you say more? That sounds very interesting!
1 reply 0 retweets 1 like -
we were doing analytics, with rolling buckets of time. naively we would have just let this all run off the all clock, implicitly, but instead, we created a source (a stream) for time, injecting the new "minute" in.
1 reply 0 retweets 2 likes -
Replying to @rektide @yoshuawuyts and
enabled two things: a. if one of our data producers was running slow, we could delay our new minute, and the analytics wouldn't run. b. if the analytics engine had an outage or issue, we could stop time, then inject minutes very quickly. also useful for performance simulation.
1 reply 0 retweets 2 likes
Heh, interesting! Thanks for sharing! I was wondering what the granularity there would be; I could see this be problematic if there was an event every millisecond haha xD But what you're saying seems useful and cool!
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.