revisiting the cap theorem
DESCRIPTION
As an assignment I had to present a follow-up paper to the CAP theorem: CAP Twelve Years Later: How the "Rules" Have Changed. I didn't really like the paper and tried to justify that ;-)TRANSCRIPT
data
Service Access Points
Data-Centric Services
Distributed Service
data
1 accept
2 don’t
availability consistencyof service of data
enter CAP
Eric Brewer
also
SA
C A
PEventually Consistent
Sorry We’re Closed
Harry Potter Was Here
Bad Idea ?
But, what is a Partition ?
cap.key
5 minutes remaining
From ‘draft’ to ‘final’
Wait Cancel
(*) Microsoft chooses Consistency
(*)
PNUTS
• P are rare!
• C+A until P!
• detect P explicitly!
• handle P explicitly
• limit operations!
• user interface!
• higher level of abstraction
• delayed intensions !
• logging, merging, …!
• compensate for mistakes