@jan And, it looks like they're doing both...?
"Because of the replication lag we mentioned earlier, however, you might not see the change you just made! This experience is very confusing for a user and also leads to double posting. We got around this concern by setting a cookie in your browser with the current time whenever you write something to our databases. The load balancer also looks for that cookie and, if it notices that you wrote something within 20 seconds, will unconditionally send you to California." Though, instead of always choosing California over Virgina, I'd store whether or not they should hit A or B, I guess. Anyways, thanks for all the pointers! -ZZ On Tue, Feb 24, 2009 at 10:11 AM, Jan Lehnardt <[email protected]> wrote: > > On 24 Feb 2009, at 17:03, Zachary Zolton wrote: > >> Thanks for the reply! It looks like they go into the more advanced >> Bayou consistency, and Byzantine failure modes, but I don't think I'll >> need to cover that soon... >> >> But a more important question: >> >> If I have two couch servers: A and B >> >> And, I want to load-balance users between them, would it be the >> responsibility of the web/app servers to ensure that a user session >> "sticks" to either A or B, after performing a write/update? At least >> until the data has had a chance to replicate between servers...? >> >> (I'm guessing this is what all the "monotonic updates discussion is >> about...?) > > That would be one solution*, yes. Another would be to employ a write-through > memcache. > > * http://www.facebook.com/note.php?note_id=23844338919&id=9445547199&index=0 > > Cheers > > Jan > -- > > >> >> On Tue, Feb 24, 2009 at 9:54 AM, Jan Lehnardt <[email protected]> wrote: >>> >>> On 24 Feb 2009, at 16:49, Zachary Zolton wrote: >>> >>>> As a developer (without an advanced degree :^P) trying to understand >>>> Eventual Consistency, I happened upon these slides: >>>> >>>> http://www.cs.berkeley.edu/~istoica/classes/cs268/06/notes/20-BFTx2.pdf >>>> >>>> I know consistency models are a hot topic around here, so I thought >>>> I'd ask if this would make a good introductory text for me to explain >>>> the techniques to some colleagues of mine. Or does anyone take >>>> theoretical issue with it contents? >>> >>> I skimmed the contents and it looks cool to me for an introduction. >>> >>> Cheers >>> Jan >>> -- >>> >>> >> > >
