Hi Patrik,
Thanks for your reply! I think I'll go with implementing the dynamic
solution for my project: we're fine with causal eventual consistency, and
we think being available during intra-DC netsplits will be beneficial.
Another reason I didn't mention is that we need a particular schema
Hi Igor,
There are several reasons, such as:
- we believe that it's useful to have the strong consistency with
single-writer within one DC, and "only" have the asynchronous replication
between DCs
- failures and partitions within one DC can be handled by SBR, fits the
model of
Hello Patrik,
excellent news!
May I ask why you chose to have static topology over dynamic one? With the
latter one, I mean lifting the "only one writer" requirement locally as
well, thus enabling always-available operation --- even during netsplit,
where you could have multiple writers.