Keith Adams: can transactions apply across tables?
Mike Franklin: Is the temporary merged _tx_ object permanent?
Bob English: what if client crashes between mask of original objects and the
super object is written? (1. and 2. slide 9)
John: Masks are leased?
Marcos Aguilera: Periodically write heartbeats to t.
Finkelstein: how many messages are involved in doing all of these client side
steps?
Marcos Aguilera: atomic writes, but not serializability
Finkelstein: yes, you do have isolation
John: concerned with counts of RPC RTTs for client & 2pc methods
Marcos Aguilera: wondering whether we actually need transactions: do we expect
multiple servers to contend?
will applications commonly be doing write sharing? read sharing
should be common, but writes?
Mike Franklin: Will there be write-write conflicts?
Keith Adams: Would like the feature to handle write-write conflicts.
Confusion of masks being on write side of transactions only.
Marcos Aguilera: Arguing for transactions - e.g. facebook wall concurrent accesses.
John: Question - NoSQL systems not doing transactions across multiple objects.
So, do we need this feature?
John: Question - Alternative transaction methods (i.e. not mini-transactions)?
John: Question - What's least functionality in servers so users can get what
they want with best possible performance?
Keith Adams: Some people really hate specific instances of eventual consistency
(e.g. "message read/unread" bit not taking effect in email app)
Bob English: On relaxed consistency: Transition between 0 and 1 the big deal.
People don't care about 4 vs. 5 (e.g. do I have any new emails vs
do I have 4 or 5 new emails?)