On 23/08/12 5:41pm, Malcolm Edgar wrote:
Is this the correct approach. The reason I am using Cayenne 3.0.x, is so I
can perform testing against our applications, which I think is a good way
of verifying this code.
I don't recall seeing much change in the bits of code you are touching from 3.0
to the current trunk. But Andrus will know.
I can't really comment about how they implement ACID, there is the issue of
latency and the speed of light! I think their use of MVCC helps with this
type of problem, which then becomes an application problem.
I watched their video. Very interesting. Like Andrus, I also don't see how they
can accomplish ACID and have asynchronous communication between nodes. Seems
like a layer of magic is missing somewhere, but I guess we'll all know more
when they release their product and specs.
The applications I develop don't have a lot of contention, my main concern
is about availability. I haven't figured out how to do automatic fail over
yet, beyond loosing a connection to a Transaction Engine (TE) and then
going back to a Broker to establish a new connection to a (TE). My concern
is if you loose a TE and Broker, then I have to write the code to reconnect
to another Broker.
For mysql you just use this in the JDBC connection
jdbc:mysql:replication
http://dev.mysql.com/doc/refman/5.1/en/connector-j-reference-replication-connection.html
But of course, mysql master-master replication sucks. And promoting a slave to
master requires scripting outside of mysql itself. At least the failover part
in JDBC is easy.
Cheers
Ari
--
-------------------------->
Aristedes Maniatis
GPG fingerprint CBFB 84B4 738D 4E87 5E5C 5EFA EF6A 7D2E 3E49 102A
