Bakery algorithm is really interesting -- AFAIK it's the only mutex
algorithm that doesn't assume an existing atomic operation like CAS.
On Mon, Sep 13, 2010 at 11:46 AM, Thorsten von Eicken
> I've been musing about how to implement locks using just cassandra, i.e.
> without sql db or zookeeper. I wrote up what I've come up with on the wiki
> (it's a bit too long for an email) at
>. I'm wondering whether I've
> overlooked something, especially I'm not yet comfortable enough with the
> semantics of failing cassandra write operations to figure out whether these
> can cause problems, so I suspect some further iterations on the algorithm
> will be necessary.
Project Chair, Apache Cassandra
co-founder of Riptano, the source for professional Cassandra support