While a bit dated and aging now, you might enjoy reading "A Fundamental Note On Distributed Computing" and perhaps some of the original papers on Jini.
I also have two books on Distributed Algorithms and Distributed Systems that I recommend.
I'm not convinced that Distributed Transactions are necessarily impossible, just that they have a significant timing problem. The first few chapters of Jim Grey's "Transactions" book were very influential on my thinking.
(Oversummarizing: transactions are possible when any operation can be reversed, and when you can track in a realistic way whether your aggregated operation was a success or a failure. I think one can do that, to a limited degree, in highly scalable and distributed systems, but the time-window when things do not have a coherent state or worldview can be quite long, and determining whether all the actors have completed or returned to the original state can be difficult. It CAN be known, but it cannot ALWAYS be known.)
no subject
I also have two books on Distributed Algorithms and Distributed Systems that I recommend.
I'm not convinced that Distributed Transactions are necessarily impossible, just that they have a significant timing problem. The first few chapters of Jim Grey's "Transactions" book were very influential on my thinking.
(Oversummarizing: transactions are possible when any operation can be reversed, and when you can track in a realistic way whether your aggregated operation was a success or a failure. I think one can do that, to a limited degree, in highly scalable and distributed systems, but the time-window when things do not have a coherent state or worldview can be quite long, and determining whether all the actors have completed or returned to the original state can be difficult. It CAN be known, but it cannot ALWAYS be known.)