The paper Eventually-Serializable Data Services (PODC'96) presented an eventually-serializable data service.

In the abstract, it says:

… and generalizes their algorithm (in the related work Lazy replication) by allowing general operations …

In section 1.1 (Background of our Work), it says:

… Also, their algorithm (in the related work Lazy replication) requires that all operations be either read-only queries or write-only updates …

My questions is as follows:

Is the protocol in this paper suitable for transactions (as a kind of general operations) services?


In section 3, the authors give an abstract algorithm for implementing eventually-serializable data service. The algorithm is based on lazy replication, in which the operations received by each replica are gossiped in the background. The key property is that the operations requested in a partial order will be eventually total ordered through the gossip protocol. I want to extend this eventually-serializable notion to support transactions. Does this algorithm itself support transactions already? And what are the challenges to support transactions instead of ordinary operations?


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.