Transactional solutions for microservices
summary
Argues that saga pattern gives transactional guarantees with eventual consistency rather than two‑phase commit, and maps that to the MicroProfile LRA spec and two real implementations: an LRA coordinator built on Narayana (HTTP headers carrying saga IDs, file or SQL storage) and a Dynflow executor (Ruby workflow producing execution plans and compensating reverts). Shows orchestration vs choreography, explains steps/plan/run/finalize lifecycle, notes demos failed but both systems support undo by calling service compensate endpoints and trade strict atomicity for survivable distributed transactions.
AI-written summary · about
chapters (11)
- 0:00Introduction to the topic
- 1:26Transaction management basics
- 3:50Transactions in microservices
- 6:39The saga pattern explained
- 10:15Approaches to saga design
- 13:54Implementation frameworks
- 20:38Handling failures and recovery
- 24:19Developer API and examples
- 30:35Advanced error handling
- 37:00Building blocks and summary
- 41:44Q&A session
related talks
31mAddressing the transaction challenge in a cloud-native world
39mA different flavor of the distributed transaction
17mSaga Pattern for Distributed Transactions
53mA different flavor of the distributed transaction
41mTransactions in Your Micro-Services Architecture
55mTackling transactions in Microservices applications
46mLost in transaction? Strategies to deal with (in-)consistency in distributed systems
46mAddressing the transaction challenge in a cloud-native world