CODEWITHSUNDEEP
distributed-transactions
maki
maki

Reputation: 621

what are the essential differences between SAGA and TCC for the distributed transaction problem?

I am reading the SAGA pattern to solve the distributed transaction problem. Although the names of the concepts in SAGA are different from those in TCC(try-commit-cancel), I don't find essential different ideas between the 2 patterns.

Both TCC and SAGA rely on the local transaction, try/retry/cancel/commit of the local transaction to guarantee the whole transaction. Could somebody explain the key differences that differentiate the 2 patterns and make both models valuable?

I lean SAGA from this document:

https://learn.microsoft.com/en-us/azure/architecture/reference-architectures/saga/saga

Upvotes: 1

Views: 2223

Answers (2)

yedf
yedf

Reputation: 186

The key differences is the data visibility. If you use Saga to do a cross-bank transfer, then the client may see unexpected balance change if the transfer failed. But TCC can avoid that.

It is explained in detail in this article:

https://dev.to/yedf2/best-practice-for-tcc-distributed-transaction-in-go-402m

Upvotes: 2

maki
maki

Reputation: 621

I think TCC is one special kind of SAGA Orchestration.

Upvotes: 0

Related Questions