In Postgres-XC, GTM assigns sequence value to all the transactions in its cluster. XC is a kind of tightly-coupled distributed system. In a loosely-coupled distributed system, where each database is autonomous, we may need another mechanism.
I've learned that logical replication (used to be bi-directional replication) people are doing this kind of work.
Can anyone point me to some reading material on how auto-generated sequence primary keys are handled on distributed systems? I think the advice used to be to use GUIDs, but I thought I read somewhere that PostgreSQL now assigns a pool of numbers to each node when a sequence is implemented. I have searched the PostgreSQL 9.1.5 Documentation, but apparently my search terms are not quite what it takes, or dreamt that up.
PostgreSQL itself does not support a distributed architecture. You may be thinking of Postgres-XC?
Sequences are local to each instances and it is not a pool, it is a 64bit allocation for each sequence within the local node, generally constrained only when called from the serial (big serial being 64 bits) type to 32 bits.
Sincerely,
Joshua D. Drake
Thanks, Melvin
-- Command Prompt, Inc. - http://www.commandprompt.com/509-416-6579 PostgreSQL Support, Training, Professional Services and Development High Availability, Oracle Conversion, Postgres-XC, @cmdpromptinc For my dreams of your image that blossoms a rose in the deeps of my heart. - W.B. Yeats