This year has been truly amazing for me! I had the opportunity to speak at the LAOUC (Latin America Oracle Users Group Community) in Mexico, Costa Rica, Guatemala, Panama, and Brazil. My session covered a topic that brings together two powerful areas: vectorized data replication and distributed AI processing with Oracle GoldenGate (OGG).
I delivered the talk in Spanish, which was a big challenge for me, but I’m writing this post in English to broaden the reach and share the knowledge with more people. If this topic sparks your interest, keep reading—you won’t regret it!
So, here’s a quick summary of my presentation, and let’s start by celebrating! ?
30 Years of Continuous Evolution
Oracle GoldenGate has just reached its 30th anniversary, three decades of continuous improvement and innovation. Since Oracle acquired GoldenGate in 2009, the product has become even more robust, establishing itself as the leading solution for real-time data replication across industries worldwide.
Ok, but… Why Oracle GoldenGate?
GoldenGate is not only the market leader, but also the tool that continues to innovate and adapt to technological changes. It is a mature and comprehensive solution, enabling full traceability of all events that occur during replication. Most importantly, it ensures data integrity with no data loss, and aren’t data the most valuable asset of any company?
OGG 23ai: The Ultimate Microservices Architecture
The 23ai release marks a turning point with the consolidation of the microservices architecture. The question is not if you will migrate, but when… Additionally, Oracle GoldenGate 23ai offers a new, more intuitive, and optimised interface, making daily administration simpler and more efficient.
Use cases with vector data
I also demonstrated two real-world use cases of heterogeneous replication between Oracle and PostgreSQL:
1. “AS IS” replication: replicating vectorized data without any code changes.
EXTRACT E_GGBR01
USERIDALIAS ogg_connect_oracle
EXTTRAIL g1
TABLE app01.src_docs_01;
REPLICAT R_GGBR01
USERIDALIAS ogg_connect_pg
MAP app01.src_docs_01, TARGET app01.trg_docs_asis_01;
2. Transforming into vectors: converting structured data into vector data directly at the target, leveraging the capabilities of Oracle Database 23ai.
EXTRACT E_GGBR02
USERIDALIAS ogg_connect_pg
EXTTRAIL g2
TABLE app01.src_docs_02;
REPLICAT R_GGBR02
USERIDALIAS ogg_connect_ora
MAP app01.src_docs_02, TARGET app01.trg_docs_02,
COLMAP (USEDEFAULTS, EMBEDDING = @DBFUNCTION('ADMIN.GET_EMBEDDING_OGG(:a2)'));
CREATE OR REPLACE FUNCTION GET_EMBEDDING_OGG (p_text IN VARCHAR2) RETURN VECTOR
IS
v_embedding_ogg_vector VECTOR;
BEGIN
-- Generate embedding based on the input columns
SELECT VECTOR_EMBEDDING ( ALL_MINILM_L12_V2 USING p_text as data ) as data
INTO v_embedding_ogg_vector
FROM DUAL;
RETURN v_embedding_ogg_vector;
END GET_EMBEDDING_OGG;
These scenarios highlight how GoldenGate is ready not only to handle traditional data, but also to support new data formats that enable generative and distributed AI applications.
The full presentation is available for download here: