1、和四个伙伴在pgvector采用IVFFlat和HNSW算法产生分歧，然后又讨论到了是否使用Azure AI Search，最后一番拳脚交流后大家，确定使用Azure CosmosDB for PostgreSQL 的HNSW-ANN，主要原因是Azure AI Search也采用的HNSW作为主流算法，并且在特定情况下支持KNN，最重要的是IVFFLat需要牺牲性能。。。。没钱
2、 @azgy 完成管道的数据隔离和用户数据标识，以及数据删除。
3、@zsbgnw12 (最帅不过你我)完成弹性伸缩和接口安全测试
4、 @zshandong 完成Azure设施部署和最终测试
5、由我完成基础设施部署
最后祝大家玩的开心。
1、After some disagreements among four partners regarding the use of IVFFlat and HNSW algorithms in pgvector, we discussed whether to use Azure AI Search. After a heated exchange, we decided to use HNSW-ANN in Azure CosmosDB for PostgreSQL. The main reason is that Azure AI Search also adopts HNSW as a mainstream algorithm and supports KNN in specific scenarios. Most importantly, IVFFlat sacrifices performance... and we are short on budget.
2、@azgy completed data isolation in the pipeline, user data identification, and data deletion.
3、@zsbgnw12 (the most handsome of us all) completed elastic scaling and interface security testing.
4、@zshandong completed Azure infrastructure deployment and final testing.
I completed the infrastructure deployment.
