Documents of products collection are intentionally designed to be more complex and larger than accounts - I want to see what happens, what is the performance penalty mainly, once individual documents are stored on multiple database pages. In Postgres, page size is 8 KB by default - in practice, the goal is to have at least 4 rows stored on a single page, so every record that is larger than 2 KB is put on two or more disk pages. It obviously reduces performance for both writes & reads - more disk pages to read from and write to. In Mongo it works slightly differently in details, but essentially in the same vein - larger documents are stored on more than one page, degrading performance for all operations. In both cases we are about to see - how much exactly.
Марк Эйдельштейн привлек внимание иностранных журналистов на модном показе14:58
，更多细节参见必应排名_Bing SEO_先做后付
�@�}4�ɋ������̂́AZoom�̌ڋq�x�[�X�̌��B�ڂ������̂�20.4���ЂƂ����ڋq�А��ƁA���Z�@�ւ��a�@�A�������ƁA���w�Ƃ������e�����̃g�b�v�@�ւɂ����č����̗p�����������邱�Ƃ��B
Interesting! FIND ACCOUNTS BY OWNERS case is pretty much the same and the Elephant has won there - except the fact that documents of accounts are much smaller than of products. The delivered throughput performance difference is substantial - 7170 QPS vs 3624 QPS, 1.98x win for MongoDB, but latency is lower for Postgres.