tag:blogger.com,1999:blog-8436330762136344379.post7240075643958675421..comments2024-03-18T20:09:27.497-04:00Comments on Metadata: Spanner: Google's Globally-Distributed DatabaseMurathttp://www.blogger.com/profile/07842046940394980130noreply@blogger.comBlogger1125tag:blogger.com,1999:blog-8436330762136344379.post-7234323738446451512013-07-18T20:46:43.054-04:002013-07-18T20:46:43.054-04:00I wondered the same thing when I read the Spanner ...I wondered the same thing when I read the Spanner paper. My conclusion was that it simplifies externally consistent reads (which is related, now that I think about it). Consider Process 1 commits a transaction, sends a message to Process 2, which then reads the results of that transaction. The Spanner implementation guarantees that Process 2 will see the results of Process 1's write, without needing to do anything like communicate version vectors or other "state" like some protocols require.<br /><br />However, its been a while since I read the paper, so this could be inaccurate. :) Thanks for the detailed summary.Evan Joneshttps://www.blogger.com/profile/04769274706597594656noreply@blogger.com