Today I saw Google BigTable ppt (2005-10-18:jeff Dean gave a talk at the University of Washington about Big table-their sys TEM for storing large amounts of data in a semi-structured manner), feel nothing new
1. A method of horizontally dividing table common parallel database
2. A cell is multidimensional information, which is actually a snapshot of information at multiple points in time
3. Family column and general column are stored separately because the columns of the family column are mutable (?).
4. The way to locate the tablet, this is great I think it's kind of like a Linux multilevel index
5. Most of them are distributed database methods such as lock service, master, etc.
6. Compression I don't know, there's nothing to say
There's a place that doesn't understand when a-machine goes-down, the master redistributes its log chunks to another machines to process (and these machine s store the processed results locally). The machines that pick up the tablets then query the master for machine for their data. Recently acquired tablet) and then go directly to the the location of the processed results (to update their
If one is down, master will distribute his log. Master does not have the Log,master how to distribute. Are all the logs put together. But the PPT said is a machine (can have multiple tablets) a log.