Geek asks and answers

Development of a database of timers in Kassandra

I am considering creating a Cassandra timers database to store millions of daily daily data that could potentially have up to 100B data points.

I looked at this article: http://rubyscale.com/blog/2011/03/06/basic-time-series-with-cassandra/

This design is very durable. Therefore, in essence, I can put daily timestamps in the form of columns and, if necessary, outline the columns by adding a day to the row.

I have two questions:

  • I look at storing up to 20,000 temporary tables (daily). Is it even worth it to outline the lines, for example. year with this number of columns? Are there any advantages / disadvantages to planing rows to reduce the number of columns to 365 per year?
  • Another idea that I have is, instead of outlining columns row by row, is to create a family of columns for each year. Thus, when accessing data for several years, I would have to query several column families, rather than one column family, and join the results on the client side. Will this approach speed up the process or rather slow down everything?
+5
source share
2 answers

If you are ever going to manage a huge number of records, there is one problem with your approach.

1 , node. node , Cassandra, . , , /.

, , . bucketing , .

( ) :

  • [ROW_BASE_NAME] + [] + someHashFunction ( )% 10
  • [ROW_BASE_NAME] + [DAY] + random.nextInt(10)
  • [ROW_BASE_NAME] + [DAY] + nextbucket < --- , .

. . , , .

/, multi_get ( ) .

+4

More articles:


All Articles