f*****e
发帖数: 2992 | 1 【 以下文字转载自 DataSciences 讨论区 】
发信人: fatalme (don't ever give it up), 信区: DataSciences
标 题: 这个cassandra paging的解决方案怎么样？
发信站: BBS 未名空间站 (Fri Sep 26 15:52:43 2014, 美东)
As paging is a big issue of Cassandra. I have a strategy to page using
secondary index: Add a column "page_number" corresponding to each column you
want to sort and page, create an index on the "page_number" column or just
use it as the clustering column. The "page_number" is a good candidate for
indexing, since it has low cardinality: every entry in the same page has the
same page number. If there are 100 entries in one page, then there are 100
entries with the same page number. Thus is low cardinality. We can order the
entries according to each column and set the corresponding "page_number".
You can retrieve one page and do the corresponding sorting and send to
display. Is this a good way or not? |
g*****g
发帖数: 34805 | 2 Not a good idea if you ever delete any entry. |
f*****e
发帖数: 2992 | 3 就是不删啊，就是显示webmetrics history的。都是history了，不会增删改。如果经
常改动该怎么做？我们公司有个senior说secondary index is deprecated。我估计他
是妒忌我的想法。娃哈哈哈。 |
w**z
发帖数: 8232 | 4 you only need to store the starting column in the index cf since C* columns
are stored continuesly.
you
just
the
100
the
【在 f*****e 的大作中提到】
: 就是不删啊，就是显示webmetrics history的。都是history了，不会增删改。如果经
: 常改动该怎么做？我们公司有个senior说secondary index is deprecated。我估计他
: 是妒忌我的想法。娃哈哈哈。
|
