This is an automated email from the ASF dual-hosted git repository.
xxyu pushed a commit to branch doc5.0
in repository https://gitbox.apache.org/repos/asf/kylin.git
The following commit(s) were added to refs/heads/doc5.0 by this push:
new e8b9bdf068 Minor remove useless image
e8b9bdf068 is described below
commit e8b9bdf068714ff6d1633a75f20ab97cec8aa371
Author: Mukvin <boyboys...@163.com>
AuthorDate: Fri Nov 25 11:16:20 2022 +0800
Minor remove useless image
---
website/docs/modeling/data_modeling.md | 5 +----
1 file changed, 1 insertion(+), 4 deletions(-)
diff --git a/website/docs/modeling/data_modeling.md
b/website/docs/modeling/data_modeling.md
index 955c0368fe..a36d76c58d 100755
--- a/website/docs/modeling/data_modeling.md
+++ b/website/docs/modeling/data_modeling.md
@@ -24,10 +24,7 @@ If it takes 1 minute to query 100 million entries of data
records, querying 10 b
### Accelerate query with Kylin pre-computation
-Kylin leverages pre-computation to avoid the computing pressure brought by the
growing data volume. That is, Kylin will precompute the combinations of defined
model dimensions and then store the aggregated results as indexes to shorten
query latency. In addition, Kylin uses parallel computing and columnar storage
techniques to improve computing and storage speed.
-
-
-
+Kylin leverages pre-computation to avoid the computing pressure brought by the
growing data volume. That is, Kylin will precompute the combinations of defined
model dimensions and then store the aggregated results as indexes to shorten
query latency. In addition, Kylin uses parallel computing and columnar storage
techniques to improve computing and storage speed.
With pre-computation, the number of indexes will be determined by the
dimension cardinality only, and will no longer undergo exponential growth as
data volume increases. Taking the data analysis of online transactions as an
example, with Kylin pre-computation, even if the volume of transaction data
increases by 10 times, the query speed against the same analytical dimensions
changes little. The computing time complexity can be kept at O(1), helping
enterprises to analyze data more efficiently.
