维度表包含有关事实的描述性信息。
A dimension table contains descriptive information about facts.
表级谓词通常也存在于一些维度表中。
Table-level predicates typically exist on some dimension tables as well.
在创建维度表时，需要记住如下问题。
Here are a few things to remember when creating dimension tables.
Time维度表包含10,000行。
维度表可能不包含事实，而只包含元数据。
在设计维度表时，需要计划变更维度数据。
As you're designing your dimension table, you need to plan for changing dimensional data.
Customer维度表包含10,000行。
Location维度表包含10,000行。
第三个方法是，在捕获旧值的维度表中增加字段。
The third method is to add a field to the dimension table that captures old values.
查询至少在一个维度表上包含一个或更多的过滤器。
The query includes one or more filters on at least one of the dimension tables.
日期维度表。
时间维度表。
维度模型中的事实表与其他所有维度表之间存在单一连接。
The fact table in the dimensional model has a single join with all the other dimension tables.
这个查询对每个事实表和维度表之间的连接使用散列连接。
This query plan uses hash join for each join between the fact table and a dimension table.
维度表的层次与它需要连接到中心事实表的连接数目有关。
The level of a dimension table corresponds to number of links it needs to connect to the central fact table.
每个散列连接都是右深查询，构建的表是一个小的维度表。
It's a right-deep query plan such that for each hash join, the build table is a small dimension table.
大型维度表可能产生资源密集的Cartesian连接。
Large dimension tables can produce resource intensive Cartesian joins.
性能的提高取决于连接键被下推时所在维度表的过滤选择度。
The performance improvement depends on the selectivity of the filter on the dimension table from which the join keys are pushed down.
我们已经讨论过事实表，因此本专栏将关注如何设计维度表。
We have already covered fact tables, so in this column we will look at how to design dimension tables.
维度建模包括三个关键部分：事实表、维度表、以及汇总表。
A dimensional model has three key parts: the fact tables, the dimension tables, and the summary tables.
维度表需要作为主键数据的通用键；这将作为事实表的外键。
A dimension table needs a generalized key as the primary key data; this becomes the foreign key in the fact table.
雪花式维度的优势之一是，因为规范化了维度表而实现了空间节省。
One advantage of the snowflake dimension is the space savings, because you normalize your dimension tables.
当连接执行时，合并选中的索引，此时维度表中的任何列都可使用。
When a join is executed that incorporates an index that is selective, any column in the dimension table can be used.
一般说来，维度表应该被非规范化和扁平化，并根据需要使用大量列。
Generally speaking, a dimension table should be denormalized and flat, with as many columns as needed.
我们已经讨论过事实表和维度表，因此本专栏将关注如何设计汇总表。
We have already covered fact and dimension tables, so in this column, we will look at how to design summary tables.
您所要面对的最重要的设计决策之一，是如何捕获维度表的层次结构信息。
One of the most important design decisions you face is how to capture hierarchy information in a dimension table.
由于每个维度表过滤的选择度，每次连接都会减少事实表大小的十分之一。
Each join reduces the size of the fact table by one-tenth due to selectivity of the table filter on each dimension table.
有些维度表可能非常大，而且是DesignFilters的候选项。
Some dimension tables can be quite large and are good candidates for Design Filters.
事实表和任何维度表之间的连接都是等值连接，这使得散列连接成为可能。
The join between the fact table and any dimension table is an equi-join, which makes a hash join possible.
所有的扫描过滤器都应用后，每个维度表选出的行数少于事实表选出的行数。
The number of rows selected from each dimension table is less than the number of rows selected from the fact table after all scan filters are applied.
应用推荐