担心立方体尺寸输出?
我第一次学习立方体,是读一本厚厚的红书。团队有一个担忧:
在创建 SSAS 多维数据集时,我们是否应该担心多维数据集查询结果的潜在最大大小输出?换句话说;假设用户能够组合多种度量,那么结果数据库查询是否可能会产生数 GB 的数据? (我们的db大约是10GB)
进一步; 可以采取哪些措施来确保没有任何一个用户能够以这种方式劫持数据库并确保为所有用户提供流畅的服务?
我们现有的立方体解决方案基于 Mondrian 和 JPivot Web 前端。这展示了巨大的输出问题。我们的新多维数据集将使用 SSAS 2008,但仍需在 Web 方面做出决定。
(在StackOverflow搜索类似问题失败)
I am learning about cubes for the first time, reading a thick red book. The team have a concern:
When creating a SSAS cube, should we be worried about the potential max size output from cube query results? In other words; Given that the users are able to combine several measures, is it so that the resulting database query might yield several GB of data? (Our db is about 10 GB)
Further;
What actions can be undertaken to ensure no single user can hijack the db in this way and ensure that smooth service continues to all users?
Our existing cube solution is based on Mondrian with a JPivot web front. This exhibits the huge output problem. Our new cubes will use SSAS 2008, still to decide on web front.
(Searching for a similar question failed at StackOverflow)
如果你对这篇内容有疑问,欢迎到本站社区发帖提问 参与讨论,获取更多帮助,或者扫码二维码加入 Web 技术交流群。
绑定邮箱获取回复消息
由于您还没有绑定你的真实邮箱,如果其他用户或者作者回复了您的评论,将不能在第一时间通知您!
发布评论
评论(1)
首先要考虑的是您是否将多维数据集配置为使用 ROLAP、MOLAP 或 HOLAP 存储模式。
MOLAP 将从您的数据源检索数据并将其永久存储在多维数据集将使用的文件中。这将大大提高查询速度,具体取决于源表返回数据的效率。 ROLAP 不会从源表中检索任何数据。每次查询多维数据集时,它都会访问源来获取结果。 HOLAP 解决方案介于两者之间。
关于大小的问题,您是否需要显示数据库表中所有列的聚合总数?如果限制多维数据集可用的度量和维度的数量,则会减小输出文件的大小。
The first thing to consider is whether you'll configure the cube to use ROLAP, MOLAP or HOLAP storage mode.
MOLAP will retrieve data from your data sources and permanently store them in files that will be used by the cube. This will help query speed considerably, depending on how efficient your source tables return data. ROLAP does not retrieve any data from the source tables. Everytime you query the cube, it will access the source to obtain the results. HOLAP solutions are somewhere between the two.
Regarding the concern for size, will you need to display aggregated totals for all of the columns from the database table? If you limit the number of measures and dimensions available through the cube, that will decrease the size of the output files.