VertiPaq 是什么及其工作原理
我正在学习列存储索引（Denali CTP3 功能）并了解到它使用 VertiPaq 架构进行数据压缩。我很想知道它是什么、它是如何工作的、它的架构。我查了谷歌但没有满意的结果。任何人都可以让我详细了解它是什么、它是如何工作的、其背后的算法/架构等。
以及它如何帮助数据压缩
I am learning the Column Store index (Denali CTP3 feature) and got to know that it uses VertiPaq architecture for data compression. I got interested to know as what it is , how it works, it's architecture. I checked in google but no satisfactory result. Could any one please let me know in detail as what it is , how it works , the algo/architecture behind this etc.
And how it helps in data compression
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（2）
我就此写了一篇博文，希望能够回答您有关列存储索引的问题：
http://www.jamesserra.com/archive/2011/08/sql-server-%e2%80%9cdenali%e2%80%9d-project-apollo/
请让我知道您是否还有疑问。
I wrote a blog post on this that hopefully will answer your questions on column store indexes:
http://www.jamesserra.com/archive/2011/08/sql-server-%e2%80%9cdenali%e2%80%9d-project-apollo/
Please let me know if you still have questions.
压缩部分工作得很好，因为同一列中的数据通常变化不大。例如，想象一下（简化）一个存储来自多 (4) 个选择输入的值的列。即使表中有 800 万条记录，列存储中也只有 4 个唯一值。这使得列值更容易压缩。这反过来又使得索引更容易装入内存，从而查询速度更快。
来源： 有关列存储技术的更多详细信息
The compression part works so well because very often data in the same column doesn't vary much. Imagine for example (simplification) a column that stores values from a multiple (4) choice input. There are going to be just 4 unique values in the column store, even if there are 8 million records in the table. That makes the column values easier to compress. That in turn makes it easier to fit the index into memory and thus faster to query.
Source: More details on columnstore technology