OLAP是把多维数据的组织与聚合分析能力与计算引擎结合、支撑快速交互式查询的数据处理技术，核心是多维模型下的预聚合与切片钻取，能力包括多维数据模型、切片切块、钻取旋转与聚合计算。它解决的是业务方要自助分析时每次都等数仓跑批的问题，并不解决底层数据本身不准确与维度定义混乱的问题。
OLAP的核心是多维与聚合，它可以帮助企业实现以下目标：
让分析查询在秒级返回。常用维度组合的结果被预先计算并缓存，用户切换分析视角时不必等待现场计算。多维模型通常以星型或雪花型结构组织，事实表存放度量值，维度表存放分析角度，模型设计直接决定查询性能上限。
支持从不同维度观察同一件事。切片与切块让同一指标可以在时间、区域、产品线等多个维度上自由组合对比。切片切块在维度集合上选取子集，钻取在层级之间上下移动，旋转在不同维度之间切换观察轴向，三者组合覆盖绝大多数分析动作。
让细节可以逐层下钻。钻取操作从汇总值下探到明细记录，分析者能一路追到原始单据。聚合计算把高频组合的结果提前算好并存储，避免每次查询都现场扫描事实表，这是分析型系统响应速度的关键来源。
让视角可以自由旋转。维度之间的轴向可以互换，观察口径的调整不必等数据部门重新开发。
总之，OLAP可以帮助企业实现分析方式从固定报表到自助多维探索的转变，缩短业务取数等待时间，提高数据洞察的灵活性与决策响应速度，推动数据资产从存储走向实际使用。
翻译
OLAP is the combination of multidimensional data organization and aggregation capability with compute engines to support fast interactive queries. Its core is pre-aggregation and slicing and drill-down under multidimensional models, with capabilities covering multidimensional data models, slicing and dicing, drill-down and rotation, and aggregation computation. It solves the problem of business users waiting for warehouse batches whenever they want to analyze data themselves, but it does not solve the problems of inaccurate source data and confused dimension definitions.
At the heart of OLAP is multidimensionality and aggregation, which can help companies achieve the following:
Return analytical queries in seconds. Results for common dimension combinations are precomputed and cached, so switching analytical perspectives does not require waiting for on-the-fly computation.
Support observing the same thing from different dimensions. Slicing and dicing let one metric be freely combined and compared across time, region, product line and other dimensions.
Allow drilling down to detail level by level. Drill-down operations descend from aggregated values to individual records, letting analysts trace all the way back to source documents.
Allow perspectives to be rotated freely. Axes between dimensions can be swapped, so adjustments to the observation angle do not wait on the data team to build something new.
In short, OLAP can help enterprises shift analysis from fixed reports to self-service multidimensional exploration, shorten business users’ waiting time for data, improve the flexibility of data insight and decision response speed, and push data assets from storage toward actual use.
