The ultimate cost for mass data optimization - HBase integrated cold cloud separated

Foreword

With the continuous development of business, database storage business will continue to grow. Normally the amount of data billions of dollars, we need to consider doing sub-library sub-table, or choose a better scalability NOSQL / NewSQL databases such as HBase table can support a single PB-level data, enough to meet the storage needs of most businesses. However, for a large number of storage bottlenecks class of business, the cost of storage system design is still the need to focus on the most important, the separation of hot and cold applications and solutions born.

Suitable for hot and cold separate business

Data billing / order class system is very suitable for hot and cold separation, the data of such systems over time tend to accumulate a mass of data, but also because of the importance of data, these data should be stored permanently. However, users often only information on the latest consumer orders or bills, more than six months of basic orders will not be accessed.
image.png
Data monitoring system also showed a very significant cold stratification characteristics. Users usually only view real-time monitoring, historical data back only when the fault is only possible to make inquiries. And if the real-time data with historical data mixed together, not only will the cost of storage is very high, and will slow down the real-time search

Guess you like

Origin yq.aliyun.com/articles/718395