Snapshot: data snapshot (data disclosure method)

Catalog

Data snapshot

Store snapshots

How to implement a snapshot

Data snapshot

Data snapshot , It is an immutable fast data copy storage for e-commerce business data . It is convenient to check the price and other attributes of the goods at the moment of ordering , It will not change due to changes in the price system . Prevent goods from being sold out 、 Incomplete data such as offline , Go to the bottom of the data .

Example ： Baidu search snapshot

Search engine snapshot

Search engine snapshot , It refers to when visitors cannot open a search result , Or when the opening speed is particularly slow , The information provided for visitors was previously saved in the search engine The server The plain text of the corresponding web page content on . however , The snapshot content saved by search engines generally only includes text data , Non text data such as pictures and other multimedia will not be saved . therefore , When the source website cannot be accessed , Pictures and other multimedia will not be displayed in the snapshot .

Store snapshots

The global network storage industry association SNIA（StorageNetworking Industry Association） To the snapshot （Snapshot） Is defined as ： A fully available copy of the specified data set , This copy includes the corresponding data at some point in time （ Copy the start time ） Image of . A snapshot can be a copy of the data it represents , It can also be a copy of the data .

The function of snapshot is mainly to backup and restore online data . When the storage device has application failure or file damage, it can be quickly Data recovery , Restore data to an available point in time . Another function of snapshot is to provide another data access channel for storage users , When the original data is processed by online application , Users can access snapshot data , You can also use snapshots for testing and other work . all The storage system , Low end, high school, whatever , As long as it applies to online systems , Then snapshot becomes an indispensable function . 

How to implement a snapshot

Currently, there are two main technologies for snapshot , One is to copy when writing for the first time (Copy OnFirst Write,COFW), Sometimes referred to as copy on write （CopyOn Write,COW）. That is, when data is first written to a storage location , First, read out the original content , Write to another location （ Storage space reserved for snapshots , In this article, we call it snapshot space ）, Then write the data to the storage device . The next write operation for this location will no longer perform the copy on write operation . This technology is often used for the first time in computer related technology , Its basic principles are similar , Only the object-oriented is different , Applicable to different occasions . from COW We can know the execution process of , This implementation requires a read operation when writing to a storage location for the first time （ Read the data of the original location ）, Two write operations （ Write original location and write snapshot space ）, If writing is frequent , Then this method will be very expensive IO Time . So it can be inferred that , If you expect I/O Most of them are mainly read operations , Write less , The snapshot implementation technology in this way is an ideal choice , Because it takes less time to complete the snapshot . besides , If an application is prone to write hotspots , That is, write only for a limited range of data , that COW The snapshot implementation method of is also an ideal choice . Because its data changes are limited to a range , Multiple writes to the same copy of data will only result in one copy on write .