When should I use Apache Druid

Many companies have put Druid It can be applied to many different application scenarios . Please visit   Use Apache Druid The company   Page to find out which companies use Druid.

If your usage scenario conforms to some of the following features , that Druid It will be a very good choice ：

• The insertion frequency of data is very high , But the update frequency is very low .
• Most queries are aggregate queries （aggregation） And report query （reporting queries）, For example, we often use “group by” Inquire about . There are also some retrieval and scanning queries .
• The query latency is limited to 100ms To In a few seconds .
• Your data has a time component （ attribute ）. For time-related attributes ,Druid Carry out special design and optimization .
• You may have multiple data tables , But queries are usually only for a large distributed data table , however , The query may also need to query multiple smaller  lookup  surface .
• If your data has a high cardinality （high cardinality） Data field , for example URLs、 user IDs, But you need to quickly count and sort these fields .
• You need to get from Kafka,HDFS, text file , Or object storage （ for example ,AWS S3） Load data in .

If your usage scenario is some of the following ,Druid  No A better choice ：

• For an existing record , Use primary key （primary key） Perform low latency update operations .Druid Support streaming insertion （streaming inserts） data , But it doesn't support streaming updates very well （streaming updates） data . Druid The update operation of is completed through background batch processing .
• Your system is similar to an offline reporting system , Query latency is not an important consideration in system design .
• The table needs to be updated in the usage scenario （Fact Table） Make a connection query , And for this query, you can introduce a high delay to wait for the query to complete .

https://www.ossez.com/t/apache-druid/13604