As people work with the data in your databases, they make changes. They may be inserting new information, deleting information they no longer need, or modifying other information or all of the above. As the data changes, your indexes are directly affected, both rowstore and columnstore, albeit in different ways. For your rowstore indexes, the acts of modifying the data can lead to page splits, pages getting rearranged, and pages getting emptied, all summed up in a term called fragmentation. In your columnstore indexes, you’re not dealing with the same kind of fragmentation, but you are dealing with the deltastore, marking data that has been deleted, or keeping updated and inserted data ready for when it gets pivoted and compressed. Fragmentation can, in some circumstances, lead to performance degradation. Unfortunately, fixing fragmentation can also lead to performance degradation.

错误:搜索内容不能为空,请输入英文关键词
错误:关键词超出字数限制,请精简
高级检索

Dealing with Index Fragmentation

  • Grant Fritchey

摘要

As people work with the data in your databases, they make changes. They may be inserting new information, deleting information they no longer need, or modifying other information or all of the above. As the data changes, your indexes are directly affected, both rowstore and columnstore, albeit in different ways. For your rowstore indexes, the acts of modifying the data can lead to page splits, pages getting rearranged, and pages getting emptied, all summed up in a term called fragmentation. In your columnstore indexes, you’re not dealing with the same kind of fragmentation, but you are dealing with the deltastore, marking data that has been deleted, or keeping updated and inserted data ready for when it gets pivoted and compressed. Fragmentation can, in some circumstances, lead to performance degradation. Unfortunately, fixing fragmentation can also lead to performance degradation.