Shulou(Shulou.com)06/01 Report--

ORACLE CheckpointQueue and incremental checkpoint

The main purpose of checkpoints is to refresh dirty blocks in a manner that has minimal impact on the daily operation of the database. Dirty blocks continue to produce, how to refresh the dirty blocks to the disk? Before 8i, Oracle periodically locked all modification operations and brushed all dirty blocks in Buffer cache to disk. This way of refreshing dirty blocks is called a complete checkpoint, which greatly affects the efficiency. After 9i, a complete checkpoint occurs only when the database is closed.

The new checkpoint location is Node 4, which corresponds to the current dirty block with the earliest dirty time. Nodes 1, 2 and 3 have been removed from the checkpoint queue. Because their corresponding dirty pieces are no longer dirty. In general, the blocks after the checkpoint location in the control file are dirty. But sometimes there are exceptions, because the checkpoint location is only updated every three seconds. As in the figure above, the dirty blocks corresponding to Node 1, Node 2 and Node 3 have been refreshed, but because the three-second interval has not arrived, the checkpoint location still points to Node 1. Only when three seconds have elapsed will the checkpoint location be updated to Node 4.

Let's stop here with regard to checkpoint queues and checkpoint locations, and before giving a comprehensive introduction to what incremental checkpoints are, let's talk about an important role of checkpoint queues.

Let's first summarize what happens inside Oracle when the user modifies the block:

1. If the block is not in Buffer cache, read the block into Buffer cache

2. Mr. Cheng redoes the record, writes it into the log cache, and writes it to the log file when the user submits it.

3. Modify blocks in Buffer cache

4. The dirty mark bit of the block is set in the Buffer cache to mark that the block becomes a dirty block, and a new node is added at the end of the checkpoint queue to record the information of the new dirty block, including the position of the dirty block in the Buffer cache and the redo record position corresponding to the dirty block generated in step 2.

5. After the user submits, the corresponding redo record is written to the log file from the redo cache.

Let's discuss the setting of incremental checkpoints.

We can see a lot of information about checkpoints in V$instance_recovery:

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.