A checkpoint can occur at any point in the physical-logging process. The database server performs two types of checkpoints: full and fuzzy. After a full checkpoint occurs, all modified pages in shared memory are flushed to disk. The database server is physically consistent because all changes to the data since the prior checkpoint are recorded on disk and in the logical log. For information, see Full Checkpoint
After a fuzzy checkpoint occurs, the database server might not be physically consistent, but the checkpoint completes much quicker and does not tie up the database server during heavy update activity. All changes to the data since the last full checkpoint or fast recovery are recorded in the logical log. For information, see Fuzzy Checkpoint.
Home | [ Top of Page | Previous Page | Next Page | Contents | Index ]