1 – transactions_field
The total number of transactions (both read/write and read-
only transactions). This statistic represents all transactions
(committed or rolled back) completed by all users of the
database, including transactions that read from the database
as well as transactions that modify the database.
2 – checkpoints_field
The total number of checkpoints. The total checkpoints
category is further broken down into categories of reasons for
checkpointing. The statistics for these categories are included
in the "AIJ growth," "txn limit," "time limit," "rollback," "AIJ
backup," and "global" subfields.
Note that there may be more reasons for checkpoints than there
are total checkpoints. For example, you might have a total count
of 100 for checkpoints, but when you add the number of checkpoint
reasons ("AIJ growth," "txn limit," "time limit," "rollback,"
"AIJ backup," and "global"), the total could be greater than 100.
This occurs because a single checkpoint may be triggered by more
than one event. For example, a checkpoint may occur because of
time and AIJ file growth. Although the total count columns for
the "interval: seconds" and "interval: AIJ blks" fields are both
incremented by one, the total count column for "checkpoints" is
only incremented by one.
3 – AIJ growth field
The number of checkpoints for all processes due to the .aij file
growth checkpoint limit.
4 – ____txn_limit_field_
The number of checkpoints for all processes due to the logical-
defined transaction limit.
5 – ____time_limit_field_
The number of checkpoints for all processes due to the time
interval checkpoint limit.
6 – ____rollback_field_
The number of checkpoints automatically triggered by rollback of
transactions that updated the database.
7 – AIJ backup field
The number of system-generated checkpoints due to periodic
backups to tape of the .aij file by the AIJ spooler.
8 – ____global_field_
The number of system-wide checkpoints, issued from an RMU
Checkpoint, RMU Backup, or RMU Backup After_Journal command.
9 – interval: AIJ blks field
This field displays the sum of the intervals between checkpoints
due to AIJ growth in block checkpoints for all processes. For
example, if Process 1 checkpoints at virtual block number (VBN)
100, then checkpoints again at VBN 250, the AIJ block interval
category is incremented by 150. If Process 2 checkpoints at VBN
125, then checkpoints again at VBN 200, the AIJ block interval
is incremented by an additional 75. Statistics for the other two
interval categories are displayed in the "interval: tx count" and
"interval: seconds" fields.
If CHECKPOINT INTERVAL IS 1000 BLOCKS is specified with the SQL
ALTER DATABASE statement, each process checkpoints when the .aij
file has grown 1000 blocks since the process' last checkpoint.
Keep in mind that checkpointing influences recovery time. The
main reason to consult checkpoint statistics is to find the
average interval per checkpoint. You can use the information in
the total count column to compute this average. For each category
of checkpoint reason, use the average interval per checkpoint to
help you decide if a checkpointing interval should be adjusted,
and by how much.
If most of the checkpoints for a database are triggered by a
particular checkpoint limit, that limit may be set too high,
or the other two limits may be set too low. You can determine
the average interval per checkpoint for each type of checkpoint
limit. After you have this information, you can reset the limits
so that each type of checkpoint limit triggers approximately the
same number of checkpoints, which results in optimal performance.
To compute the average interval in AIJ blocks, divide the
total count for the AIJ block interval by the total number of
checkpoints minus the number caused by AIJ backups. Although
checkpoints caused by AIJ backups are counted in the total
number of checkpoints, they are not counted in the total of AIJ
block intervals. If the total count of AIJ block intervals is
70000, the total count of checkpoints is 100, and the number
of checkpoints caused by AIJ backups is 1, then the average AIJ
block interval is 707:
70000/(100 - 1) = 707
The help for the "interval: tx count" field explains how to
determine the average interval for transaction checkpoints.
The help for the "interval: seconds" field explains how to
determine the average interval for time checkpoints.
10 – interval:_tx_count_field
This field displays the sum of the intervals between checkpoints
due to the transactions count checkpoint for all processes. For
example, if Process 1 checkpoints after 20 transactions, the
transactions count category is incremented by 20. If Process
2 checkpoints after 30 transactions, the transactions count
category is incremented by an additional 30. Statistics for the
other two interval categories are displayed in the "interval: AIJ
blks" and "interval: seconds" fields.
The transactions limit for checkpoints is determined by the
setting of the RDM$BIND_CKPT_TRANS_INTERVAL logical name. If
RDM$BIND_CKPT_TRANS_INTERVAL is defined as a system logical set
to 10, each process will checkpoint after 10 transactions unless
a user redefines the RDM$BIND_CKPT_TRANS_INTERVAL logical to a
different value. That is, if a user defines RDM$BIND_CKPT_TRANS_
INTERVAL as a process logical and sets a value of 5, that user
will checkpoint after 5 transactions.
Keep in mind that checkpointing influences recovery time. The
main reason to consult checkpoint statistics is to find the
average interval per checkpoint. You can use the information in
the total count column to compute this average. For each category
of checkpoint reason, use the average interval per checkpoint to
help you decide if a checkpointing interval should be adjusted,
and by how much.
If most of the checkpoints for a database are triggered by a
particular checkpoint limit, that limit may be set too high,
or the other two limits may be set too low. You can determine
the average interval per checkpoint for each type of checkpoint
limit. After you have this information, you can reset the limits
so that each type of checkpoint limit triggers approximately the
same number of checkpoints, which results in optimal performance.
To compute the average transactions interval, divide the
total count for transaction intervals by the total number of
checkpoints. If the total count for transaction intervals is
800 and the total number of checkpoints is 100, then the average
number of transactions between checkpoints is 8.
800 / 100 = 8
The help for the "interval: AIJ blks" field explains how to
determine the average interval for .aij file growth checkpoints.
The help for the "interval: seconds" field explains how to
determine the average interval for time checkpoints.
11 – interval:_seconds_field
This field displays the sum of the intervals between time in
seconds checkpoints for all processes. For example, if Process
1 checkpoints after 500 seconds, the time in seconds category is
incremented by 500. If Process 2 checkpoints after 600 seconds,
the time in seconds category is incremented by an additional 600.
Statistics for the other two interval categories are displayed in
the "interval: AIJ blks" and "interval: tx count" fields.
If CHECKPOINT TIMED EVERY 600 SECONDS is specified with the
SQL ALTER DATABASE statement, each process checkpoints every
10 minutes.
Keep in mind that checkpointing influences recovery time. The
main reason to consult checkpoint statistics is to find the
average interval per checkpoint. You can use the information in
the total count column to compute this average. For each category
of checkpoint reason, use the average interval per checkpoint to
help you decide if a checkpointing interval should be adjusted,
and by how much.
If most of the checkpoints for a database are triggered by a
particular checkpoint limit, that limit may be set too high,
or the other two limits may be set too low. You can determine
the average interval per checkpoint for each type of checkpoint
limit. After you have this information, you can reset the limits
so that each type of checkpoint limit triggers approximately the
same number of checkpoints, which results in optimal performance.
To compute the average time interval, divide the total count for
seconds interval by the total number of checkpoints. If the total
count for the seconds field is 59,300 and the total number of
checkpoints is 100, the average number of seconds between each
time-triggered checkpoint is 593.
59,300 / 100 = 593
The help for the "interval: AIJ blks" field explains how to
determine the average interval for .aij file growth checkpoints.
The help for the "interval: tx count" field explains how to
determine the average interval for transaction checkpoints.
12 – checkpoint_stall_field
This field displays the checkpoint duration in seconds.
13 – flushed_buffers_field
This field displays the number of buffers flushed to disk during
a checkpoint operation.