onstat -g rah command: Print read-ahead request statistics

Use the onstat -g rah command to display information about read-ahead requests.


Syntax:

>>-onstat-- -g--rah--------------------------------------------><

Example output

Figure 1. onstat -g rah command output

Read Ahead

# Qs                 1
# threads            2         
# Requests           58690
# Continued		        0    
# Memory Failures    0         
Last Thread Add      04/06/2013.14:34
Way behind           0

Partition ReadAhead Statistics
   
          Buffer   Disk   Hit    Data       Index        Idx/Dat     Log/PageList Last Committed 
Partnum   Reads    Reads  Ratio  # Reqs Eff # Reqs  Eff  # Reqs  Eff # Pages Eff  # Reqs  Eff   # Resch 
0x200003  4312677  110    99     0      0   0       0    0       0    0      0     12906  100   0
0x300002  23740584 1427   99     0      0   0       0    0       0    0      0     6681   100   7
0x400002  17818942 966    99     0      0   0       0    0       0    0      0     25849  100   57

Output description

Qs: Number of queues for read-ahead requests
# threads: Number of read-ahead threads
# Requests: Number of read-ahead requests
# Continued: Number of times a read-ahead request continued to occur
# Memory Failures: Number of failed requests because of insufficient memory
Last Thread Add: Date and time when the last read-ahead thread was added
Way behind: How many page list requests were dropped because the read-ahead daemon is too far behind
Partnum: Partition number
Buffer reads: Number of bufferpool and disk pages that were read
Disk Reads: Number of pages that were read from disk
Hit Ratio: Cache hit ratio for the partition
# Reqs: Number of read ahead requests. (There are 5 instances of this output field: for data, the index, index data, log pages, and last committed rows.)
Eff: Efficiency of the read-ahead requests. This is the ratio been the number of pages requested by read-ahead operations to the number of pages that were already cached and for which a read-ahead operations was not needed. Values are between 0 and 100. A higher number means that read ahead is beneficial. (There are 5 instances of this output field: for data, the index, index data, log pages, and last committed rows.)
Resch: The number of requests for last committed rows that are rescheduled because the updates to a multi-piece row are not complete.