There are a couple of AIX commands needed to monitor the effectiveness of the JFS filesystem cache.

The first is mpstat.  The 'mpstat -w 2' command has two columns, min and maj, these columns tell you the number of minor and major page faults on a per logical processor, but only for file-systems reads, as writes do not cause page faults as the data to be written is already in memory.

min = Minor page faults (page faults with no IO).
maj = Major page faults (page faults with disk IO).

 

mpstat -w 2

cpu min maj mpc int    cs  ics rq   mig  lpa   sysc  us   sy   wa   id   pc  %ec   lcs
0   479  0   0 3122  3610  177  3   371 100.0  3890 5.0 87.8  0.0  7.2 0.39  23.0  691
1   614  0   0 1536  1854  107  0   391 100.0  1981 9.9 81.1  0.0  9.0 0.35  20.8  474
2     0  0   0   39    25    0  0    11 100.0    25 0.0  0.2  0.0 99.8 0.07   3.9   21
3     0  0   0   16     1    0  0     0 100.0     0 0.0  0.1  0.0 99.8 0.07   4.0   18
4   748  0   0 1826  2623  133  2   361 100.0  5801 5.7 85.9  0.0  8.4 0.36  21.0  398
5   476  0   0 2284  2786  115  0   391 100.0  3009 6.3 81.7  0.0 12.0 0.31  18.2  567
6     0  0   0   14     0    0  0     0 100.0     0 0.1  0.0  0.1 99.9 0.07   4.4   17
7   616  0   0  107     1    0  0     3 100.0  1638 3.5 17.2  0.0 79.3 0.09   5.2  107
ALL 2933 0   0 8944 10900  532  5  1528 100.0 16344 5.7 70.6  0.0 23.7 1.71 213.5 2293

 

iostat -F /export 4

System configuration: lcpu=8 ent=2.00 fs=21

FS Name:     % tm_act    Kbps     tps  Kb_read Kb_wrtn     time
/export          -     173123.0 43280.8 692492      0    08:01:28
/export          -     162620.2 40655.2 650480      0    08:01:32
/export          -     145422.0 36355.5 581688      0    08:01:36
/export          -     171614.2 42903.8 686456      0    08:01:40

If you compare the IO rates for Kb_read against the iostat -DR command, you get the effectiveness of your JFS2 cache.
In the following example, we see zero disk IO, therefore all those filesystem reads are coming from the JFS2 cache.

iostat -DRT hdiskpower0 4

System configuration: lcpu=8 drives=602 paths=4 vdisks=2

hdiskpower0   xfer:  %tm_act     bps   tps  bread  bwrtn        time
                        0.0      0.0   0.0   0.0    0.0       08:01:28
                        0.0      0.0   0.0   0.0    0.0       08:01:32
                        0.0      0.0   0.0   0.0    0.0       08:01:36
                        0.0      0.0   0.0   0.0    0.0       08:01:40

 

Whereas in the following example, the reads are coming from the disk.
You see the bread column from iostat -DR is non-zero, so data is read from the disk.

FS Name:     % tm_act   Kbps     tps  Kb_read Kb_wrtn     time
/export          -     2666.9  755.2  10667        0    12:37:36
/export          -     2135.5  724.5  8542         0    12:37:40
/export          -     2220.0  727.2  8879         0    12:37:44
/export          -      412.3  114.8  1649         0    12:37:48

hdiskpower0 xfer: %tm_act    bps   tps    bread  bwrtn    time
                     25.5   2.7M  137.2    2.7M    0.0   12:37:35
                     21.5   3.0M  358.5    3.0M    0.0   12:37:39
                     24.0   3.5M  330.5    3.5M    0.0   12:37:43
                     14.2   1.2M  173.0  870.4K    0.0   12:37:47