Skip to Main Content

GoldenGate

Announcement

For appeals, questions and feedback about Oracle Forums, please email oracle-forums-moderators_us@oracle.com. Technical questions should be asked in the appropriate category. Thank you!

Cache Statistics interpretation and Tuning

Arturo GutierrezOct 5 2017 — edited Oct 5 2017

Hello,

I've a extractor with a configuration of cache of this:

CACHEMGR CACHESIZE 128M

The process is up since:

Last Started 2017-05-22 08:47   Status RUNNING

About 5

If we see cache statistics:

send extract EXT_E0 CACHEMGR CACHESTATS

Sending CACHEMGR request to EXTRACT EXT_E0 ...

CACHE OBJECT MANAGER statistics

CACHE MANAGER VM USAGE

vm current     =   3.57M vm anon queues =   3.57M

vm anon in use =      0    vm file        = 0

vm used max    = 148.13M   ==> CACHE BALANCED

CACHE CONFIGURATION

cache size            = 128M   cache force paging = 209M

buffer min            =  64K buffer max (soft)  =   1M

pageout eligible size =   1M

================================================================================

RUNTIME STATS FOR SUPERPOOL

CACHE Transaction Stats

trans active    =      0 max concurrent =     22

non-zero total  = 50.12M   trans total    = 51.66M

CACHE File Caching

filecache rqsts        = 10.100K   bytes to disk      = 278.40G

file retrieves         = 283.68K   objs filecached    = 96

queue entries          =  10.81K queue processed    =  17.17K

queue entry not needed =    190    queue not signaled =     45

fc requesting obj      =      0

CACHE MANAGEMENT

buffer links   = 308.95K  anon gets =      0

forced unmaps  =   4.65K cnnbl try   =  16.13K

cached out     =   6.78K

Because I see Cache file caching activity:

Filecache rqsts        = 10.100K   bytes to disk      = 278.40G

This means, that from the start to up, the extract accessed 10.100K time to disk and write 278.40 GB. !

With these statistics I can already know that there have been a number of times that the extractor has dumped transactions from the cache to the disk, due to the memory pressure.

Then increase the cache will reduce I/O to disks.

however, it is possible to know more information about the nature of the transactions that have caused the disk dump. For example, maximum size of the transaction, duration etc., if the write was done for a a large transacction, o BR interval.

Which is you opinion?

Thanks

Arturo

Comments
Locked Post
New comments cannot be posted to this locked post.
Post Details
Locked on Nov 2 2017
Added on Oct 5 2017
2 comments
656 views