Kofax Communication Server 10.0.0 User Manual
Page 60
![background image](/manuals/658232/60/background.png)
Environment Guide
Version 10.00.00
60
© Copyright Kofax. All information is subject to change without notice.
Update kB/sec
848
Update Write kB/sec
216
TCOSS Cache
Database
Cache Misses/sec
1
Directory
Cache Misses/sec
1
Document
Cache Misses/sec
1.4
Processor
_Total
% Processor Time (0 ..100)
33
Process
TCOSS
% Processor Time (0 ..400)
32
The TCOSS performance counter “Update kB/sec” shows the amount of data read from the primary master,
the performance counter “Update Write kB/sec” gives the amount written to the secondary master. The
difference between the two is the data which turned out to be equal on the secondary master‟s disk so it
was not written.
The additional disk reads (848 kB/sec, 53 Reads/sec) resulting from the update of the secondary master
cause no problem for the physical disk, the disk read time does not go up, probably because these disk
reads are in sequential order, not random like the other TCOSS disk reads.
In the second run of this test case the queue length of outbound faxes was above 700 and the TCOSS
cache worked less efficiently. The update was running a bit slower in this case (698 kB/sec):
Object
Instance
Counter
Value
Physical Disk
D: (TCOSS)
Disk Writes/sec
99.5
Disk Write Bytes/sec
628,000
Disk Reads/sec
59
Disk Read Bytes/sec
909,000
%Disk Write Time
62
%Disk Read Time
6.5
TCOSS Disk
Read kB/sec
189
Reads/sec
15
Update kB/sec
698
Update Write kB/sec
181
TCOSS Cache
Database
Cache Misses/sec
1
Directory
Cache Misses/sec
1
Document
Cache Misses/sec
13
Processor
_Total
% Processor Time (0 ..100)
33
Process
TCOSS
% Processor Time (0 ..400)
32
The about 12 additional, random TCOSS disk reads cause the disk read time to go up a bit.
5.9.4
Test Results: Primary Mastery Failure
To test this situation the TCSRV service was stopped on the primary master while the tandem system was
in a healthy, fault-tolerant state. The secondary master started stand-alone, it took 21 minutes to boot all 38
line servers. After about 15 minutes the primary master was restarted, it took the primary about 20 minutes
to boot all line servers. About an hour later the update of the primary master from the secondary was
completed.
The following performance values were measured on the primary master during the re-synchronization
phase:
Object
Instance
Counter
Value
Physical Disk
D: (TCOSS)
Disk Writes/sec
118
Disk Write Bytes/sec
952,000
Disk Reads/sec
8.7
Disk Read Bytes/sec
143,000
%Disk Write Time
71