jrwev...@gmail.com
unread,Oct 10, 2012, 3:28:35 AM10/10/12You do not have permission to delete messages in this group
Either email addresses are anonymous for this group or you need the view member email addresses permission to view the original message
to
We are a bit concerned about the reported amount of idle time on our Production ASE 15.7 Server. Since the upgrade from 15.0.3 the amount of reported idle time on the server has dropped enormously suggesting that the server is far busier than it used to be. We use either the monEngine MDA table or the sp_monitor stored proc to measure the server idle time for reporting purposes. From what we can see, the new metric “IO CPU Time” seems to be attributed the difference between what used to be reported as Idle Time and what is now reported on 15.7. Here’s an example …
My MDA monitoring tool provides the following values for our 15.7 Production Database Server which are the delta values from the fields in monEngine for a 10 minute period this morning.
EngineNumber SystemCPUTime UserCPUTime CPUTime IOCPUTime IdleCPUTime Yields Connections DisIOChecks DiskIOPolled DiskIOCompleted ContextSwitches Server Time
0 299 2 38 161 98 91595 1223 0 0 121629 206912 47 8:50:01 AM
1 299 1 34 163 101 85954 1223 0 0 0 196872 47 8:50:01 AM
2 299 2 36 163 98 88675 1223 0 0 0 199889 47 8:50:01 AM
0 300 3 44 160 94 99555 1240 0 0 150756 230476 47 8:55:02 AM
1 301 3 44 161 93 94048 1240 0 0 0 227997 47 8:55:02 AM
2 301 3 44 161 93 96351 1240 0 0 0 230812 47 8:55:02 AM
The server has 3 engines, all reporting about 40 CPU seconds, 160 IO CPU Seconds and therefore 100 Idle CPU Seconds for each 300 Second period. Calls to the sp_monitor stored proc suggest the same sort of usage. A 5 minute sysmon around the same time suggests …
===============================================================================
Sybase Adaptive Server Enterprise System Performance Report
===============================================================================
Server Version: Adaptive Server Enterprise/15.7.0/EBF 20373 SMP ESD#02 /
P/x86_64/Enterprise Linux/ase157esd2/3109/64-bit/FBO/Sat
Jul 7 05:36:19 2012
Run Date: Oct 09, 2012
Sampling Started at: Oct 09, 2012 08:50:29
Sampling Ended at: Oct 09, 2012 08:55:29
Sample Interval: 00:05:00
Sample Mode: No Clear
Counters Last Cleared: Sep 22, 2012 22:18:29
Server Name: WARM4
===============================================================================
Kernel Utilization
------------------
Engine Utilization (Tick %) User Busy System Busy I/O Busy Idle
------------------------- ------------ ------------ ---------- ----------
ThreadPool : syb_default_pool
Engine 0 14.7 % 1.0 % 51.5 % 32.8 %
Engine 1 14.6 % 0.9 % 52.0 % 32.5 %
Engine 2 14.7 % 0.8 % 52.0 % 32.5 %
------------------------- ------------ ------------ ---------- ----------
Pool Summary Total 44.0 % 2.7 % 155.4 % 97.9 %
Average 14.7 % 0.9 % 51.8 % 32.6 %
------------------------- ------------ ------------ ---------- ----------
Server Summary Total 44.0 % 2.7 % 155.4 % 97.9 %
Average 14.7 % 0.9 % 51.8 % 32.6 %
Average Runnable Tasks 1 min 5 min 15 min % of total
------------------------- ------------ ------------ ---------- ----------
ThreadPool : syb_default_pool
Global Queue 0.0 0.0 0.0 0.0 %
Engine 0 0.2 0.2 0.2 26.5 %
Engine 1 0.3 0.2 0.2 44.1 %
Engine 2 0.2 0.2 0.1 29.4 %
------------------------- ------------ ------------ ----------
Pool Summary Total 0.8 0.5 0.5
Average 0.2 0.1 0.1
------------------------- ------------ ------------ ----------
Server Summary Total 0.8 0.5 0.5
Average 0.2 0.1 0.1
CPU Yields by Engine per sec per xact count % of total
------------------------- ------------ ------------ ---------- ----------
ThreadPool : syb_default_pool
Engine 0
Full Sleeps 37.5 0.4 11247 3.8 %
Interrupted Sleeps 296.0 3.4 88810 30.1 %
Engine 1
Full Sleeps 38.7 0.4 11598 3.9 %
Interrupted Sleeps 280.1 3.3 84017 28.5 %
Engine 2
Full Sleeps 36.5 0.4 10950 3.7 %
Interrupted Sleeps 293.6 3.4 88068 29.9 %
------------------------- ------------ ------------ ----------
Pool Summary 982.3 11.4 294690
------------------------- ------------ ------------ ----------
Total CPU Yields 982.3 11.4 294690
Thread Utilization (OS %) User Busy System Busy Idle
------------------------- ------------ ------------ ----------
ThreadPool : syb_blocking_pool : no activity during sample
ThreadPool : syb_default_pool
Thread 2 (Engine 0) 36.3 % 6.8 % 56.9 %
Thread 3 (Engine 1) 36.6 % 6.3 % 57.1 %
Thread 11 (Engine 2) 35.9 % 6.9 % 57.2 %
------------------------- ------------ ------------ ----------
Pool Summary Total 108.7 % 20.0 % 171.2 %
Average 36.2 % 6.7 % 57.1 %
ThreadPool : syb_system_pool
Thread 8 (NetController) 0.8 % 6.6 % 92.6 %
Thread 9 (DiskController) 0.2 % 0.5 % 99.3 %
Thread 10 (CtlibController) 0.0 % 0.1 % 99.9 %
------------------------- ------------ ------------ ----------
Pool Summary Total 1.0 % 7.2 % 391.8 %
Average 0.3 % 1.8 % 97.9 %
------------------------- ------------ ------------ ----------
Server Summary Total 109.8 % 27.2 % 963.0 %
Average 10.0 % 2.5 % 87.5 %
Adaptive Server threads are consuming 1.4 CPU units.
Throughput (committed xacts per CPU unit) : 62.8
A couple of things I’m a bit confused about …
1) thread utilization (OS %) suggests an idle time of nearly 60%. Why is this high in relation to the reported idle time ?
2) thread utilisation (Tick %) suggests a > 50% I/O Busy time which seems very high given that the I/O component should be offloaded to I/O threads which incidentally seem largely idle.
Is it possible that someone can shed some light on what we’re seeing here ? Thanks.