|
Home > Archive > Server + > September 2002 > another one, see ya in the AM!
You are viewing an archived Text-only version of the thread.
To view this thread in it's original format and/or if you want to reply to
this thread please [click here]
| Author |
another one, see ya in the AM!
|
|
| twister166 2002-09-08, 5:29 pm |
| Your department has added 50 new users to the network. All of them are using the same application located on the main server in the data center. During peak usage hours the response of the application has slowed to an unacceptable level.
The server is configured with two CPUs, 1 GB of memory, a 100 GB SCSI disk array, and one Fast Ethernet NIC. The performance baselines taken during peak hours are:
Historic Current
- Average CPU utilization 50% 60%
- Average CPU queue 1.5 1.7
- Average network utilization 25% 40%
- Average network queue 0.2% 0.2
- Memory hard page faults 0.1 pages/s 0.12 pages/s
- Memory soft page faults 100 pages/s 200 pages/s
- Average disk subsystem utilization 10% 8%
- Average disk subsystem queue 0.1 0.05
Where is the bottleneck most likely located and how can it be resolved?
A. Memory is the bottleneck. More RAM needs to be installed.
B. CPUs are the bottleneck. Fast CPUs need to be installed.
C. The network is the bottleneck. Replace the hub with a switch.
D. The disk I/O is the bottleneck. Add an additional SCSI controller. | |
| Tech Ranger 2002-09-08, 6:10 pm |
| quote:
Memory hard page faults 0.1 pages/s 0.12 pages/s
Is this a typo? Do you mean 1 and 12? If not, are you sure of all your entries? | |
|
| well, if you had 1 page fault in 10 seconds, the page fault rate would be .1 fault per second...
i think you can only get 35-37% capacity with ethernet so going from 25% to 40% is a problem. i'd buy a new switch (answer c). | |
| twister166 2002-09-08, 6:55 pm |
| quote:
Originally posted by Tech Ranger
Is this a typo? Do you mean 1 and 12? If not, are you sure of all your entries?
I did not develope the Q, I copied about 20 Q's from various sites that I visited... I think the number is ok. Just FYI and for other people who are not familiar with the page faults: (From M$)
If a process requests a page in memory and the system cannot find it at the requested location, this constitutes a page fault. (If the page is elsewhere in memory the fault is called a soft page fault. If the page must be retrieved from disk, the fault is called a hard page fault.) The system also counts a page fault on a file access if the requested page is not found in the File System Cache and must be retrieved from storage. The page fault counters do not distinguish between hard and soft faults, so you must combine counter information to deduce the number of hard faults. | |
| Tech Ranger 2002-09-08, 6:58 pm |
| I didn't know Ethernet maxes out at around 35%, or if I knew it I forgot it. Does anybody else agree with that contention? | |
| twister166 2002-09-08, 7:03 pm |
| quote:
Originally posted by Tech Ranger
I didn't know Ethernet maxes out at around 35%, or if I knew it I forgot it. Does anybody else agree with that contention?
Based on Cisco, Ethernet max out at 40% and 70% for TR. Network Essential from M$, says 30% and 60% respectely. | |
| charlie130 2002-09-08, 9:25 pm |
| From reading the ti leaves, my answer is (A)
Too many memory faults, hard or soft...
Keep up the good work Twister.. | |
| azimuth40 2002-09-08, 11:48 pm |
| I don't like the performance figures given, they seem strange. But I am guessing that we need a switch placed in here so my answer is C.
Strangeness is that hard page faults increased but disk utilization decreased. Soft faults increased much more than hard fault rate so I am guessing memory won't solve the problem. If the disk utilization had increased above 10% then that would be time to look at the disk subsystem and/or memory because the VMM is probably working too hard; but again disk dropped to 8%, unless that is a typo. Disk subsystem queue is also very low so that says it is performing well. If that value was over two then again time to look at it. CPU whould have to be over 80 to even give it a second notice.
Yep I go with C for the moment.
 | |
| twister166 2002-09-09, 6:14 pm |
| My answer to this question was C as well.
Memory was not in stress, as the hard page fault are small, disk % is small too. Even the softpage fault was increase significantly, but soft page fault can be in the hundreds and without problem. The only thing that is tight is the network, so C.
Again, I don't have an answer for the question that I am posting. In fact, the questions that I am posting had an answer that I did not agree. The original answer was A. |
|
|
|
|