Lab Question

 < Free Open Study > 



 

This exercise is designed to test your knowledge of monitoring and troubleshooting MetaFrame XP.

You are a consultant with XYZ Consulting Services. You've been sent out on a job to assist a company in troubleshooting their Windows 2000 Advanced Server/MetaFrame XPe server farm. The administrators on site tell you they have followed a step-by-step process when planning, testing, and implementing the server farm. They cannot understand why there are so many performance problems.

The administrators inform you there has only been one change in their environment since they originally implemented the server farm: the addition of several new users. They go on to say that the new users were the source of many complaints about applications loading slowly, and they only ran one specific application, which had just recently been installed. As for when the complaints started, some users had been complaining about slow logons since before the new users arrived.

The administrators on site have been monitoring the servers using System Information Tool, Event Viewer but have not discovered what the problem is. There are no errors in the Event Log and no installed devices appear to be faulty. They have kept a log of the complaints they have been receiving over the last week. It is your job to track down what their problem(s) may be.

As you read through the log, you see the following items:

  • Monday, users complaining of slow logons and applications loading slowly.

  • Tuesday, user logons still slow but applications loading a little faster.

  • Wednesday, a third of the users are out of the office for a seminar and those left behind haven't complained too much about applications loading slowly, but logging on is still slow. A few times during the day the network just crept along.

  • Thursday, logons still slow, applications loading slowly again, but network seems to be faster today.

  • Friday, logons still slow, applications loading a bit faster, especially as the day progressed. The network appeared to be okay.

if the onsite administrators were careful in their planning, testing, and implementation, it s likely that they sized their servers correctly; at least in the beginning. here s a list of the clues you should have picked up on: * the addition of several new users. * the addition of a new application installed on the servers. * the fact that the new users were among the users complaining about how slow their applications loaded, even though they only used the server farm for the one application. * slow logons had been happening since before the new users arrived, so it is most likely a separate problem * the problem with the network also appears unrelated to the new users, but it could be significant that it occurred when many of the users were out of the office. * monday and tuesday appear to be the days when users complain most about the programs loading slowly. * wednesday, when many users were out of the office, there were few complaints about applications. * friday, the application behavior was somewhat better. * slow logons appear to be the only consistent problem. plausible answers and resolutions: since the server farm is running metaframe xp e , you could ask to install resource manager on the servers to be monitored. with resource monitor, you can track application usage and system health. after running the resource manager for a few days, you should be able to see some trends. it is quite likely that the new users application is a resource hog. it could also be possible that the additional users themselves are overloading the servers. in resource monitor, check the network interface; bytes total/sec can tell you if the servers are receiving more traffic than the nic can handle. if this is the case, it might be wise to add an additional server, but in this scenario, it doesn t sound necessary. memory available bytes should also be monitored. this metric can tell you if your servers are low on memory. an application that is a resource hog could be causing the problem. if the memory available bytes counter is high, the disk use and page file %usage will also be high because the system must use the page file when supplementing its physical memory with virtual memory. if the counters are indeed high, adding additional ram to the servers may solve the problem of applications loading slowly. the slow logons were a gimme. this is typically the case when user profiles are out of control. make sure that temp files, temporary internet files, favorites, user documents, outlook .pst files, and any other data that can be saved elsewhere is. logons are guaranteed to speed up dramatically once this is done. as for the slow network response and faster application loading on wednesday, a good guess would be that the users who were out of the office for the day included at least some of the new users running the memory-hungry application, while the other half were probably managers. users left with no supervisor looking over their shoulder might take the opportunity to browse the internet, play online games or download mp3s. this theory could be tested by employing network monitor the next time a similar situation occurred.

Answers

 

If the onsite administrators were careful in their planning, testing, and implementation, it's likely that they sized their servers correctly; at least in the beginning. Here's a list of the clues you should have picked up on:

  • The addition of several new users.

  • The addition of a new application installed on the servers.

  • The fact that the new users were among the users complaining about how slow their applications loaded, even though they only used the server farm for the one application.

  • Slow logons had been happening since before the new users arrived, so it is most likely a separate problem

  • The problem with the network also appears unrelated to the new users, but it could be significant that it occurred when many of the users were out of the office.

  • Monday and Tuesday appear to be the days when users complain most about the programs loading slowly.

  • Wednesday, when many users were out of the office, there were few complaints about applications.

  • Friday, the application behavior was somewhat better.

  • Slow logons appear to be the only consistent problem.

Plausible answers and resolutions:

Since the server farm is running MetaFrame XPe, you could ask to install Resource Manager on the servers to be monitored. With Resource Monitor, you can track application usage and system health. After running the Resource Manager for a few days, you should be able to see some trends. It is quite likely that the new users' application is a resource hog. It could also be possible that the additional users themselves are overloading the servers.

In Resource Monitor, check the Network Interface; Bytes Total/sec can tell you if the servers are receiving more traffic than the NIC can handle. If this is the case, it might be wise to add an additional server, but in this scenario, it doesn't sound necessary. Memory Available Bytes should also be monitored. This metric can tell you if your servers are low on memory. An application that is a resource hog could be causing the problem. If the Memory Available Bytes counter is high, the Disk Use and Page File %Usage will also be high because the system must use the Page File when supplementing its physical memory with virtual memory. If the counters are indeed high, adding additional RAM to the servers may solve the problem of applications loading slowly.

The slow logons were a gimme. This is typically the case when user profiles are out of control. Make sure that temp files, temporary Internet files, Favorites, user documents, Outlook .pst files, and any other data that can be saved elsewhere is. Logons are guaranteed to speed up dramatically once this is done.

As for the slow network response and faster application loading on Wednesday, a good guess would be that the users who were out of the office for the day included at least some of the new users running the memory-hungry application, while the other half were probably managers. Users left with no supervisor looking over their shoulder might take the opportunity to browse the Internet, play online games or download MP3s. This theory could be tested by employing Network Monitor the next time a similar situation occurred.



 < Free Open Study > 



CCA Citrix MetaFrame XP for Windows Administrator Study Guide Exam 70-220
CCA Citrix MetaFrame XP for Windows Administrator Study Guide (Exam 70-220)
ISBN: 0072193190
EAN: 2147483647
Year: 2001
Pages: 169

flylib.com © 2008-2017.
If you may any questions please contact us: flylib@qtcs.net