Real Problems – Real Solutions - February, 2002

Do you have a story to tell? Submit it to us and perhaps you can win a TechNet mug!

Attention to Detail is the Great Divider

The key factor in network administration is attention to detail. The following Real Problems – Real Solutions article is based on the top 10 tips submitted by Carl Fransen. Carl raises some very good points and to help administrators we added some extra information and links to supplement the original submission.

Carl starts by stating what the goal for all administrators should be. "Your goal when administering the network is to provide your clients with reliable and dependable access to their services and data 7 by 24." So keep this in mind as we go through the other tips.

So where do we start? Well, Carl talks about security and the latest updates. "Keep up to date with security updates and, when need be, service packs. Regularly apply Microsoft security updates to all servers as soon as they come out. When a new service pack comes out, test it before you apply it to the environment." So how do I know where to find these updates and how do they come to me? Well the best place to start is the TechNet Security web site. To have this information sent to you when updates occur subscribe to the alerter server at www.microsoft.com/technet/security/bulletin/notify.mspx. As for the service packs, all of these are posted at support.microsoft.com/support/servicepacks/default.asp.

Carl goes on to say "Good backups = Good job security". "Having a proven and tested disaster recovery plan is essential. It is your job to maintain all user data, including during and after a disaster. All Backups are tested regularly and tapes stored offsite. Rescue diskettes are all up to date and placed in the A: drive of all servers".

Remember our goal about uptime. Have a look at the How To articles.

What else does Carl suggest; well he recommends "Check the logs daily, event logs on each server will tell you of upcoming problems and current issues before users discover the problems. Take time in the morning to check each server's logs and test out specific systems". There are resource kit tools to help, try Dumpel to dump the log to a tab-separated text file and well as using it to filter. The operations guide has lots of information on preventative maintenance and troubleshooting, have a look online at www.microsoft.com/technet/prodtechnol/windows2000serv/reskit/w2rkbook/CoreTechnologies.asp This bring us nicely to what Carl calls "Documentation is Good", he expands on that with "Document all unique features of the network". Keep all of the diagrams, IP ranges, etc. up to date. Write the documentation that another NT Administrator would be able to assume your job. Documentation will also help you enforce system procedures such as passwords, granting access, etc Keep track of events in your logs." You can use products like Access to hold event information for easy retrieval later. Use tools like VISIO to help map and maintain your networks (https://www.microsoft.com/technet/prodtechnol/visio/visio2000/maintain/stepdiag.mspx). Too many things happen on today's network that go undocumented.

Carl then focuses on servers; how, what and who uses them. Remember, "Servers are Servers NOT workstations" Carl says. "Microsoft makes it easy for someone to use a server as a workstation. However, servers are only to be used as servers. This means that no one runs ANY applications locally on the servers. Any applications that need to be run will be run from a workstation". It's also worth noting that the same application run on a server and a workstation will perform differently due to the nature of the priority configuration. Foreground tasks on workstations are given higher priority than the same task on a server, which must share the priority with all other services running. On the performance front Carl reminds us "Servers use server class hardware.Servers can run on any hardware, however to provide the users with the maximum possible performance and uptime always purchase the appropriate hardware from a Tier 1 (preferred) or 2 vendor. The servers should be placed in a secured area where only NT Administrators have physical walk up access. No users have the right to physically touch a server". What about the subject of access? What does Carl tell us on that front? "Administration access only for NT Administrators! Admin access is given based on skill and job responsibility. Only those who have the responsibility and the skills to directly maintain the systems on a day to day basis should be considered. Hierarchical company rank, laziness, or ignorance does not grant a user domain admin access." Once you have determined who is an administrator and who is not then, as Carl tells us, "All NT Administration should be done from desktops NOT servers. Microsoft provides a very good selection of client based admin utilities for all of their products. Everything from adding users to Exchange, forcing domain replication, disabling accounts, setting scheduled jobs, copying data, etc is done from the desktop. The goal is to never have to leave your desk to administer an NT server."

Ok, one final word from Carl, I couldn't leave this one out and it's a good one to end on. "Reboots do NOT fix servers. Reboots should only be used for scheduled maintenance or a complete disaster. There is a reason why something is misbehaving and we are provided with tools to fix the problem, reboots temporarily make the problem go away, they do not fix problems". How many times do you reboot because the server is low on memory? There is probably a reason why this has happened and usually it's a memory leak from an application. If you have these scenarios use the Performance Monitor to check memory usage. There are some good articles on TechNet regarding this subject; https://support.microsoft.com/default.aspx?scid=kb;
en-us;302558&sd=tech to start with.

I'd like to thank Carl for his submission and his insight into System administration.