|
Home > Archive > Server + > September 2002 > Scenario Q
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]
|
|
| twister166 2002-08-31, 11:54 am |
| You are a new hired server manager, on the first day of your new job, you walk into the server room, you saw:
- 3 Prolient 1600R rack servers stacked on each other on a table.
- 3 Moinitors and 3 sets of keyborad and mouse setting on the left, top and the right of the server.
- Look behind the servers, all equipemts are pluged into two power strips that are pluged into the wall outlet.
- 25 footer patch cables ran accross the room from the opposit wall where 3 8-port Linksys hubs that are uplinked and the pacth panel.
What are your recommendations to the MIS director under the following different setups:
- Budget is tight, less than US $1000 to spend.
- Moderate budget, less than US $10,000 to spend.
- Good budget, less than us $150,000 to spend.
I know our answers may not be exact the Comptia answers, but our real life experiences may help the new comers outside of the book. I don't have a CompTia answer, but open discussion will be good for everyone, me included. | |
| Tech Ranger 2002-08-31, 6:44 pm |
| To get things started we can put up a rack, get a KVM switch, and a UPS.
Also, we could put 2 of the monitors, KB's and mice to use elsewhere or return them.
I would integrate the hubs, the servers, the KVM switch, the remaining monitor, the UPS, the KB, and mouse onto the rack. If you want to take things further, we would need a lot more info. We could talk about storage solutions. We might want to replace the hubs with a switch or router. This all depends on the objectives here. | |
| twister166 2002-09-01, 12:50 pm |
| I think that is what CompTia's solution to start as well.
Get the UPS, rack, KVM's and fix the cables to start up.
Let's the put scenario a bit further, you have noticed that hub's collision light is constantly blinking, and at time it stays on for 2 to 3 seconds, what would you do first? | |
| Tech Ranger 2002-09-01, 7:00 pm |
| I would fire up the Network Analyzer and pull out the records on any baselines that had been done. | |
| twister166 2002-09-04, 5:40 pm |
| quote:
Originally posted by Tech Ranger
I would fire up the Network Analyzer and pull out the records on any baselines that had been done.
Me too, I will get my Sniffer, and check for broadcast storm, re-transmition rate, runt and giants. Try to determine if the collision is from the high LAN utilization or a bad NIC/Cable.
Let's take this down more:
After you exam the LAN Analyzer, you noticed the Ethernet Utilization sporatic between 0 to 80%, average utilization is 28%. You also found server IP address are associated with many retransimition errors, what do you do next? | |
| Tech Ranger 2002-09-04, 6:48 pm |
| Run diagnostics on the NIC associated with the suspect IP address. | |
| twister166 2002-09-04, 7:29 pm |
| Well, I guess it is just you and I are playing this game... 
Well, I will do is run a week long baseline monitoring and see the average utilization level, due to the collision, the retrans is normal. If the average utilization maintain high at 28%, I probably will go with replacing the HUBS. By replace the HUBS to swithes, I can create seperate collision domain and increase the bandwidth at the same time. The collision are also possibly caused by the crosslinks of 3 HUBS caused by contension of single port to the server hub.
I will probably replace the hubs with a 24-port switch.
Let's say, if we put in a new 24-port switch. What is the first thing you do to the servers? | |
| Tech Ranger 2002-09-04, 7:43 pm |
| Give each one dedicated bandwidth. | |
| twister166 2002-09-04, 8:04 pm |
| quote:
Originally posted by Tech Ranger
Give each one dedicated bandwidth.
Well, actually that is done within the function of the switch, server has no choice.
I will make sure that the each NIC setting is at auto negotiate for the duplex. If I can force the 100 Full Duplex, I force it instead of auto detect. If it is Win2k, it may not need a reboot, but if the server is NT4 a reboot is likely.
Now the server is running at 200 full-duplex with a new switch, check with the LAN Analyzer, the average utilization is now 4%, retrans are minimal. Users are happy! 
Remeber the brand new UPS that we put in? Well, the location is in South FL, and thunderstorm has been clock work every day, 2 days in a roll, lighting has come close to the building, and the UPS had just dropped dead twice when large light struck close to the building. The servers went down as a result. What is the cause, and what can we do about it? | |
| Tech Ranger 2002-09-04, 8:15 pm |
| The best that I can come up with is that somehow the lightning affected the grounding of the UPS in a detrimental way. We need to look for a new way to ground. I guess if you can't go down, you have to go up. | |
| twister166 2002-09-04, 8:23 pm |
| You are correct! The building wiring has poor grounding, the surge produced too much heat and the UPS decided to turn itself off, so itself does not get damaged. Call the electrician and have the wire inspected, and put a good surge protector between the wall outlet and the UPS may help for the short term, you would much rather brown out the surge protecter instead of the UPS.
Do you want to keep on going?
| |
| Tech Ranger 2002-09-04, 8:28 pm |
| I wish someone else would join the fun, but I am still up for it. | |
| twister166 2002-09-04, 8:38 pm |
| Ok then, here we go!!!!! Broooommmm...
After the boss has seen the UPS gone down and have electrician check the wire. Now he wants a Server Maintanance Plan that includes an UPS section, what should be in the UPS section? | |
| Tech Ranger 2002-09-04, 8:48 pm |
| Mfr., vendor, warranty info, battery replacement history, maintenance history, VA rating, load info, whether its stand-by or in-line. Specs regarding the management software. | |
|
|
| Tech Ranger 2002-09-04, 9:11 pm |
| Number of outlets, transfer time, absorbtion capacity of surge protection circuitry, physical dimensions. | |
| twister166 2002-09-05, 6:46 am |
| We should verify how long (in minutes) the equipments will stay powered on before it shuts down, and verify it does shut down ok. We also need to include a regular UPS power test monthly.
I will continue this when I get back from the A+ exam. | |
| ordpeople 2002-09-05, 5:51 pm |
| I think Comptia would want you to come up with an installation plan first. |
|
|
|
|
