Fiber from home office switch to equipment rack?

MinPins

Limp Gawd
Joined
Aug 30, 2007
Messages
368
I just installed a 10 port Cisco SG300-10MP in my office, 9 of the ports are already in use. Connected to it are

  • PC
  • Laptop
  • Notebook
  • WIFI AP
  • B&W laser printer
  • Color inkjet printer
  • IP camera
  • Squeezebox SB3
  • Cat5e connection to basement HP 1810G-24 switch
All of these devices are connected via a single Cat5e cable that was installed when the house was built. I had to re-terminate both ends of it when I moved in because the existing terminations were done very poorly. I have no idea how or where the cable is routed, nor what the quality of this Cat5e cable is. It did pass a test using my cable tester.

This single Cat5e is connected to my main HP switch in the basement equipment rack.

Here is what I am pondering... :confused:

Both of these switches have mini-GBIC ports that can utilize SFP transceivers and fiber. I priced out 2 MGBSX1 compatible modules along with a 30M multimode fiber patch 'cable' online at under $150. Although it would require at least a couple hours of work (including digging around in the attic and crawl space and drilling some holes), but running the fiber to the basement is very doable.

My question to my far more experienced forum members is if the performance gains would be worth the effort?

My server is in the basement, and despite having an Intel GB NIC in my PC I am seeing read speeds of ~23MBs and writes of ~49MBs to/from my it. My server has teamed GB NICs (802.3ad LACP Layer 2) connected to trunked (LACP) ports on my HP switch.

Would running the fiber appreciably increase my available bandwidth and throughput?

p.s. I have absolutely 0 experience with fiber, but I do own an impressive assortment of tools and have experience wiring houses and running Cat and coax cables.

Thank you for your input.
 
iometer.
Speeds are from office PC to a mapped share on my NAS server over the Cat5e described above.

Ran tests again and averaged 28.36 MBs read, 53.24 MBs write over 5 tests with this setup.
 
Last edited:
Try running Iperf or Jperf if you prefer a GUI and seeing what you get with that.
 
If you really doubt the single cat5 cable, you could *shudder* get a new one for $10, and test and see if it's any better.

Because here's the shocker - the fiber's only 1Gb too. So if you have a fully working Gb line, it doesn't matter if it's copper or glass. it's still 1Gb.

Your problem is the network configuration itself. You're getting gigabit speeds, but you're sharing the line with 8 other devices. Fiber isn't going to fix that.
 
The amount of work involved to install either fiber or copper would be nearly identical.
As I mentioned I have no experience using fiber, but if it's bandwidth and throughput are the same as copper then it would be silly to put in the expensive fiber instead of a Cat 6 cable.

I downloaded Jperf the other day, will give it a try later today.
 
Well, first thing is first, what speeds do you see if you take your PC down into your basement and hook it up direct into the switch?

Typically you use fiber where you need more speed or greater distance than copper will get you. Fiber won't be quicker than copper at the same speed "because it's fiber".

Also what is the disk layout in your server? It doesn't matter what your network configuration and connectivity is if disk is the bottleneck.
 
This sounds like a cool project.. Reminds me of a cable run out at the local airport, i thought it was all high speed fiber, but it was only a 10/100..
 
Back
Top