0

I've been tasked with debugging why my company's network connection is running slow. Having no previous experience in this field I have absolutely no clue where to begin or what tools to use.

The comapny has around 30 employees and the entire network goes through a single switch (from what I've gathered so far).

But how would I approach this task? I'm guesing there's a ton of tools I can use to monitor throughput at various points and such, but having no experience I was wondering if someone could give me any ideas as to where I'd start.

Thanks in advance.

2
  • 1
    Can you edit to include some more details? What model is the switch? What computers / servers / devices are connected, and what kind of things do they do? (Office work, video editing, etc). And what are the symptoms of the slowness? Slow server access? intranet access? internet access? file copying? Logons only? Printing only? Everything, but only from one office? Oct 29, 2014 at 20:30
  • Sorry for not being more specific, as I only recently started working for this company I have yet to experience it fully myself. From what I've understood the symptoms are pretty much everything currently, internet is slow and intranet too. But I see your point ;) Will gather more information and formulate a better question
    – Rasmus
    Oct 29, 2014 at 20:54

1 Answer 1

3

"having no experience" is a bad starting point for complex issues. Unfortunately, more often than not, performance issues tend to be complex. And there is no single diagnostic toolset which, when run, would tell you to "swap cable connecting to port 13 and change the server's registry value HKLM\foo to bar".

You typically would start by defining "slow" and "normal operation" in more formal terms than a gut feeling. Transfer rates for a given sets of data and protocols, round-trip times to defined destinations and packet loss / retransmit percentages usually are good metrics for this.

After you have defined the thresholds, you would need to figure out the conditions in which your network is under-performing. Is it affecting certain stations? Just at isolated points in time? Just for certain protocols, destinations or payloads? Obviously, gathering this information would require active monitoring of the defined metrics throughout your network or at least a sufficiently large control sample of data. At this point you probably will have 2-5 hypotheses of what might be going wrong which you might test by tweaking the relevant configuration and looking at the data collected before and after the change.

You might need to replace components on speculation, update firmwares, drivers, software versions, change protocol settings and run and evaluate network traces in the process. This certainly is not a task for people with low frustration tolerance as there likely will be many setbacks, especially since you don't have a comprehensive model of how things should be running when you're starting.

1
  • I couldn't agree more that my starting point isen't ideal :/ Thanks for your answer though, hopefully it will be helpful on what sounds like a long journey of debugging this issue ;)
    – Rasmus
    Oct 29, 2014 at 20:49

Not the answer you're looking for? Browse other questions tagged .