<?xml version="1.0" encoding="UTF-8"?><rss version="2.0"
	xmlns:content="http://purl.org/rss/1.0/modules/content/"
	xmlns:dc="http://purl.org/dc/elements/1.1/"
	xmlns:atom="http://www.w3.org/2005/Atom"
	xmlns:sy="http://purl.org/rss/1.0/modules/syndication/"
		>
<channel>
	<title>Comments on: Switches and Cables</title>
	<atom:link href="http://etbe.coker.com.au/2008/08/22/switches-and-cables/feed/" rel="self" type="application/rss+xml" />
	<link>http://etbe.coker.com.au/2008/08/22/switches-and-cables/</link>
	<description>Linux, politics, and other interesting things</description>
	<lastBuildDate>Thu, 09 Feb 2012 01:09:24 +0000</lastBuildDate>
	<sy:updatePeriod>hourly</sy:updatePeriod>
	<sy:updateFrequency>1</sy:updateFrequency>
	<generator>http://wordpress.org/?v=3.3.1</generator>
	<item>
		<title>By: etbe</title>
		<link>http://etbe.coker.com.au/2008/08/22/switches-and-cables/comment-page-1/#comment-15580</link>
		<dc:creator>etbe</dc:creator>
		<pubDate>Sun, 31 Aug 2008 07:30:01 +0000</pubDate>
		<guid isPermaLink="false">http://etbe.coker.com.au/?p=721#comment-15580</guid>
		<description>Steve: You assume a centralised and competent network admin team that actually controls everything.  All it takes is one guy to go in there to fix a server in an emergency and change a few cables without documenting it and then it becomes a total mess.  The patch panel and big central switch idea means that it&#039;s difficult and complicated to do things right, easy to do things wrong, and time consuming to fix problems once they have been created.

I think that there should be a design goal to make it easy to do things right and easy to fix problems once they occur.</description>
		<content:encoded><![CDATA[<p>Steve: You assume a centralised and competent network admin team that actually controls everything.  All it takes is one guy to go in there to fix a server in an emergency and change a few cables without documenting it and then it becomes a total mess.  The patch panel and big central switch idea means that it&#8217;s difficult and complicated to do things right, easy to do things wrong, and time consuming to fix problems once they have been created.</p>
<p>I think that there should be a design goal to make it easy to do things right and easy to fix problems once they occur.</p>
]]></content:encoded>
	</item>
	<item>
		<title>By: Steve Parker</title>
		<link>http://etbe.coker.com.au/2008/08/22/switches-and-cables/comment-page-1/#comment-15575</link>
		<dc:creator>Steve Parker</dc:creator>
		<pubDate>Sat, 30 Aug 2008 22:49:09 +0000</pubDate>
		<guid isPermaLink="false">http://etbe.coker.com.au/?p=721#comment-15575</guid>
		<description>I prefer the rack installations which use a patch panel at the top (or bottom) of each rack, leading to whatever (usually large) switch(es) as appropriate.

This keeps the network tech within the rack as simple as possible (it&#039;s just a wire), keeps the labelling simple (&quot;Server gandalf, NIC #2  Switch 4, Port 43&quot;), and allows for centralised management of the network infrastructure (allowing you to use things like tagged VLANs usefully too). Of course, that assumes a centralised and competent network admin team; without that, you may as well try to keep everything as flat and simple as possible, still try to avoid Alex&#039;s &quot;Issue #2&quot; above.

With the above example, you maintain docs saying that in Cab 3, Gandalf&#039;s NIC2 goes to local Patch Panel port 23, which is connected to Cab 1&#039;s Patch Panel port 3.23, which goes to Switch 4 Port 43.</description>
		<content:encoded><![CDATA[<p>I prefer the rack installations which use a patch panel at the top (or bottom) of each rack, leading to whatever (usually large) switch(es) as appropriate.</p>
<p>This keeps the network tech within the rack as simple as possible (it&#8217;s just a wire), keeps the labelling simple (&#8220;Server gandalf, NIC #2  Switch 4, Port 43&#8243;), and allows for centralised management of the network infrastructure (allowing you to use things like tagged VLANs usefully too). Of course, that assumes a centralised and competent network admin team; without that, you may as well try to keep everything as flat and simple as possible, still try to avoid Alex&#8217;s &#8220;Issue #2&#8243; above.</p>
<p>With the above example, you maintain docs saying that in Cab 3, Gandalf&#8217;s NIC2 goes to local Patch Panel port 23, which is connected to Cab 1&#8242;s Patch Panel port 3.23, which goes to Switch 4 Port 43.</p>
]]></content:encoded>
	</item>
	<item>
		<title>By: etbe</title>
		<link>http://etbe.coker.com.au/2008/08/22/switches-and-cables/comment-page-1/#comment-15443</link>
		<dc:creator>etbe</dc:creator>
		<pubDate>Fri, 22 Aug 2008 22:22:18 +0000</pubDate>
		<guid isPermaLink="false">http://etbe.coker.com.au/?p=721#comment-15443</guid>
		<description>Alex: 1)  In my experience users are more than willing to power-cycle devices.  My experience (being involved in supporting Internet gateways for a number of companies) is that when things stop working everything in the area ends up power cycled, the DSL modem, the router, and the PC on the desk (in that order).  Sometimes it&#039;s just a remote web site that has gone down or the company has been black-listed for having a virus-ridden Windows box that sent out spam, but everything gets rebooted anyway.

2)  Some users might, but there are always a few computer hobbyists in every office.

3)  I agree that it would be bad to seek out switches that need to be power cycled every month.  But you seem to end up with some anyway.  If the switch in question cost $5000 then you are stuck with it and just need to automate the reboot.  If it cost $100 then you have a choice of having the nearest person power-cycle it (which they probably won&#039;t even tell you about) or just spending another $100.  Both options are viable.

I agree with your point about some of the wiring being good.  But while it may be technically good to have an ordered and labeled set of 100 cables, having an architecture that allows an ordered and labeled set of 5 cables is better.

Let&#039;s not forget that organisation structures change over time.  While you might have a team of people who can do a really good wiring job now, next year you might have people who are less skilled - and then things go horribly wrong.  Designing things so that they can be maintained by someone with less skill than you is a good idea for many reasons.

In regard to your second points:

1) So you have several switches in one box?  I&#039;d rather just have several switches so that one failure CAN&#039;T cause other failures.

2) Well, it&#039;s not as if that doesn&#039;t happen already...

3) Sure, and someone doesn&#039;t work while waiting for the person with the wire testing tools to arrive and they disturb others while waiting.</description>
		<content:encoded><![CDATA[<p>Alex: 1)  In my experience users are more than willing to power-cycle devices.  My experience (being involved in supporting Internet gateways for a number of companies) is that when things stop working everything in the area ends up power cycled, the DSL modem, the router, and the PC on the desk (in that order).  Sometimes it&#8217;s just a remote web site that has gone down or the company has been black-listed for having a virus-ridden Windows box that sent out spam, but everything gets rebooted anyway.</p>
<p>2)  Some users might, but there are always a few computer hobbyists in every office.</p>
<p>3)  I agree that it would be bad to seek out switches that need to be power cycled every month.  But you seem to end up with some anyway.  If the switch in question cost $5000 then you are stuck with it and just need to automate the reboot.  If it cost $100 then you have a choice of having the nearest person power-cycle it (which they probably won&#8217;t even tell you about) or just spending another $100.  Both options are viable.</p>
<p>I agree with your point about some of the wiring being good.  But while it may be technically good to have an ordered and labeled set of 100 cables, having an architecture that allows an ordered and labeled set of 5 cables is better.</p>
<p>Let&#8217;s not forget that organisation structures change over time.  While you might have a team of people who can do a really good wiring job now, next year you might have people who are less skilled &#8211; and then things go horribly wrong.  Designing things so that they can be maintained by someone with less skill than you is a good idea for many reasons.</p>
<p>In regard to your second points:</p>
<p>1) So you have several switches in one box?  I&#8217;d rather just have several switches so that one failure CAN&#8217;T cause other failures.</p>
<p>2) Well, it&#8217;s not as if that doesn&#8217;t happen already&#8230;</p>
<p>3) Sure, and someone doesn&#8217;t work while waiting for the person with the wire testing tools to arrive and they disturb others while waiting.</p>
]]></content:encoded>
	</item>
	<item>
		<title>By: Alex</title>
		<link>http://etbe.coker.com.au/2008/08/22/switches-and-cables/comment-page-1/#comment-15442</link>
		<dc:creator>Alex</dc:creator>
		<pubDate>Fri, 22 Aug 2008 22:01:49 +0000</pubDate>
		<guid isPermaLink="false">http://etbe.coker.com.au/?p=721#comment-15442</guid>
		<description>I think you overestimate the average end-user&#039;s willingness to do any work that is not part of their job description. In an ideal world, switches for each group of desks would work, but in the real world there are problems:

1) In any company over about 20 employees, the vast majority of employees will have zero vested interest in the company, and will therefore either be reluctant or utterly unwilling to do anything outside of their job description, even if it just involves power-cycling a switch.

2) End users are downright terrified of technology. They&#039;ll treat the switch on their desk as though it&#039;s an angry cobra and will refuse to get near it.

3) If the switch needs to be reset every month, the IT department in the company is not doing its job.

Wiring is a dirty fact of modern technology. Some of it is awful, and some of it just LOOKS awful. (Many of the pictures in the blog you referenced were of REALLY good wiring jobs, they just look bad because there are so many wires.) The fact of the matter is that wiring is an &quot;expense&quot; of technology, and it is up to us, as technology professionals, to deal with that wiring so that our end users can focus on their jobs.

The basic business responsibilities perspective notwithstanding, however, there are still a few purely technical issues:

1) Once an organization grows beyond a 24-port switch, it&#039;s time to invest in good switches that have individually configurable 8-12 port modules. A mis-configuration will now only knock out a few users, and those modules can be swapped out quickly if necessary.

2) Having individual switches on desks encourages quick and dirty solutions like stringing switches together. Put any more than 3 switches in a line and you start getting nasty network issues.

3) With a proper naming scheme for the wires and a few simple wire testing tools it&#039;s quite easy to diagnose wiring issues.

Just my observations. Most of the ideas you present are very good overall, but in my experience, they just don&#039;t work nearly as well in PRACTICE as it seems that they should. :)</description>
		<content:encoded><![CDATA[<p>I think you overestimate the average end-user&#8217;s willingness to do any work that is not part of their job description. In an ideal world, switches for each group of desks would work, but in the real world there are problems:</p>
<p>1) In any company over about 20 employees, the vast majority of employees will have zero vested interest in the company, and will therefore either be reluctant or utterly unwilling to do anything outside of their job description, even if it just involves power-cycling a switch.</p>
<p>2) End users are downright terrified of technology. They&#8217;ll treat the switch on their desk as though it&#8217;s an angry cobra and will refuse to get near it.</p>
<p>3) If the switch needs to be reset every month, the IT department in the company is not doing its job.</p>
<p>Wiring is a dirty fact of modern technology. Some of it is awful, and some of it just LOOKS awful. (Many of the pictures in the blog you referenced were of REALLY good wiring jobs, they just look bad because there are so many wires.) The fact of the matter is that wiring is an &#8220;expense&#8221; of technology, and it is up to us, as technology professionals, to deal with that wiring so that our end users can focus on their jobs.</p>
<p>The basic business responsibilities perspective notwithstanding, however, there are still a few purely technical issues:</p>
<p>1) Once an organization grows beyond a 24-port switch, it&#8217;s time to invest in good switches that have individually configurable 8-12 port modules. A mis-configuration will now only knock out a few users, and those modules can be swapped out quickly if necessary.</p>
<p>2) Having individual switches on desks encourages quick and dirty solutions like stringing switches together. Put any more than 3 switches in a line and you start getting nasty network issues.</p>
<p>3) With a proper naming scheme for the wires and a few simple wire testing tools it&#8217;s quite easy to diagnose wiring issues.</p>
<p>Just my observations. Most of the ideas you present are very good overall, but in my experience, they just don&#8217;t work nearly as well in PRACTICE as it seems that they should. :)</p>
]]></content:encoded>
	</item>
</channel>
</rss>

