<?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: Should Servers Come With Batteries?</title>
	<atom:link href="http://www.datacenterknowledge.com/archives/2009/11/27/should-servers-come-with-batteries/feed/" rel="self" type="application/rss+xml" />
	<link>http://www.datacenterknowledge.com/archives/2009/11/27/should-servers-come-with-batteries/</link>
	<description>News and analysis about data centers, cloud computing, managed hosting and disaster recovery</description>
	<lastBuildDate>Fri, 19 Mar 2010 19:34:02 +0000</lastBuildDate>
	<generator>http://wordpress.org/?v=abc</generator>
	<sy:updatePeriod>hourly</sy:updatePeriod>
	<sy:updateFrequency>1</sy:updateFrequency>
		<item>
		<title>By: Conor</title>
		<link>http://www.datacenterknowledge.com/archives/2009/11/27/should-servers-come-with-batteries/comment-page-1/#comment-10153</link>
		<dc:creator>Conor</dc:creator>
		<pubDate>Thu, 21 Jan 2010 15:30:28 +0000</pubDate>
		<guid isPermaLink="false">http://www.datacenterknowledge.com/?p=18896#comment-10153</guid>
		<description>This idea doesn&#039;t scale operationly or logically.

Problem 1: Operational monitoring - the NOC would have to monitor the charge level of every battery. This also means battery discharge tests. 100 racks would be a full time job for a NOC staff member.

Problem 2: Different discharge times, a server running at 3A will drain the battery faster than another rack at 1A. Each server in a datacenter would run at a different power consumtion rate. This would mean that uptime would be different for each server. To over come this you would have to interconnect the batteries in each server, which is in effect a centralised solution.

Problem 3: The building air conditioning would require a back up anyway to maintain the right temp. So you would only be replicating that system.</description>
		<content:encoded><![CDATA[<p>This idea doesn&#8217;t scale operationly or logically.</p>
<p>Problem 1: Operational monitoring &#8211; the NOC would have to monitor the charge level of every battery. This also means battery discharge tests. 100 racks would be a full time job for a NOC staff member.</p>
<p>Problem 2: Different discharge times, a server running at 3A will drain the battery faster than another rack at 1A. Each server in a datacenter would run at a different power consumtion rate. This would mean that uptime would be different for each server. To over come this you would have to interconnect the batteries in each server, which is in effect a centralised solution.</p>
<p>Problem 3: The building air conditioning would require a back up anyway to maintain the right temp. So you would only be replicating that system.</p>
]]></content:encoded>
	</item>
	<item>
		<title>By: John</title>
		<link>http://www.datacenterknowledge.com/archives/2009/11/27/should-servers-come-with-batteries/comment-page-1/#comment-9203</link>
		<dc:creator>John</dc:creator>
		<pubDate>Wed, 16 Dec 2009 10:25:45 +0000</pubDate>
		<guid isPermaLink="false">http://www.datacenterknowledge.com/?p=18896#comment-9203</guid>
		<description>Saw a couple comments about the AC/DC/AC/DC conversion process.

At one ISP I worked for, we colocated our servers in a Sprint telco.  For not much extra, we were able to tap their 48V, battery backed power setup.  Essentially, we were now on the same rock-solid power that the telphone company used for voice and data circuits.

Granted, our servers did not have DC power supplies.  We did have to purchase a 48v dc-ac converter to power our rack.  But I remember thinking how could it would be to have 12V power supplies.

Now, having worked with major data centers, we run a lot of 15-25A circuits.  Typically there are three circuits per rack of 110V AC.

I contemplate how thick and expensive the wiring was for 48VDC.  While we could do a bit smaller at 24V or 12V, it&#039;s still a much thicker wire to deal with.   It is much harder to deal with DC voltage than it to deal with AC voltage.  The voltage drop over distance is one of the big reasons we went with AC to the home.

However, telcos have managed for around 100 years operating everything off of DC power.  AC to the building perhaps, but then DC voltage for everything internal.  We could look to them for experience and apply it to the data centers of tomorrow.

This would also lead to supplemental power from wind/water turbines in areas that can support it.</description>
		<content:encoded><![CDATA[<p>Saw a couple comments about the AC/DC/AC/DC conversion process.</p>
<p>At one ISP I worked for, we colocated our servers in a Sprint telco.  For not much extra, we were able to tap their 48V, battery backed power setup.  Essentially, we were now on the same rock-solid power that the telphone company used for voice and data circuits.</p>
<p>Granted, our servers did not have DC power supplies.  We did have to purchase a 48v dc-ac converter to power our rack.  But I remember thinking how could it would be to have 12V power supplies.</p>
<p>Now, having worked with major data centers, we run a lot of 15-25A circuits.  Typically there are three circuits per rack of 110V AC.</p>
<p>I contemplate how thick and expensive the wiring was for 48VDC.  While we could do a bit smaller at 24V or 12V, it&#8217;s still a much thicker wire to deal with.   It is much harder to deal with DC voltage than it to deal with AC voltage.  The voltage drop over distance is one of the big reasons we went with AC to the home.</p>
<p>However, telcos have managed for around 100 years operating everything off of DC power.  AC to the building perhaps, but then DC voltage for everything internal.  We could look to them for experience and apply it to the data centers of tomorrow.</p>
<p>This would also lead to supplemental power from wind/water turbines in areas that can support it.</p>
]]></content:encoded>
	</item>
	<item>
		<title>By: Tom Carty</title>
		<link>http://www.datacenterknowledge.com/archives/2009/11/27/should-servers-come-with-batteries/comment-page-1/#comment-8966</link>
		<dc:creator>Tom Carty</dc:creator>
		<pubDate>Wed, 09 Dec 2009 21:39:04 +0000</pubDate>
		<guid isPermaLink="false">http://www.datacenterknowledge.com/?p=18896#comment-8966</guid>
		<description>IThe fire risk of batteries can be minimised by fireproof casing, if the tempature gets too large, the battery ejects and the case closes, with maybe internal CO2 emmision to close the air supply shorting the fire.

Problem solved.

As for cooling, servers need the cases to minimise dust, but could refrigerated air not be circulated to aid in cooling? (Closed and sealed door allows for smallest area for tempature control)</description>
		<content:encoded><![CDATA[<p>IThe fire risk of batteries can be minimised by fireproof casing, if the tempature gets too large, the battery ejects and the case closes, with maybe internal CO2 emmision to close the air supply shorting the fire.</p>
<p>Problem solved.</p>
<p>As for cooling, servers need the cases to minimise dust, but could refrigerated air not be circulated to aid in cooling? (Closed and sealed door allows for smallest area for tempature control)</p>
]]></content:encoded>
	</item>
	<item>
		<title>By: nate</title>
		<link>http://www.datacenterknowledge.com/archives/2009/11/27/should-servers-come-with-batteries/comment-page-1/#comment-8618</link>
		<dc:creator>nate</dc:creator>
		<pubDate>Mon, 30 Nov 2009 18:53:35 +0000</pubDate>
		<guid isPermaLink="false">http://www.datacenterknowledge.com/?p=18896#comment-8618</guid>
		<description>What about rack level DC-DC UPS systems? Rather than using batteries in each server(curious what their run time would be given the size and the draw of the typical server), with so many DC options available I find it surprising that there aren&#039;t DC-DC UPSs. From what I understand the main driver for this is to improve efficiency going from AC(utility)-&gt;DC(UPS)-&gt;AC(rack)-&gt;DC(server)</description>
		<content:encoded><![CDATA[<p>What about rack level DC-DC UPS systems? Rather than using batteries in each server(curious what their run time would be given the size and the draw of the typical server), with so many DC options available I find it surprising that there aren&#8217;t DC-DC UPSs. From what I understand the main driver for this is to improve efficiency going from AC(utility)-&gt;DC(UPS)-&gt;AC(rack)-&gt;DC(server)</p>
]]></content:encoded>
	</item>
	<item>
		<title>By: Ernie</title>
		<link>http://www.datacenterknowledge.com/archives/2009/11/27/should-servers-come-with-batteries/comment-page-1/#comment-8607</link>
		<dc:creator>Ernie</dc:creator>
		<pubDate>Mon, 30 Nov 2009 13:58:05 +0000</pubDate>
		<guid isPermaLink="false">http://www.datacenterknowledge.com/?p=18896#comment-8607</guid>
		<description>Heat from the servers will decrease the life of the batteries. If you have to increase the cooling capacity, what energy have you saved? Most of the new UPS products on the market have an energy efficiency of 93%, I.E. the Liebert NX. 
New high density servers require very specific cooling add batteries to the mixture and the situation gets a little more complicated. Data Centers that want to go &quot;green&quot; should look into a power efficiency and cooling audit. 

The audit take into consideration both the present condition and future expansion of the center. Not following cold and hot isle configuration and improper air flow accounts for up to 50% energy loss.</description>
		<content:encoded><![CDATA[<p>Heat from the servers will decrease the life of the batteries. If you have to increase the cooling capacity, what energy have you saved? Most of the new UPS products on the market have an energy efficiency of 93%, I.E. the Liebert NX.<br />
New high density servers require very specific cooling add batteries to the mixture and the situation gets a little more complicated. Data Centers that want to go &#8220;green&#8221; should look into a power efficiency and cooling audit. </p>
<p>The audit take into consideration both the present condition and future expansion of the center. Not following cold and hot isle configuration and improper air flow accounts for up to 50% energy loss.</p>
]]></content:encoded>
	</item>
	<item>
		<title>By: Paul Latham, Romonet</title>
		<link>http://www.datacenterknowledge.com/archives/2009/11/27/should-servers-come-with-batteries/comment-page-1/#comment-8602</link>
		<dc:creator>Paul Latham, Romonet</dc:creator>
		<pubDate>Mon, 30 Nov 2009 11:43:06 +0000</pubDate>
		<guid isPermaLink="false">http://www.datacenterknowledge.com/?p=18896#comment-8602</guid>
		<description>I heard the biggest problem with adding batteries to your servers is the fire risk. If a battery goes up (and they do - remember Sony&#039;s laptops habit of bursting into flames) then the resulting fire is quite difficult to put out and\or risks taking out the entire DC every time.</description>
		<content:encoded><![CDATA[<p>I heard the biggest problem with adding batteries to your servers is the fire risk. If a battery goes up (and they do &#8211; remember Sony&#8217;s laptops habit of bursting into flames) then the resulting fire is quite difficult to put out and\or risks taking out the entire DC every time.</p>
]]></content:encoded>
	</item>
	<item>
		<title>By: anon</title>
		<link>http://www.datacenterknowledge.com/archives/2009/11/27/should-servers-come-with-batteries/comment-page-1/#comment-8590</link>
		<dc:creator>anon</dc:creator>
		<pubDate>Mon, 30 Nov 2009 04:41:46 +0000</pubDate>
		<guid isPermaLink="false">http://www.datacenterknowledge.com/?p=18896#comment-8590</guid>
		<description>to answer your question, probably not.   this approach is on wrong side of scale curve.  physics dictates that  these distributed and smaller battery systems are not nearly as efficient as large scale power conditioning and distribution.  it does have resilliency benefits, but probably only for virtual servers where three 9&#039;s is sufficient.  no way that small batteries and power supplies have longevity at load vs industrial scale</description>
		<content:encoded><![CDATA[<p>to answer your question, probably not.   this approach is on wrong side of scale curve.  physics dictates that  these distributed and smaller battery systems are not nearly as efficient as large scale power conditioning and distribution.  it does have resilliency benefits, but probably only for virtual servers where three 9&#8217;s is sufficient.  no way that small batteries and power supplies have longevity at load vs industrial scale</p>
]]></content:encoded>
	</item>
	<item>
		<title>By: Greening the data center, the hard way. &#171; The Server Room</title>
		<link>http://www.datacenterknowledge.com/archives/2009/11/27/should-servers-come-with-batteries/comment-page-1/#comment-8485</link>
		<dc:creator>Greening the data center, the hard way. &#171; The Server Room</dc:creator>
		<pubDate>Sat, 28 Nov 2009 00:30:58 +0000</pubDate>
		<guid isPermaLink="false">http://www.datacenterknowledge.com/?p=18896#comment-8485</guid>
		<description>[...] } Tonight I was reading about Facebook following Google&#8217;s lead in putting batteries in servers in the data center. This just goes to show you&#8217;re never to big to have or share a boneheaded [...]</description>
		<content:encoded><![CDATA[<p>[...] } Tonight I was reading about Facebook following Google&#8217;s lead in putting batteries in servers in the data center. This just goes to show you&#8217;re never to big to have or share a boneheaded [...]</p>
]]></content:encoded>
	</item>
	<item>
		<title>By: Facebook Following Google&#8217;s Lead on Server Hardware</title>
		<link>http://www.datacenterknowledge.com/archives/2009/11/27/should-servers-come-with-batteries/comment-page-1/#comment-8452</link>
		<dc:creator>Facebook Following Google&#8217;s Lead on Server Hardware</dc:creator>
		<pubDate>Fri, 27 Nov 2009 19:50:32 +0000</pubDate>
		<guid isPermaLink="false">http://www.datacenterknowledge.com/?p=18896#comment-8452</guid>
		<description>[...] Should Servers Come With Batteries? [Data Center Knowledge] [...]</description>
		<content:encoded><![CDATA[<p>[...] Should Servers Come With Batteries? [Data Center Knowledge] [...]</p>
]]></content:encoded>
	</item>
	<item>
		<title>By: Frank Sanborn</title>
		<link>http://www.datacenterknowledge.com/archives/2009/11/27/should-servers-come-with-batteries/comment-page-1/#comment-8448</link>
		<dc:creator>Frank Sanborn</dc:creator>
		<pubDate>Fri, 27 Nov 2009 19:25:54 +0000</pubDate>
		<guid isPermaLink="false">http://www.datacenterknowledge.com/?p=18896#comment-8448</guid>
		<description>These changes in design could be beneficial for people and business who live and operational in less developed parts of the world..  In fact if we were to design our systems to run under condition there.  I bet we would have the most efficient designs available at a lower overall cost and and much lower operational cost.</description>
		<content:encoded><![CDATA[<p>These changes in design could be beneficial for people and business who live and operational in less developed parts of the world..  In fact if we were to design our systems to run under condition there.  I bet we would have the most efficient designs available at a lower overall cost and and much lower operational cost.</p>
]]></content:encoded>
	</item>
</channel>
</rss>
