none
Will the Server StandBy? RRS feed

  • Question

  • Will it always run, or will it Wake on Lan? It seems like it would be unnessary to run fulltime.
    Monday, March 5, 2007 12:48 AM

Answers

  • WHS is an "always on" server.  Personally, I like this as there are a lot of useful things that an always available server can do.  (Unfortunately, many will have to wait for future releases.  Unless you're willing to try unsuported configurations.  )

    The objections I see to leaving a server on seem to be either power or reliability.  The small form-factor server that I'm testing WHS on idles around 30 watts, so I'm contributing less to global warming than Mr. Gore's jet.  The reliability issue got really interesting last week when several studies of hard drive usage said that hard drive failures are related more to the make and model of the drive than to temperature (within reason) and usage.  This makes a case for always on servers.

    obtw, I have a Buffalo NAS box configured to shutdown when idle.  However, when any computer on our network is running, the NAS unit starts up and shuts down about every 20 minutes.  (ARP messages make wake-up-on-lan not very practical.) I'd wager that this is much harder on the drives than always on.

    Murray

    Monday, March 5, 2007 1:18 AM

All replies

  • WHS is an "always on" server.  Personally, I like this as there are a lot of useful things that an always available server can do.  (Unfortunately, many will have to wait for future releases.  Unless you're willing to try unsuported configurations.  )

    The objections I see to leaving a server on seem to be either power or reliability.  The small form-factor server that I'm testing WHS on idles around 30 watts, so I'm contributing less to global warming than Mr. Gore's jet.  The reliability issue got really interesting last week when several studies of hard drive usage said that hard drive failures are related more to the make and model of the drive than to temperature (within reason) and usage.  This makes a case for always on servers.

    obtw, I have a Buffalo NAS box configured to shutdown when idle.  However, when any computer on our network is running, the NAS unit starts up and shuts down about every 20 minutes.  (ARP messages make wake-up-on-lan not very practical.) I'd wager that this is much harder on the drives than always on.

    Murray

    Monday, March 5, 2007 1:18 AM
  • Thanks MurMan99, I understand the benefits. I am aware of the WOL issues and was wondering if they were going to be troublesome with WHS.
    Thursday, March 8, 2007 12:47 AM
  •  MurMan99 wrote:

    WHS is an "always on" server.  Personally, I like this as there are a lot of useful things that an always available server can do.  (Unfortunately, many will have to wait for future releases.  Unless you're willing to try unsuported configurations.  )

    The objections I see to leaving a server on seem to be either power or reliability.  The small form-factor server that I'm testing WHS on idles around 30 watts, so I'm contributing less to global warming than Mr. Gore's jet.  The reliability issue got really interesting last week when several studies of hard drive usage said that hard drive failures are related more to the make and model of the drive than to temperature (within reason) and usage.  This makes a case for always on servers.

    I think that 30W is way too much for a sensible home server. Imagine for example million people leaving their WHS's running at home and count how many watts that is. And I think that million people is quite much less than the actual amount of users probably purchase this kind of thing. And that would require their servers to be at the low end of power consumption. MS should find a way or urge manufacturers to find a way to make a server run with much lower power consumption when idle. Or then there should be a way to wake it up reliably when needed. Wake on lan doesn't seem to do that. I also didn't understand your reasoning how make and model of a hard drive make a case for always on servers?

    Other reason for shutting idle servers down at home is the increased fire risk. Computers are no different from washing machines, etc. which should not be left running alone at home.

    Tuesday, March 13, 2007 9:29 PM
  • Hello,

    Why not something like that : hibernating when no computer is connected and wake up when WHS find a least one computer on the lan ? Is it possible ?

    I'm agree with you ; computer is dangerous... and usually, we need it on evening and on week-end... not 24/24, 7d/7d...

    Bye,

    Tuesday, March 13, 2007 9:47 PM
  •  Shannin wrote:
    I think that 30W is way too much for a sensible home server.

    If you take the position that 30W is not sensible, then what is your target?  For me it is sensible.  If you tour the average house of someone with several computers you will find lots of 'always on' devices.  Any device with a remote control or a clock is always on.  Nearly all devices with a wall-wart power supply is always on because the power supply doesn't fully switch off.  We're talking 1-10W per device.  A doorbell transformer takes about 3W when idling.  My garage door opener idles around 6W.  There's the router and cable modem at 10W each.  Alarm system at 15W.  Even a night light is 3 - 7.5W.

    My point is that the power consumption of all of these devices adds up.  The total is much greater than 30W in my home.  I'm replacing an 18W Buffalo Linkstation and a 10W print server with WHS.  Therefore I concluded that my 30W server is sensible.   You may not agree, but you have the reason for my conclusion.

     Shannin wrote:
    Imagine for example million people leaving their WHS's running at home and count how many watts that is.
     Any small positve integer multiplied by a big number yields a bigger number.  What's missing from your analysis is perspective.

     Shannin wrote:
    Other reason for shutting idle servers down at home is the increased fire risk. Computers are no different from washing machines, etc. which should not be left running alone at home.

    Fire risk from computers and washing machines?  Yikes.  Show me the data!  (Laptop battery fires excluded.)

     Shannin wrote:
     I also didn't understand your reasoning how make and model of a hard drive make a case for always on servers?

    That wasn't what I said.  I was referencing studies of hard drive reliability that concluded that hard drives were more reliable when operated continuously at a relatively hot temperature. This suggests that always on servers will have higher disk reliablity than servers that switch the hard drives off or power down completely.   The study also concluded that the make and model was a bigger factor in detemining the reliablity than most environmental factors.  This is why I mentioned make and model of the hard drive.

    Tuesday, March 13, 2007 11:53 PM
  •  MurMan99 wrote:
    We're talking 1-10W per device.  A doorbell transformer takes about 3W when idling.  My garage door opener idles around 6W.  There's the router and cable modem at 10W each.  Alarm system at 15W.  Even a night light is 3 - 7.5W.

    These are very interesting numbers! How did you measure them? What equipment did you use?

    Erik.

    Wednesday, March 14, 2007 5:59 AM
  • Most of them were measured using an RMS wattmeter.   They are somewhat specialized - you won't find them at Radio Shack.   (RMS, root mean square, is the mathematical term for dealing with ac waveforms so that the measurement is equivalent to a dc measurement.  For devices like switching power supplies, that don't have sinusoidal current waveforms, it's imperative to use a true RMS meter.)  If you measure voltage and current separately and multiply them, you'll get Volt-Amperes (VA) which is not the same as watts.   VA will be a higher number than watts (except for resistive loads,like heaters)  and doesn't measure the power consumption of the device.

    The night light value came from the stamped value on the bulb.  For most electronic devices, the power in watts on the nameplate is usually the maximum rated power that the power supply can safely supply.  The actual power consumption will be lower, sometimes much lower.

    obtw, all of these measurements came about from a serious discussion last year about the true cost of the power "vampires" that suck power in your home.

    Wednesday, March 14, 2007 6:35 AM
  •  MurMan99 wrote:
     Shannin wrote:
    I think that 30W is way too much for a sensible home server.

    If you take the position that 30W is not sensible, then what is your target?  For me it is sensible.  If you tour the average house of someone with several computers you will find lots of 'always on' devices.  Any device with a remote control or a clock is always on.  Nearly all devices with a wall-wart power

    I am very aware that there are several devices already consuming power all the time. That doesn't mean we have to make more of them. Since you know that standby power consumption is a huge waste of energy, why don't you see that this is just one more device like that? Sensible standby products consume less than 1W in standby mode. http://standby.lbl.gov/DATA/1WProducts.html

     MurMan99 wrote:

    My point is that the power consumption of all of these devices adds up.  The total is much greater than 30W in my home.  I'm replacing an 18W Buffalo Linkstation and a 10W print server with WHS.  Therefore I concluded that my 30W server is sensible.   You may not agree, but you have the reason for my conclusion.

    I am not talking about you but in general. Reducing is always good (which happened in your case), but most of the people won't be reducing devices with WHS, just adding one more to the pile. And if the one used is an old discarded PC, you can imagine what it spends because it probably doesn't support even the basic power saving modes.

     

     Shannin wrote:
    Imagine for example million people leaving their WHS's running at home and count how many watts that is.

     Any small positve integer multiplied by a big number yields a bigger number.  What's missing from your analysis is perspective.

    You just proved my point that even if one home adds just a small amount to power consumption (I still remember that your consumption decreased), counting them all together spends a lot of energy.

     Shannin wrote:
    Other reason for shutting idle servers down at home is the increased fire risk. Computers are no different from washing machines, etc. which should not be left running alone at home.

    Fire risk from computers and washing machines?  Yikes.  Show me the data!  (Laptop battery fires excluded.)

    Computers are not a huge source of electrical fires, but they still are. The risk differs in every country, you can check your own from the net.  Some data from somewhere: http://www.usfa.dhs.gov/citizens/all_citizens/home_fire_prev/electrical.shtm . Imagine old devices stored in a closet (since they do not require keyboard etc.) in the middle of dust and they are almost never cleaned up. I have seen computers burning from power supply (flames coming out) after they have gathered enough dust inside. This was during my work in a computer repair shop for several years. After that I have seen for example UPS devices almost melting down because of internal fire caused by overvoltage (which in turn was caused by a tree falling on a power line).

     Shannin wrote:
     I also didn't understand your reasoning how make and model of a hard drive make a case for always on servers?

    That wasn't what I said.  I was referencing studies of hard drive reliability that concluded that hard drives were more reliable when operated continuously at a relatively hot temperature. This suggests that always on servers will have higher disk reliablity than servers that switch the hard drives off or power down completely.   The study also concluded that the make and model was a bigger factor in detemining the reliablity than most environmental factors.  This is why I mentioned make and model of the hard drive.

    Okay, you just shortened your story too much, couldn't guess what the studies said. This fact about hard drives reacting badly to constant spin downs and spin ups has been proven empirically for ages in server rooms. 

     

    Wednesday, March 14, 2007 8:03 AM
  • My main system has been running 24/7 365 for the last 10 years. Only time it is off is for upgrades. No issues here. Mind you it is not in a closet and gets cleaned out regularly. The WHS box is on a hardwood floor next to a desk with plenty of airspace. It is not replacing anything so my power consumption has gone up but what can you do? As nice as it would be to have the WHS box only power on when needed I think in the long run this would be more harmful to the components than running 24/7 and thus offset the extra cost of running 24/7 by having to replace parts. This may end up not even being an issue if places like HP and such build custom WHS boxes for consumers. They can tailor them to low power consumption. The WHS is designed for typical home users who want a low maintenance backup. I doubt they will be buying the parts and building it themselves.
    Wednesday, March 14, 2007 1:00 PM
  • interesting discussion...

    MurMan99, where could we look for an affordable device to "audit" the power consumption of our devices at home ? This is something that everynow and then populates my mind, but I have never done anything to make a diference, either in my wallet nor on the environment. But I do consider myself concious and like to believe I waste little amounts of energy.

    It's just that sometimes thigs get out of perspective, i.e. I have a 50"plasma TV that I watch on average for 2 hours/day. Everytime I walk by it, I can feel its heat. Also I do crank up the sound when listening to music and I can almost feel the receiver sucking on electricity to drive those speakers... So maybe, micro managing a little thing won't necessarilly help that much, although in the end every Watt counts.

    On a separate note, I used to run Six computers back in the old days doing my SETI@Home project... and also did some Folding@home too. Cheers,

    Friday, March 16, 2007 8:57 PM
  •  RodSpina wrote:
    MurMan99, where could we look for an affordable device to "audit" the power consumption of our devices at home ? This is something that everynow and then populates my mind, but I have never done anything to make a diference, either in my wallet nor on the environment. But I do consider myself concious and like to believe I waste little amounts of energy.

    The Kill-a-Watt is what you're probably looking for. It's not super-accurate, but it'll give you a decent idea what your appliances are consuming.
    Friday, March 16, 2007 11:01 PM
    Moderator
  • I had never seen the Kill-a-Watt until Ken suggested it.  It looks pretty cool and the price is definitely right ($23 on Amazon) but I'm not sure that it would work for this kind of measurement.

    Here's my concern about it.  (I promise to keep this as simple as possible, but it does get into electrical theory and I'm a EE.)  Power in watts is the current multiplied by the voltage.  Measuring it in ac circuits is complicated because the current and voltage have waveforms that can be shifted in time or have different waveshapes.  This means that you have to do the multiplication over the entire waveform.  It's relatively easy to deal with time shift (phase shift) but it's not easy to deal with the different waveshapes.  Devices like heaters and motors (the main components of appliances) have current waveforms that are sine waves just like the voltage, so there's only the phase shift to measure.  But many electronic devices (nearly all computers) have switching power supplies that create really ugly current waveforms that are constantly changing.  This requires a wattmeter that is relatively sophisticated and expensive.  The one I used was borrowed and looked expensive.

    I read the Kill-a-Watt manual and datasheet looking for clues about its capabilities.  It talked mostly about appliances and didn't mention electronic devices.  The specs are limited, which to me means you need to assume the worst.   My guess is that it assumes that the current waveforms are sinusoidial and measures the zero-crossings to calculate the phase shift.  This would not work well for electronic devices.  My other concern is that the accuracy of measuring devices depends on the range.  The quoted accuracy of 0.2% (which is unrealistically good for a low-cost measuring device) is probably 0.2% of full-scale.  If full scale is 1500 W, the accuracy could be + or - 3 W, a real problem for low power measurements.

    Maybe Ken or someone else knows something to the contrary about Kill-a-Watt.  Please post it because my opinion is based mostly on the missing specs from the manufacturer. 

    I looked around the net for other low-cost wattmeters and found the "Watts Up Pro" for $129. at the PowerMeterStore.com.  It's specs say that it does high-frequency sampling of voltage and current (a good way to deal with ugly current waveforms) so it should be good for electronic device measurements.  It has 2% accuracy above 10W and 5% accuracy below 10W.  (This seems like a realistic spec. for a device in this price range.)  It also has a USB interface so you can do measurements over time.  I'm thinking about buying one.  But first, I've got to do my taxes this weekend. 

    Saturday, March 17, 2007 1:14 AM
  • Good points, but I never said it was super accurate. Very much the opposite, as a matter of fact.

    I'll observe that +/- 3 watts is 5% at 60 watts, and all most people really care about is an approximate measurement. I don't care if my workstation is consuming 396 watts or 398 watts, I only want to know that it's consuming a bit less than 400 watts, because that number will help me to determine how big a UPS I need for a given runtime. (In)Accurate to within 10% would do well enough for that.
    Saturday, March 17, 2007 1:53 AM
    Moderator
  •  MurMan99 wrote:
    The small form-factor server that I'm testing WHS on idles around 30 watts

    Murray, I'm interested in learning more about your server configuration to get it so low.  I'm running a low power Via processor (~1Ghz), onboard graphics and NIC, 512MB DDR, and a single HDD, and my system pulls about 43W idling as measured by my Kill-A-Watt.  Would you mind providing more details about your system build?  I'm looking to reduce the power draw as much as possible for an always-on box, and I thought that this Via was about as power-sipping as you can get.
    Saturday, March 17, 2007 2:54 PM
  • Shutrbug -

    I'm using a Compaq small-form factor PC, the Evo D510.  Its configuration is similiar to your system, except that it uses a P4 processor.  Based on the processors alone, your machine should draw less power.

    But there are factors other than the CPU.  Our motherboards are very different.  The onboard graphics and chipsets are a factor.  The efficiency of the power supplies could be different.  I replaced the HD on the Evo with a new WD Caviar 320GB.  This drive is very quiet and idles at 8W according to WD specs.   Hard drives are a major consumer of power and could be a major difference.  My DVD and floppy drives, USB keyboard and mouse are disconnected.  The Evo has only one fan (in the power supply.)

    Our measuring devices are very different.  If the Kill-A-Watt works like I speculated above, it's reading will be higher than actual for devices with switching power supplies.

    Considering the differences in devices and measurement, I don't think our readings are that far off. 

    Murray

    Saturday, March 17, 2007 4:45 PM
  • I've tried to get more info about the Kill-a-Watt without success.  The US distributor is P3 and they haven't responded to my email.  It is built in Taiwan by Prodigit Electronics Co. (www.prodigit.com)   There's a two page datasheet, but nothing that describes how it operates.  They rate power readings at 2% accuracy with 0.5% typical.   This is more like it.  0.2% sounded too good to be true.

    I read all the reviews on the web that I could find.  There are none that would qualify as an independent technical review of the product.  Not one review compared its reading with a lab quality wattmeter.  They pretty much all read the same: "There's this cool device called Kill-a-Watt and I got one and I measured lots of things and got lots of readings that were interesting and I don't really understand it, but you should get one too."  Disappointing. 

    I purchased a WattsUp? Pro device today.  (What's with these product names?)  It samples vottage and current at 4KHz and integrates them to get true RMS power readings.  This should be fast enough (barely though) to track the current waveform of a switching power supply.  It also can store up to 1000 readings on a time basis that can be loaded via USB.  (Power usage is seldom constant.)  I found several reviews that were better written than the Kill-a-Watt reviews, but still no lab tests.   I'll use it for a few days and post back what I find out.

    If there is anyone in the San Diego area that owns a Kill-a-Watt, maybe we could get together and compare the readings on indentical loads.  Contact me at murman99@earthlink.net

     

    Tuesday, March 20, 2007 12:34 AM
  • I'm hoping that they come out with low power servers.  You just can't count the amount of energy it uses, but how much heat it puts off.  Here in Las Vegas we have already seen 90 degree days so it's just not the energy the machine uses, but the amount of energy the AC uses to keep the room cool.  I came by this thread because I came home to find out that my WHS failed.  So I ran spinright and it said my hard drive was over heating and now it's trying to fix the bad spot.  I've had this server running for about two months.  A lot of corportate servers are stored in specially AC rooms so they aren't really relying on the systems cooling power.
    Thursday, April 5, 2007 2:50 AM
  • I've written a small Windows Home Server add-in to address this very issue, it's still pretty new so there might be a bug or two that I've missed, but you're welcome to give it a try.

     

    Info and download here:

    http://runtime360.com/2007/12/23/windows-home-server-power-saving/

    Sunday, December 23, 2007 5:57 PM