Re: OT: servers vs. lightning and power outages
On Mon, Mar 14, 2011 at 09:40:34PM -0500, Stan Hoeppner wrote: Jerome BENOIT put forth on 3/14/2011 7:58 PM: Just curious: why the UPS must be at the bottom ? Your attempt at being a smart ass aside, I'll answer your question. The answer, however, isn't the simple one you're expecting. I stated bottom of the server rack explicitly because some SAs install all of their UPS gear in a single dedicated rack, or two, or xx, and feed PDUs in their server racks, often ignorantly over loading their UPS output circuits in the process. And yes, these SAs will load a 42U rack bottom to top with UPS units. I had to do this on one occasion myself due to an ill conceived pre-existing equipment layout. Thankfully the 4 Liebert 3000s I had to mount in the top of the UPS rack only weighed about 80 lbs apiece. The first one was tough, but the last 3 were relatively easy. Ego boost prize goes to the first OP to reply correctly explaining why this was the case. You needed several hands to hold the first Liebert in place as you screwed it into the rack. The next one could rest on top of the first. I note that Liebert does not offer a slide-in rail kit. -dsr- -- http://tao.merseine.nu/~dsr/eula.html is hereby incorporated by reference. You can't defend freedom by getting rid of it. -- To UNSUBSCRIBE, email to debian-user-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/20110318171323.ga24...@tao.merseine.nu
Re: OT: servers vs. lightning and power outages
Dan Ritter put forth on 3/18/2011 12:13 PM: On Mon, Mar 14, 2011 at 09:40:34PM -0500, Stan Hoeppner wrote: Jerome BENOIT put forth on 3/14/2011 7:58 PM: Just curious: why the UPS must be at the bottom ? Your attempt at being a smart ass aside, I'll answer your question. The answer, however, isn't the simple one you're expecting. I stated bottom of the server rack explicitly because some SAs install all of their UPS gear in a single dedicated rack, or two, or xx, and feed PDUs in their server racks, often ignorantly over loading their UPS output circuits in the process. And yes, these SAs will load a 42U rack bottom to top with UPS units. I had to do this on one occasion myself due to an ill conceived pre-existing equipment layout. Thankfully the 4 Liebert 3000s I had to mount in the top of the UPS rack only weighed about 80 lbs apiece. The first one was tough, but the last 3 were relatively easy. Ego boost prize goes to the first OP to reply correctly explaining why this was the case. You needed several hands to hold the first Liebert in place as you screwed it into the rack. I actually only needed two hands (plus a knee and then a shoulder) for the first unit, as I'm a bit larger than the average guy. I do seem to recall a little grunting was involved. ;) Once I had my left shoulder under the front edge and my left hand under the middle, I was able to get the first screw in pretty easily. With half the weight relieved the other screw went in very smoothly. The next one could rest on top of the first. You win the ego boost prize. :) The 2nd one went smoothly, and the 3rd and 4th were slightly tougher due to holding ~80 lbs overhead, however brief that period was. If I weren't 6'4 200 lbs. I'd have never dreamed of making this a one man job. My boss at the time was about 5'4 and 130 lbs. He couldn't even lift one of the Lieberts off the ground. Coincidentally, this is why all the rackable UPS units were on the ground to begin with. The previous and existing staff were unable to lift them, even working as a team. I note that Liebert does not offer a slide-in rail kit. They do offer slide rail kits today http://www.emersonnetworkpower.com/en-US/Products/ACPower/RackmountUPS/Documents/SL-23180.pdf http://www.emersonnetworkpower.com/en-US/Products/ACPower/RackmountUPS/Documents/SL-23185.pdf though I don't know if they did when the 3000s I was installing were originally purchased. Rails aside I thought they were decent UPS units. The 3000s we had were online models, not line interactive, so battery life was lower than the typical line interactive models most shops deploy. -- Stan -- To UNSUBSCRIBE, email to debian-user-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/4d839d31.7050...@hardwarefreak.com
Re: OT: servers vs. lightning and power outages
On 15/03/11 03:40, Stan Hoeppner wrote: Jerome BENOIT put forth on 3/14/2011 7:58 PM: Just curious: why the UPS must be at the bottom ? Your attempt at being a smart ass aside, I do not think so: I was really curious. I plan to play with UPS soon. I'll answer your question. Thanks a lot. The answer, however, isn't the simple one you're expecting. hence my question. I stated bottom of the server rack explicitly because some SAs install all of their UPS gear in a single dedicated rack, or two, or xx, and feed PDUs in their server racks, often ignorantly over loading their UPS output circuits in the process. And yes, these SAs will load a 42U rack bottom to top with UPS units. I had to do this on one occasion myself due to an ill conceived pre-existing equipment layout. Thankfully the 4 Liebert 3000s I had to mount in the top of the UPS rack only weighed about 80 lbs apiece. The first one was tough, but the last 3 were relatively easy. Ego boost prize goes to the first OP to reply correctly explaining why this was the case. Unrelated to that incident, I prefer to install one properly sized UPS (and battery packs if needed) in the bottom of each server rack. There are a few benefits to doing this: 1. Cleaner cabling, less radiated EMI/RFI 2. Shorter cable runs means less resistance, thus lowering the impedance load on the UPS, yielding slightly higher overall efficiency. 3. The immense mass usually lowers the rack center of gravity sufficiently to eliminate the need to bolt the rack to the floor, while still allowing you to slide servers out for maintenance without tipping the rack over on yourself Thanks, Jerome -- To UNSUBSCRIBE, email to debian-user-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/4d7f2b64.8050...@rezozer.net
Re: OT: servers vs. lightning and power outages
Excerpts from Jason Hsu's message of Mon Mar 14 19:31:00 +0100 2011: How do you protect servers from lightning? [...] Others already answered this, but I'd like to stress the fact that you'll need proper protection on _all_ levels. Surge protectors will only do what they're intended to do if coarse protection does its job as well and so on. Match the (input) ratings of each protector to the (output) ratings (response time etc.) of the previous protection level. Make sure _all_ entry points to the house are protected (including telephone line, broadband cable, garage lighting, etc.). A proper ground system is of equal importance. Jim Gettys can tell [1]. HTH. Sascha [1] http://gettys.wordpress.com/2010/08/31/new-ground-system/ -- http://sascha.silbe.org/ http://www.infra-silbe.de/ signature.asc Description: PGP signature
Re: OT: servers vs. lightning and power outages
On Tue, 15 Mar 2011 21:43:20 +0100 Sascha Silbe sascha-ml-reply-to-201...@silbe.org wrote: Excerpts from Jason Hsu's message of Mon Mar 14 19:31:00 +0100 2011: How do you protect servers from lightning? [...] Others already answered this, but I'd like to stress the fact that you'll need proper protection on _all_ levels. Surge protectors will only do what they're intended to do if coarse protection does its job as well and so on. Match the (input) ratings of each protector to the (output) ratings (response time etc.) of the previous protection level. Make sure _all_ entry points to the house are protected (including telephone line, broadband cable, garage lighting, etc.). A proper ground system is of equal importance. Jim Gettys can tell [1]. And you can still only do so much. If you get a close hit, nothing on Earth will protect electrical equipment. Lightning does not obey Ohms' Law, but does exotic things with voltage gradients, and will leave a conductor when it gets bored with following it, through insulation if necessary. The chimney of the house next door to me was hit by a minor strike, only a few bricks loosened. I had a TV aerial in my loft, and some damaged roof felt showed where some of the power took a slight detour. The front element of the aerial was bent back at 45 degrees, but there was no other damage. The body of the coax connector of the TV set was *not* *there*. The shell was there, and the cable ended neatly, but there was no trace of the connector body. The coax plug was completely unmarked. The TV tuner and power supply were damaged. Everything in the house which was turned on was damaged, though apparently it couldn't jump open switches. *Neon* *bulbs* were blown. Unfortunately, I was up a ladder at the time, about two feet from a light bulb which was also on, and which burned very brightly for a few microseconds before expiring loudly. One of the power amps of my tuner-amplifier did not work, and on investigation, I found that a speaker cable crossed the TV aerial cable, and by looking very carefully, I could see a tiny mark on both where they crossed. So place your trust in UPSes if you will, but look carefully at cable crossings, where your expensive lightning arresters can be bypassed. Don't forget the DSL line: don't assume your router will selflessly protect the server by vaporising. Most of my lightning damage (two telephone answering machines, apart from the above incident) has arrived by telephone line. -- Joe -- To UNSUBSCRIBE, email to debian-user-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/20110315212705.37c2a...@jresid.jretrading.com
OT: servers vs. lightning and power outages
How do you protect servers from lightning? What backup power sources do you use to keep them running during power outages? Unlike a home computer for strictly personal use, you can't just shut down and unplug server equipment every time there are thunderstorms in the area, ESPECIALLY in you are located in a lightning-prone area like Florida. Yes, I know there are UPS systems with surge protection (like the APC 350 product I use), but I suspect that some products are much better (and much more expensive) than others. My APC 350 surge protector with a small battery works for me (though I still unplug my equipment from the power line and phone line when I go out of town or expect thunderstorms), but I'm not running anything on my home computer that the world needs to be able to access 24/7/365. -- Jason Hsu jhsu802...@jasonhsu.com -- To UNSUBSCRIBE, email to debian-user-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/20110314133100.c4870624.jhsu802...@jasonhsu.com
Re: OT: servers vs. lightning and power outages
Jason Hsu put forth on 3/14/2011 1:31 PM: How do you protect servers from lightning? During facility construction: www.erico.com/public/library/fep/LT1421.pdf What backup power sources do you use to keep them running during power outages? Generally, one Smart-UPS RT 10KVA RM per rack of 3U/2U/1U servers, UPS at the bottom, servers above. http://www.apc.com/products/resource/include/techspec_index.cfm?base_sku=SURT10KRMXL6U-TF5total_watts=1000 If packing the entire rack with 30 1U servers, I add another transformer to get 36 outlets total, with proper output power distribution from the UPS. This assume single PSU servers. I'm not a big believer in dual PSUs WRT SHV servers. http://www.apc.com/products/resource/include/techspec_index.cfm?base_sku=AP9626 For the 18 SAS drive FC storage arrays, I do one Smart-UPS RT 3000VA per 4 SAN chassis, along with one Smart-UPS RT 192V RM Battery Pack for ~2 hours extended runtime. http://www.apc.com/products/resource/include/techspec_index.cfm?base_sku=SURTA3000RMXL3Utotal_watts=1000 -- Stan -- To UNSUBSCRIBE, email to debian-user-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/4d7eb377.1050...@hardwarefreak.com
Re: OT: servers vs. lightning and power outages
Hi ! On 15/03/11 01:31, Stan Hoeppner wrote: Jason Hsu put forth on 3/14/2011 1:31 PM: How do you protect servers from lightning? During facility construction: www.erico.com/public/library/fep/LT1421.pdf What backup power sources do you use to keep them running during power outages? Generally, one Smart-UPS RT 10KVA RM per rack of 3U/2U/1U servers, UPS at the bottom, servers above. http://www.apc.com/products/resource/include/techspec_index.cfm?base_sku=SURT10KRMXL6U-TF5total_watts=1000 Just curious: why the UPS must be at the bottom ? If packing the entire rack with 30 1U servers, I add another transformer to get 36 outlets total, with proper output power distribution from the UPS. This assume single PSU servers. I'm not a big believer in dual PSUs WRT SHV servers. http://www.apc.com/products/resource/include/techspec_index.cfm?base_sku=AP9626 For the 18 SAS drive FC storage arrays, I do one Smart-UPS RT 3000VA per 4 SAN chassis, along with one Smart-UPS RT 192V RM Battery Pack for ~2 hours extended runtime. http://www.apc.com/products/resource/include/techspec_index.cfm?base_sku=SURTA3000RMXL3Utotal_watts=1000 Jerome -- To UNSUBSCRIBE, email to debian-user-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/4d7eb998.5000...@rezozer.net
Re: OT: servers vs. lightning and power outages
On Tue, Mar 15, 2011 at 01:58:00AM +0100, Jerome BENOIT wrote: Hi ! On 15/03/11 01:31, Stan Hoeppner wrote: Jason Hsu put forth on 3/14/2011 1:31 PM: How do you protect servers from lightning? During facility construction: www.erico.com/public/library/fep/LT1421.pdf What backup power sources do you use to keep them running during power outages? Generally, one Smart-UPS RT 10KVA RM per rack of 3U/2U/1U servers, UPS at the bottom, servers above. http://www.apc.com/products/resource/include/techspec_index.cfm?base_sku=SURT10KRMXL6U-TF5total_watts=1000 Just curious: why the UPS must be at the bottom ? Probably because they're really heavy. I had a server rack on wheels, and a UPS which had a shipping weight of 172 lbs. No way was I putting that anywhere but the bottom of the rack, for stability reasons. -Rob -- To UNSUBSCRIBE, email to debian-user-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/20110315011018.gb22...@aurora.owens.net
Re: OT: servers vs. lightning and power outages
Jerome BENOIT put forth on 3/14/2011 7:58 PM: Just curious: why the UPS must be at the bottom ? Your attempt at being a smart ass aside, I'll answer your question. The answer, however, isn't the simple one you're expecting. I stated bottom of the server rack explicitly because some SAs install all of their UPS gear in a single dedicated rack, or two, or xx, and feed PDUs in their server racks, often ignorantly over loading their UPS output circuits in the process. And yes, these SAs will load a 42U rack bottom to top with UPS units. I had to do this on one occasion myself due to an ill conceived pre-existing equipment layout. Thankfully the 4 Liebert 3000s I had to mount in the top of the UPS rack only weighed about 80 lbs apiece. The first one was tough, but the last 3 were relatively easy. Ego boost prize goes to the first OP to reply correctly explaining why this was the case. Unrelated to that incident, I prefer to install one properly sized UPS (and battery packs if needed) in the bottom of each server rack. There are a few benefits to doing this: 1. Cleaner cabling, less radiated EMI/RFI 2. Shorter cable runs means less resistance, thus lowering the impedance load on the UPS, yielding slightly higher overall efficiency. 3. The immense mass usually lowers the rack center of gravity sufficiently to eliminate the need to bolt the rack to the floor, while still allowing you to slide servers out for maintenance without tipping the rack over on yourself -- Stan -- To UNSUBSCRIBE, email to debian-user-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/4d7ed1a2.3020...@hardwarefreak.com