- Test - lists.nanog.org

Re: MAE-East still no generator
by Sean Donelan 08 Sep '96

08 Sep '96

>Just out of curiosity, does anyone have any personal experience with >losing utility-provided power more than, say, three times in a day? Plenty. When power goes out around here, its going to go up and down several more times. DRA is located between several organizations concerned with real money, Bridge Information handles real-time trading feeds, MasterCard International does credit cards and Western Union/AT&T does money orders. The standard operating procedure around here is three backup generators. Nothing can ruin your trading day faster than a power failure just at the moment you have your backup generator torn apart for maintenance. With 24 hour trading becoming the norm, all day is trading day. Two generators usually start immediately, with the third one held as a spare. DRA won't make the news. But if you hear MasterCard International is down, DRA will likely have suffered the same fate. P.S. a generator hint, don't forget to keep your fuel tanks warm. As several sites in the northeast found out a couple of winters ago, diesel fuel gels at very cold tempatures. -- Sean Donelan, Data Research Associates, Inc, St. Louis, MO Affiliation given for identification not representation

1 0

Re: MAE-East still no generator
by matthew＠scruz.net 08 Sep '96

08 Sep '96

Original message <199609081405.KAA04199(a)netaxs.com> From: Avi Freedman <freedman(a)netaxs.com> Date: Sep 8, 10:05 Subject: Re: MAE-East still no generator > > > Our real problem is getting them to fit properly in 23" racks :) > > > > ===================================== > > George Hall 916-565-4500 > > While this may not apply to 7513s, we've found that super-sturdy $25 > 23" or 19" rack shelves (and Greybar sells the Chatsworth sturdy > shelves for $75 or so) take a 7507 or 7000 (or 7505/7010) *very* well, > and there's not need to bother with the rack-harnesses. > > Avi We have facilities on both sides of the San Andreas (our main offices are in one of the downtown Santa Cruz buildings that *didn't* collapse in 1989) and don't share your view of shelves :) -matthew kaufman matthew(a)scruz.net

1 0

Re: MAE-East still no generator
by Rob Liebschutz 08 Sep '96

08 Sep '96

In the Alternative energy arena, Trace Engineering makes a 4000 and 5500 watt 48 VDC sine wave inverter that supports stacking of two inverters to double the capacity/reliability. It's probably not designed to be rackmountable though. For smaller applications, Exeltech makes physically compact solid state sine wave inverters ranging from 250 - 6000 watts. If your running AC equipment with non-switching type power supplies, like CSU/DSU's, monitors or MUXes watch out for any inverter that does not generate a true sine wave. Square waves and modified sine waves can fry such equipment. And Yes, you do need hefty wiring on DC equipment due to the voltage drop created when transmitting high current at low voltage. I have a 400 amp hour battery bank in my van with a 1200 watt inverter mounted about 10 feet away and I used 2/0 (that's pronounced 2 ought - heaver than 0 gauge) wire for a proper connection. There are formulas for calculation of the correct wire gauge. Heavy wire that is flexible enough to be routed around corners and through racks is made of braided conductors and is commonly available from welding suppliers. Trace Engineering can be reached at 206-435-8826. Exeltech can be reached at 817-595-4969. I could provide lists of vendors of such equipment if that is appropiate. Rob

2 1

Re: MAE-East still no generator
by Vadim Antonov 08 Sep '96

08 Sep '96

Robert E. Seastrom <rs(a)bifrost.seastrom.com> wrote: > From: Andrew Partan <asp(a)partan.com> > > So what are you using? AC or DC? If you are not on DC, I would > switch. Maybe you would have better luck. >You took the words right out of my mouth. -48vDC is *so* much cleaner >and less troublesome than mains AC (or even upsed AC) that I urge all >my clients to use it for *everything* whenever it's available. Typically telco setups have about 30min-1hr of AC UPS capacity and 4-6 hrs of 48V battery capacity. Last time i was at MAE-East it didn't have any backup AC power and a nice DC setup. BTW, the cheap solution for DC hookup would be using a power inverter. --vadim

4 3

Inverters
by mo＠uu.net 08 Sep '96

08 Sep '96

Gee, you can either do an Alta Vista search, or even ask MFS what they use to power AC things from CO-DC supplies. the one that seems most interesting is Exceltech out of Dallas. they have a modular inverter system which makes its ratings at 100% duty cycle (ie, all the time). they can rack and stack up to 20KVA per assembly, if my feeble memory serves me right. last time I looked, MFS used the smaller non-modular Exceltech units to power AC things they supplied on MAE-EAST (ie, at their end). Don't succumb to the stupid idea of buying a UPS without batteries and then connecting to the DC supply. most ups units smaller than a refrigerator have to be significantly derated to be used in "pure interver" service. the manufacturers scream when you ask them about this. of course, 28.8 modems, 10-base-T hubs, and everything else you need IS available DC powered. you just have to be diligent. (grin) but there are times when using an AC thing *is* more expedient. Just plug the cord into a real inverter running off house DC. Good luck. -Mike O'Dell UUNET Technologies

1 0

Re: MAE-East still no generator
by Peter Lothberg 08 Sep '96

08 Sep '96

> > So, what's the fuzz here? > > > > (Don't you all use -48Vdc in telco pops?) > > > > --Peter > > The fuzz is that most people use AC supplies in their CSUs and routers. > My best guess would be that it's so that they could move those boxes > out of the telco environement (if needed) and run them on AC. You can swap the PS in any cisco box, on some models you don't even need a screw driver. As someone already pointed out, it takes half the components to do UPS -48V than UPS 110V/230V. Where I live the comercial power is more reliable than the small UPS boxes, so the lesson is, UPS'es has to come in pairs.. --Peter

1 0

Re: MAE-East still no generator
by Peter Lothberg 08 Sep '96

08 Sep '96

> > I realize that the phone world that MFS is part of operates > > differently, but this is networking equipment we are talking about, > > not phone switches, and UPSes are very old tech at this point. > > the bottom layers of the network (L2 and below) are mostly telco > stuff and mostly (other than the gigaswitch) prefer DC power. DC Our Gigaswitches run on DC..... --Peter

1 0

Re: MAE-East still no generator
by Peter Lothberg 08 Sep '96

08 Sep '96

> >|} Check out the nice graph of MAE-East for that last 5 days. When are > >|} they going to get a generator installed? Why can't you pay for a > >|} generator when you are charging +30 people $5,700 a month? > > > >At least part of the facility (1919 Gallows, P1) is on generator, or > >hasn't had a power outage recently. > > > >mae-east uptime is 3 weeks, 3 days, 5 hours, 46 minutes > > All of our equipment is on battery-backed -48 volt DC power. Well, for > full disclosure, almost all of it, there is a dialup modem plugged into > a wall wart; but I can't find an rackmount DC-powered NEBS compliant > modem in quantities of one. > > But, there seems to be some piece of equipment somewhere that's missing > power. Even though the FDDI interface stays operational, we seem to loose > almost all the peering sessions. Even other folks who also are on DC power. > AS286 sticks around for some reason I can't explain. I guess its time to > physically walk the circuit path looking for equipment plugged into a > wall outlet. > > BTW, are other folks getting advanced notice from MFS about this? Our > sales person claims we're on the notification list, but we still don't get > advanced warning about work at mae-east. > -- > Sean Donelan, Data Research Associates, Inc, St. Louis, MO > Affiliation given for identification not representation > ICM-MAE-E uptime is 2 weeks, 4 days, 14 hours, 12 minutes System restarted by reload at 14:40:57 EDT Tue Aug 20 1996 Running buggy software Neighbor V AS MsgRcvd MsgSent TblVer InQ OutQ Up/Down State 192.41.177.1 4 1 92312 159041 6773476 0 0 2d17h 192.41.177.115 4 2548 31408 457020 6773476 0 0 2d17h 192.41.177.140 4 690 116540 212363 6773476 0 0 1d10h 192.41.177.141 4 1673 99241 158848 6773501 0 0 2d17h 192.41.177.170 4 3830 103181 211298 6773501 0 0 2d07h 192.41.177.180 4 3561 221435 210611 6773501 0 0 2d17h 192.41.177.181 4 3561 26754 210272 6773501 0 0 2d17h 198.67.131.89 4 1800 763230 569598 6773631 0 0 2d11h This is funny, I did loose my IBGP peers, maybe the MFS provided IDSU is on AC power... -:) --Peter

1 0

Re: MAE-East still no generator
by Vadim Antonov 08 Sep '96

08 Sep '96

Paul A Vixie <vixie(a)wisdom.home.vix.com> wrote: >DC >power is what the box wants internally, and there's a lot less heat >generated by making 5V and 12V out of 48VDC instead of 110VAC or 220VAC. >if you visit a telco switching center and surround yourself with 10 >Northern Telecom OC48 racks you should notice how quiet it seems. Well, it is not exactly true. The way modern power supplies work is to rectify AC to DC, modulate it with high frequency (1KHz or so) with some transistors to feed a small transformer (the higher frequency, the smaller core you need), and then rectify it again. The power loss on transistors working in saturated modes is mostly due the voltage loss (which is constant determined by physics of the PN-junction or field-effect) multiplied by the current. With lower voltage you need higher current to get the same power, so the power dissipation on switching transistors is higher. Therefore, high-voltage power supplies are more efficient than low-voltage. The reason why sync mux racks are quiet is because they don't do any fancy high-speed stuff (after all, those muxes are nothing more than glorified shift registers) and so they don't need any power-hungry circuitry. --vadim /

1 0

Re: MAE-East still no generator
by Peter Lothberg 08 Sep '96

08 Sep '96

So, can someone explain what the problem is? MAE-East is a telco facility (well...) but if I look at one of our routers; ICM-MAE-E uptime is 2 weeks, 4 days, 14 hours, 12 minutes System restarted by reload at 14:40:57 EDT Tue Aug 20 1996 So, what's the fuzz here? (Don't you all use -48Vdc in telco pops?) --Peter

2 1