January 2001 - NANOG - lists.nanog.org

RE: Forcasts, why won't anyone believe them?
by dlr＠bungi.com 17 Jan '01

17 Jan '01

[In the message entitled "RE: Forcasts, why won't anyone believe them?" on Jan 16, 19:49, Roeland Meyer writes:] > > The assumption is wrong. A server motherboard and CPU draws the same power > regardless of what it's doing. Lap-tops and the like are different and you > actually pay extra for that design. Most circuts these days are nmos > technology. Only in cmos does the power go up with the frequency. Peripheral > usage, like disk drives, are also constant since the largest power draw goes > to keeping them spinning. The seek mechnics are trivial. Floppy drives and > cd-rom drives are different. But, most servers do not keep those spinning > constantly. Ergo, for all intents and purposes, servers are a constant power > draw. They can be rated. Uh - I think you will find that disk drive actually have between two and "many" modes in how they consume power. The largest amount of power is typically drawn on startup (but this depends on the drive, of course). Power also goes up radically on seeks, because they are not trivial operations on fast-seek drives. Other modes include write-behind, and cache prefill, which may draw additional power during activation. Some drives also require more power during certain rare operations (like initial servo write). Someone else has already pointed out that all modern equipment uses CMOS, and the externaly clocked components have several modes (during refresh, for example, the DRAM draw *much* more power than when in idle). So, server power most definately is a combination of what they are doing, and how much they are doing it. It's almost never constant (except when a UNIX server is in the idle loop). What's more amusing, however, is trying to measure it on the AC side. Since most power supplies are switch-mode, with poor power factors, standard off-the-shelf techology will be anywhere from wildly wrong to totally inaccurate. --

2 2

IPSectarianism
by Dave Wardle, Critical Networks, Inc. 16 Jan '01

16 Jan '01

Back in November 2000 I read with interest a thread discussing the implications of Service Providers blocking particular protocols (I believe it was "Operational impact of blocking SMB/Netbios"). I recall one response which explicitly stated that IPsec was not blocked. Is anyone on the list aware of Service Providers (ISP/NSP...) who DO block IPsec traffic, with or without informing their customers or peers? I'm trying to assess the pros and cons of major Enterprise Customers basing their entire remote office/small office/mobile network access strategy on some type of IPsec based VPN solution. Any thoughts? Cheers Dave ------- Dave Wardle, Principal Consultant Critical Networks, Inc. ------- Email: dave(a)criticalnets.com Homepage: www.criticalnets.com ------- Cell: 831 332 1021 Tel: 831 662 1710 Fax: 831 662 1710 -------

2 1

Re: Forcasts, why won't anyone believe them?
by Sean Donelan 16 Jan '01

16 Jan '01

On Tue, 16 January 2001, Jeff Cours wrote: > It might be possible to come up with some sort of average power draw, > but Electrical Engineers really hate to give out numbers like that > because people base their designs on them instead of on the worst case > power draw, and then when something fries the EE winds up getting the > blame. That's why most engineering disciplines derate components and > allow a safety margin, which I suspect is where the fuse rating comes > from. They may hate to give out this information, but any piece of equipment with a NEBS level 3 rating already has this information. I'm not asking for vendors to do the full, burn the box, NEBS testing. But having the power drain data sheet would be very useful. This is really a customer demand issue. If customers started demanding vendors supply a full NEBS-like space and power planning data sheet, we would get the information. But customers don't, so we don't. While this may be a deliberate decision by customers, who would rather save the money on equipment and spend a bit more on "extra" electrical infrastructure, I don't think so. I suspect its because customers have never known they could ask for such a thing.

1 0

Re: Forcasts, why won't anyone believe them?
by Richard A. Steenbergen 16 Jan '01

16 Jan '01

On Tue, Jan 16, 2001 at 03:43:31PM -0800, Sean Donelan wrote: > > 10 years ago, how many folks remember going to the phone company and > telling them "I need 600 phone lines in my basement." And the phone > company replying, "No one needs 600 phone lines, here is 6." "No > really, I need 600 lines in the next 12 months." The phone company > replies, "When you need them, then we'll install them." "How long will > it take." The phone company replies, "Don't worry, we know what we're > doing, that's none of your concern." To be fair, if all the customers in a colo used the amount of bandwidth they predicted they would at the rate they would, current facilities would be out of capacity after a few dozen racks. Most customers can't predict for squat, so when a real figure comes along it takes more work to make sure its believable. -- Richard A Steenbergen <ras(a)e-gerbil.net> http://www.e-gerbil.net/humble PGP Key ID: 0x138EA177 (67 29 D7 BC E8 18 3E DA B2 46 B3 D8 14 36 FE B6)

2 1

RE: IPSectarianism
by Steve Riley (MCS) 16 Jan '01

16 Jan '01

A while back, Cox @Home made it very clear that the use of VPN protocols is prohibited. AT&T @Home (my provider) hasn't made any similar statements, so I think it's more of a policy issue for the various cable companies than it is for @Home themselves. Of course, there's always httptunnel. Stick your IPSec/PPTP/L2TP traffic inside good old http and you'll get anywhere you need to go... _______________________________________________________ Steve Riley Microsoft Communications Consulting in Denver, Colorado steriley(a)microsoft.com +1 303 521-4129 (OLD mobile) www.microsoft.com/isn/ Applying computer technology is simply finding the right wrench to pound in the correct screw. -----Original Message----- From: Dave Wardle, Critical Networks, Inc. [mailto:dave@criticalnets.com] Sent: Tuesday, January 16, 2001 7:49 PM To: nanog(a)merit.edu Subject: IPSectarianism Back in November 2000 I read with interest a thread discussing the implications of Service Providers blocking particular protocols (I believe it was "Operational impact of blocking SMB/Netbios"). I recall one response which explicitly stated that IPsec was not blocked. Is anyone on the list aware of Service Providers (ISP/NSP...) who DO block IPsec traffic, with or without informing their customers or peers? I'm trying to assess the pros and cons of major Enterprise Customers basing their entire remote office/small office/mobile network access strategy on some type of IPsec based VPN solution. Any thoughts? Cheers Dave ------- Dave Wardle, Principal Consultant Critical Networks, Inc. ------- Email: dave(a)criticalnets.com Homepage: www.criticalnets.com ------- Cell: 831 332 1021 Tel: 831 662 1710 Fax: 831 662 1710 -------

1 0

Forcasts, why won't anyone believe them?
by Sean Donelan 16 Jan '01

16 Jan '01

10 years ago, how many folks remember going to the phone company and telling them "I need 600 phone lines in my basement." And the phone company replying, "No one needs 600 phone lines, here is 6." "No really, I need 600 lines in the next 12 months." The phone company replies, "When you need them, then we'll install them." "How long will it take." The phone company replies, "Don't worry, we know what we're doing, that's none of your concern." This was followed by the phone companies complaints that the Internet wave hit them "without notice." Even though ISPs had been giving them forcasts about the demand, the phone companies didn't want to make the investment because those Internet companies might disappear in 3 months and leave a bunch of unused capacity. Internet companies and web hosters have been giving power companies forcasts for several years about their electrical demands. And once again, the reply usually was, "No one needs 10 Megawatts for an office building, here is 10 Kw." "No really, we need 10 Megawatts." The power company replies "Don't worry, we know what we're doing." If you look at the last three stage 3 alerts, there has always been extra power available in the Western electrical grid. The problem has been one of credit, not power. One thing that would help. Sun, Dell, Cisco, Compaq, Juniper, etc. Can you please start listing the true power draw of your equipment, not just the fuse rating. It would make forcasting a lot easier, if we knew ahead of time how much the equipment will really draw.

4 3

Re: [Possible OT] California, and running off of generators for extended periods
by Sean Donelan 16 Jan '01

16 Jan '01

On Tue, 16 January 2001, Jerry Scharf wrote: > Unless you have a group that is really really set up to run on local power as > your main source, it's best to use it as the backup it was designed to be. > Unless the facility is ready to make a major commitment to local power > generation, the risks outweigh the benefits, IMO. I'd agree with this statement. Although some politicians have made a deal about the "Internet" economies effects on the power situation in California, the major colo and hosting companies don't really have much effect on the "peak" load. They already have either interruptible contracts, or go on generator when the voltage starts fluctuating. When the grid is down to the last 0.5%, a lot of data center load is already shed off the grid. Which really shows you how bad things have gotten. The problems with doing this on a regular basis. "STANDBY" power systems are standby systems. It seems obvious, but folks don't always remember it. Standby plants are not designed, built or maintained for continious operation. When you order a generator from Catapiller, you have three options: Primary, Continious and Standby. The different types of generators have different costs, and standard features. Maintaining and operating a continious power plant is very different from a standby power plant. Its effects everything from OSHA regulations to the type of employees you need on staff. The Environmental Protection Agency assumes standby generators operate when the grid is off-line, so the net pollution is about zero. If you operate while the grid is on-line, the EPA will start asking for impact statements. For local pollution controls, the assumption is standby generators operate a maximum of 250 hours per year (including testing, standby duty, etc). Normally not a problem, except maybe this year. Once again, more than 250 hours of operation, and someone will want an impact statement. Some local air quality control boards may have more stringiant controls. And finally, the most important factor: cost. Power from the grid is the average cost of power including very cheap hydro and nuclear to more expensive coal and gas fueled plants. While the peak cost of operating your own diesel may be less, the average cost is usually more. If you've been reading the newspapers, you will have seen a number of people making noise about co-generation plants. Many large universities already have on-campus co-gen facilities. The problem is in the past co-gen only made sense when you had a use for the "excess" heat, such as heating the college campus. The one thing colo/hosting facilities don't need is excess heat. But, depending on how the market shapes up, it may make financial sense to operate your own power facility. If you can partner with a nearby institution which needs the heat, and can share the reliable power (e.g. a hospital, hotel, large office complex) you might make it work.

1 0

[Possible OT] California, and running off of generators for extended periods
by Matt Clauson 16 Jan '01

16 Jan '01

Allo, all... Watching what's going on in California, it occured to me, and some of my colleagues, that the major colo and data centers (for example, but not focusing on, Exodus, Colo.com, etc.) could assist in the power crunch that California is having by shifting part, or all, of their loads over to generator power -- one would think that a) diesel fuel is still cheaper than electricity in California, price hikes lately notwithstanding, and b) the goodwill generated by such an endeavor would be good for the company. Admittedly, some sites with only limited battery UPS power cannot do this, at least not so early in the day. But it seems to me that at least a few datacenters and major facilities could switch over to independent power and just keep fueling the generators. Question: Why would (or wouldn't) your company switch your datacenter loads to generator (or other off-mains) power sources, if you had the capability and/or capacity? Comments invited, especially from companies in CA, or with offices/datacenters in CA. Comments will be aggregated and (re)posted at a later date. (note: this is posted from my personal account. This inquiry is NOT related to my employer or employment, but is posed for information, speculation, and discussion among the NANOG and Internet communities.) --mec

4 3

Re: [Possible OT] California, and running off of generators for extended periods
by Richard A. Steenbergen 16 Jan '01

16 Jan '01

On Tue, Jan 16, 2001 at 10:03:25AM -0800, Sean Morrison wrote: > > Anyone using or considering using a gas turbine instead of diesel to > run their generators? Why not just go straight to nuclear powered colo? :P http://www.uic.com.au/nip60.htm -- Richard A Steenbergen <ras(a)e-gerbil.net> http://www.e-gerbil.net/humble PGP Key ID: 0x138EA177 (67 29 D7 BC E8 18 3E DA B2 46 B3 D8 14 36 FE B6)

3 2

Re: [Possible OT] California, and running off of generators for extended periods
by matthew＠tycho.net 16 Jan '01

16 Jan '01

> Question: Why would (or wouldn't) your company switch your datacenter > loads to generator (or other off-mains) power sources, if you had the > capability and/or capacity? 1. Due to price caps, it costs me less to take electricity from PG&E (which, because I run a datacenter 24x7, I don't take with time-of-use metering) than it does to run my generator on diesel. Especially so if I factor in the non-fuel costs of operation. 2. My permit from the local air quality management district restricts me to operation when the utility is offline plus a total of 60 hours per year for all testing. I would need to apply for a permit for a different type of operation (voluntary load shedding, peak shaving, or prime generation) and for that, a full environmental impact report is required. Given #1, it isn't worth my trouble. (And my AQMD is the Monterey Bay AQMD... I can't even imagine the pain of applying in the Los Angeles area) 3. Other than as a PR effort, this doesn't make as much sense as having some industry which uses much more power than the Internet data center industry do this. Perhaps an industry which can easily move their peak load to a different hour of the day. Of course, if we had *real* deregulation, my price for power as a commercial user would closely reflect the utility's cost for power, and they might even charge me differently by time of use whether or not I wanted that... then the cost equation changes for item 1, and I'd have an incentive to go look at the true costs of item 2. -matthew kaufman Tycho Networks/DSL.net matthew(a)tycho.net

4 4