Re: rack power question
A close second might be liquid cooled air tight cabinets with the air/water heat exchangers (redundant pair) at the bottom where leaks are less of an issue (drip tray, anyone? :) )... Something like what you suggest has been around for a year or two now, though using liquid CO2 as the coolant. It doesn't require particularly tight cabs. http://www.troxaitcs.co.uk/aitcs/products/ Is anyone using these over here? This is a far more significant strategy that simply using an alternative to water to carry the heat from the cabinets. The game is PHASE CHANGE, but unlike our traditional fairly complicated refrigeration system systems with oil return issues and artificiaally high head pressures simply to have a 100PSI MOPD to keep full flow through the TXV (even with low ambients outside) this is in its simplest form viewed as a PUMPED LIQUID heat pipe system, where there is no need for large pressure drops as the fluid goes around the loop. Your pump only has to cover piping loses and any elevation differences between the COLO space and the central machinery. There is NO insulation at all. The liquid being pumped out to finned coils on the back of each cabinet is at room temperature and as it grabs heat from the cabinet exhaust air (which is very efficient because you have it HOT and not needlessly undiluted with other room air) some of the liquid flashes to gas and you have a slurry that can easily be engineered to handle any size load you care to put in the rack. The more heat you add, the more gas and the less liquid you get back, but as long as there is still some liquid, the fluid stream is still at the room temperature it was at before entering the coil. It is perfectly happy trying to cool an empty cabinet and does not over cool that area, and can carry as much overload as you are prepaired to pay to have built in. At the central equipment, the liquid goes to the bottom of the receiver ready for immediate pumping again, and the gas is condensed back to liquid on cold coils in this receiver (think of a large traditional shell and tube heat exchanger that also acts as a receiver and also a slight subcooler for the liquid). The coils can be DX fed with any conventional refrigerant, or could be tied to the building's chilled water supply. Parallel tube bundles can provide redundant and isolated systems, and duplicating this whole system with alternate rows or even alternate cabinets fed from different systems lets you function even with a major failure. Read about their scenarios when a cooling door is open or even removed. The adjacent cabinets just get warmer entering air and can easily carry the load. Enough 55 degree ground water in some places might even let you work with a very big shell and tube condenser and NO conventional refrigeration system at all. If you have every single cabinet packed full, having just two systems each needing full double+ capacity would not be as good as having 3 or 4 interleaved systems, but that is simply a design decision, but one that can be partially deferred. Pipe for 4 interleaved isolated systems, and then run the ODD ones into one condensing/pumping system, and the EVEN ones into another. As cabinets fill, and as dollars become available for paranoia, add the other central units and flick a few normally padlocked preprovisioned valves and your are done. The valves stay for various backup strategies. You can accidentally leak some CO2 from one system to another and then sneak it back. There are NO parallel compressor oil return issues, just a large range between min and max acceptible charges of CO2. The big problem is that CO2 at room temperature is about 1000 PSI, so all this is welded stainless steel and flexible metal hoses. There need not be enough CO2 in any one system to present any suffocation hazard, but you DO want to be totally aware of that in the design. Unlike regular refrigerants, liguid CO2 is just dirt cheap, and you just vent it when changing a finned rear door - each has its own valves at the cabinet top main pipes.. You just go slowly so you don't cover everything or anyone with dry ice chips. Here is another site hawking those same Trox systems: http://www.modbs.co.uk/news/fullstory.php/aid/1735/The_next_generation_of_cooling__for_computer_rooms.html Over in Europe they are talking of a demo being easliy done if you already have chilled water the demo could use. A recent trade mag had a small pumped heat pipe like R134a system for INSIDE electronic systems - a miniature version of these big CO2 systems. Heat producing devices could be directly mounted to the evaporator rather than use air cooling fins or a water based system, and the condenser could be above. or below or wherever you need to put it and could function in arbitrary positions in the field. And no heat pipe wicks needed. The fully hermetic pump has a 50K hour MTBF and is in this case pumping R134a.
Re: History of the EPO (Emergency Power Off)
Many years ago when we were much, much smaller, the EPO was wired to a special EPO circuit breaker on the main panel which fed the subpanel for the datacenter room. A short on that breaker was like pressing the test switch on a GFCI breaker. Do most people who do have functional (as opposed to decorative) EPO buttons have them connected to the building/suite mains disconnect? or to the output of your UPS units? to a special EPO panel which trips the EPO cutoffs on other units? I'd guess what you are describing is what is known as a SHUNT TRIP coil in the large breaker you need to trip. This is a readily available option even on relatively small breakers - just feed it power and it trips the breaker. However it does need seperate power run through the EPO button and fed from a small dedicated 15 or 20AMP normal branch circuit breaker. Once the inspector has permanently departed, that little breaker can be accidentally left tripped and then the EPO function does not work - no wiring/unwiring skills needed. Ususal issues of liability, so decide if/how to inform other staff.
Re: Telecom Convertors for DS1/DS3 signals
I am looking for recommendations for devices that can take a DS3 signal from a carrier via coax, convert it to optical so I can use fiber across buildings to transport the signal, www.mrv.com/dl.php?prod=FD type=A4PDF72file=MRV-FD-SF will do a T3 on a single fiber, but many similar products probably for less $ will do it on 2 fibers. and then on the other side, demux the DS3 signal to individual DS1 channels and terminate them on a RJ-48 port. Any suggestions? I am only looking at one DS3's worth of data to transport across buildings. Distance is around 1000 feet. Any 1U M13 mux should be fine. Older several U high clunkers should be avoided unless you already know and love some model. The 3 obvious brands are Telco Systems, CAC, and Adtran. Each has it features and problems, and vocal fans. I'll stay mostly neutral in public but love/hate all 3 for different reasons. Ebay $900 +/- $250 should get you one, but be sure you get TWO FULL main cards on the Telco Systems or Adtran, or the 1:N spares in the CAC. You definitely want as full a box as possible with all the built in redundancy that brings. You also want to inquire about software upgrades. They range from free forever to OUCH each time depending on brand. But normally one DS3's worth of data is NOT delivered on a channelized T3, broken down to 28 T1s each of which could even be clocked at a slightly different rate. ATM IMA? or MPPP? either ok with up to maybe 8 T1s? Or what are you doing? If it is true fractional DS3, the question would be what brand CSU/DSU was or was being emulated at the far end - there is NO one standard, and an M13 MUX is totally WRONG. A cisco PA-2T3 used for $300 handles 2 T3s and with about 3 different brands being emulated (ADC, Larscomm, DigitalLink), the PA-2T3+ does more and costs more. The single port versions are often not really much less used. A PA-MC-2T3+ (or the PA-MC-T3) can take the channelised T3 and view it as 28 T1s or weird FT1 slicings, and works well for MPPP up to maybe 8 T1s worth. But NOT ATM IMA. The PA-MC-2T3+ (but not the single MC T3 one) also does full/fractional T3s the same way the PA-2T3+ does. PA-MC-T3 is channelised ONLY. Using the PA-MC... card as a cisco router port card obviates the M13 in the channelised to T1/FT1 situation unless you need an M13 to get T1s to feed the 8 port T1 IMA card. Gruber (www.gruber.com) makes a solid and cost effective 32 pair relatively new (maybe 2-3 years) telco connector to RJ-48 panel, and can make the 32 pair cables (28 is also ok). This is NOT the very fancy and extremely expensive one they once made for someone else that then sold them for some insane price years ago. Normal Male to Male cables will work with any brand, but only if using the Telco Systems box, a reverse entry (pinned normally electrically but 90degree cable exit is on other side) is modestly better on the one cable end for T1 OUT on the right side at the rear of the M13 (OUT == RECEIVE DATA, fwiw). An alternate strategy is to order all 180 degree connector hoods and then there are no cable exit direction issues but in all cases be prepared to swap screwed on ty-wrapable clips for jack posts or longsize (for cable connector to cable connector matings) self retaining 4-40 screws for their shorter cousins needed for cable to bulkhead with jackpost applications (see AMP's catalog). Peel and stick ty-down points are also your friends as is waxed lacing twine. 180 degree hoods get ugly with 22ga ABAM. I assume your total T1 loop length is quite short and the IN/OUT (Tx/Rx) signals will be similar levels. So order 26GA non shielded cat-3 (or 24GA, but NOT 22GA). That is tiny, flexible, and all you need. AMP's 24GA connectors are also used for 26GA - not sure about the other brands, but whoever makes your cables should know. You do have -48VDC handy, of course.
Re: Routers for CO OOB management network
I need to select a router to install in each of our CO's to bring together a network of T-1 between our colos. Look to Ebay or similar for 2500's with dual V.35. Either find them with DC supplies or buy DC supplies as spares. The DC version has only one inlet, but it is trivial to bridge an A/B supply with a diode bridge. We do so with our boxes. Depending on availability of DC supplies this could be very cost-effective. The 2600's DC supplies are an exact mechanical fit and have a slightly heftier power rating and may be easier to find. They sometimes appear on EBAY as either 2500 or as 2600 supplies at $20 or less each. If you get stuck, the supply manufacturer sold us 20 of them at $65 each a few years ago. You may have to be persistant, and promise to never tease ci$co about it. Remember to load all 2500s with 16 meg DRAM and 8 meg of flash (or more). If you have a lot of 4 meg flash sticks, use 2 in 1/2 your routers (and say: PARTITION FLASH 1 8 - rather than 2 4 4), and new 8 megs in the rest.The DRAM is about $10 on ebay, and the FLASH is maybe $20. Try hard for Intel flash chips (or Sharp clones that ID themselves as Intel) so you won't need to chase newer boot PLCCs if you get routers with old boot code than can't do AMD flash. Load them with unix compressed .Z images or see if you can find one of the kits that used to be around on the net to convert cisco run from FLASH images into more normal for cisco -mz gzipped types buried in a self extracting wrapper so your 25xx routers will behave like most other cisco routers. Find details in old C.D.S.C news archives. You can then load newer images onto a RUNNING router without the 2500 class B/S involving Flash-Load-Helper and remote disasters when flash is erased and tftp fails.
Re: Black box that allows just an A or B DC feed
You just need two suitably large diodes. Big ones need heatsinks. You will find some of the rack-top fuse panel makers that include diode OR-ing their main A and B feeds. No switchover time at all. The power is just there. Pretty sure Telect has some. Beware of Hendry. Good stuff, but savagely overpriced unless you are buying Bell class quantities and a huge percent of neat options in the catalog seem to never have been built! They offer every weird combo anyone might want, and then build when they see the order. You WILL wait if it isn't a popular item on some distributor's shelf. The Canadian NoranTel folks are VERY responsive and make a nice 1U 200 AMPS per A/B feed panel that has up to 50 amps for each of 4 A/B outputs as well as a pile of grasshopper outputs for smaller up to 15 AMP loads. The unit we get does not have diodes, but if they don't have something suitable, these folks are very responsive to new products fast. There may even be some fuse panel vendor with those diodes on the individual outputs., but I don't remember any. Even worse than that single input device is the sleazy device that LOOKS as though it has dual inputs but is actually single input with the 2 connections bussed together with no blocking diodes (and no woarnin to customers...) and when wired as 2 seperate inputs will gleefully backfeed A to B or B to A depending on what died. This is no problem if a fuse/breaker to just this device dies, but when a main A or B feed dies (either to the rack or worse to a BDFB/CDBD) everything else on that panel is now being backfead via the clueless device. Cisco's CERENT 15454 SONET ADMs will try this for you and the result is typically just losing the 15454 needlessly when a main fuse feeding it blows because its own fuses can't carry the rest of the rack/room/whatever that it is trying to backfeed and the wrong one of those 2 fuses to the 15454 may well go first. And a SONET ADM is not what you want to have die - especially needlessly. We have been seriously considering making a small diode-or device just for the darn 15454s. Main fuses don't go often, but techs have been known to intentionally kill an A or a B feed for safety during maintenance when they are sure everything is dual fed. Just warn them about 15454s so they can kill the correct backfeed path first. NB that the fuse panel with diode OR-ed main inputs does fix the backfeed liability, but does not give you the backup of two individual device fuses for the single input device. - Original Message - From: james [EMAIL PROTECTED] To: [EMAIL PROTECTED] Sent: Monday, June 09, 2003 1:38 PM Subject: Black box that allows just an A or B DC feed *This message was transferred with a trial version of CommuniGate(tm) Pro* I have a device that only has one DC power input. Is there a box that can accept an A B DC leg, keep both isolated, and feed my device with just A or B ? ie, if the A or B leg fails at the co-lo this blackbox will switch to what ever leg is up. Quickly, I hope ! Or are there other ways to solve this issue ? James Edwards Routing and Security Administrator [EMAIL PROTECTED] At the Santa Fe Office: Internet at Cyber Mesa Store hours: 9-6 Monday through Friday Phone support 365 days till 10 pm via the Santa Fe office: 505-988-9200 or Toll Free: 888-988-2700
Re: DWDM interconnects
DWDM comes in many flavors, and I doubt it makes much sense to hand another carrier a fiber with a lot of different lambdas on it (if that is what you were asking). There are way too many variables. OTOH, if you were simply refering to buying the use of one wave (or lambda) that is a normal product for many companys. Unlike dark fiber, they have lit the route and are just selling you a lambda off their big DWDM system. They may choose to price it differently if you are running OC192 on it than if you are running OC48, bit in any case the Sonet ADM gear is up to you. Your connection is very apt to be a short range 1310 into them and their DWDM gear then will convert it to an ITU grid color up in the 1550 range and will coordinate signal levels, etc to eliminate crosstalk with adjacent channels. All you can mess up is your traffic. Lambda sales will become even more popular at major optical switching centers as realtime open markets evolve. Finisar is finally about to ship GBIC shaped pluggable optical devices in ITU grid colors. They will have actually, I think, two speed ranges, one for OC3 and OC12 and the other one for GIG-E and OC48. The exact application depends on the card you plug this generic device into. First big batch samples this month, and full production maybe April. The SFP (Small Formfactor Pluggable) units won't be getting ITU grid colors for over a year later. Finisar has been armtwisted into protecting the largest router arrogance, especially in the 80km GBIC product space. In the more plebian range down at the 10 km 1310 units or the local MultiMode units you will find more price competition and should pay respectively less than $200 and $100 even in small quantities. Molex doesn't make 80km units but does make the rest, and cisco as well as everyone else buys from both. Don't ever say whose equipment you are go to use a GBIC with, because they then may not sell it to you! The DOJ has to fit in here somewhere. On a metro area scale, I bet someone might sell you a lambda on a passive DWDM network to some building where your service didn't compete with theirs, and where you were using the same power same brand pluggable GBIC like devices and where there was no chance for your messing up their adjacent channels with too hot a signal. These pluggable devices will open up many options. But the long haul intelligent DWDM systems are juggling way too many variables and should be under one company's management. There is too much at risk and too easy to screw up. Of course two carriers can and will do whatever they want between themselves. We were the first carrier to drag an RBOC into an interoperability test with our cisco/cerent 15454s and their whatever. Cisco had not been certified til then to interconnect to any RBOC and was very eager. VZ, well, they did it because the letter from their legal dept said to. They used the Fugitsu FLMs in various sizes to test against just one Cerent using its wide range of cards. Now VZ is using Cerents themselves. After the grief we went through to get simple RBOC OC3, OC12, and OC48 interconnections blessed, I would hate to try adding DWDM to the mix. - Original Message - From: Pete Kruckenberg [EMAIL PROTECTED] To: [EMAIL PROTECTED] Sent: Monday, January 06, 2003 6:04 PM Subject: DWDM interconnects How common are DWDM interconnects between networks (carriers)? Is DWDM considered a reliable/scalable/operable carrier interconnection technology? Is multi-vendor DWDM (whether internal to the network or for carrier interconnection) practical or sensible, especially for carrier/network interconnection? Many vendors proclaim interoperability, but does that work in the real world? Pete.
Re: DC power versus AC power
A few points: Below 50 volts, anyone can do the wiring. No licensed electrician is needed im most jurisdictions. Total fault current available determines the damage that will be done when something like a wrench falls across bus bars. Time, too, counts. If you can't vaporize the whole wrench before the upstream fuse or breaker opens, then you just have scarred metal. For any given load, the AC wiring, running higher voltage than 48VDC CO batteries, will be smaller due to the lower amperage requirements and the larger typically allowed voltage drop. The AC network's source impedance is going to be a lot higher than a local battery string. too. A 600 amp capacity BDFB panel in a COLO room even just a 100' of cable loop length (50' end to end) from the main fuse panel at the batteries may well be wired with excess ampacity just to limit the IR drop. Instead of just a single 535MCM cable (that has adequate ampacity) that might mean up to 3x535MCM or 2x777MCM in parallel for both battery and return for both A and B battery feeds! That is a lot of copper. Look at the current available from a typical 9,000 ampere hour battery through just a single 100' length of 750MCM cable that has a resistance of .00150 ohms, assuming, for simplicity no resistance for the battery. That is 48 / .00150 = 32,000 amps, and that is very comfortable short term for a battery that can deliver 9,000 amps for an hour! The cable between the battery and the main fuse panel is unprotected other than by the ability to vaporise the straps between cells that are often substantially smaller in cross section than the cable. After the main fuse panel, typically there would be a 600 amp fuse protecting the feed to a BDFB Before that fuse goes, there is still plenty of energy to vaporize common hand tools. Typical 120/208V small branch circuit breakers in small buildings and homes have an interrupting capacity rated at 10,000 amps, and should not be deployed where that can be exceeded. It will be on the label. Some carriers have reservations about the small plug in (bullet pins on rear) BDFB panel breakers that in one case size range from a few amps to 100. Apparently there have been fires, and fuses are considered safer. Once you get past dry skin resistance, I have several times seen human body resistance given as 1 ohm. I have no idea how/where that is being measured.
Re: How do you stop outgoing spam?
A twist we saw spammers using on dialup accounts in Miami could come to cyber cafes and could be ugly. They were dialing in and then using the IP address to send spam out some other connection elsewhere where RPF wasn't in use. The return packets all came back on their dialup into us, but bypassed our filters that were then only on outbound packets. Since these were wholesaled dial ports, we know there are no valid servers customers needed in RIPE annd APNIC blocks and in long ACLs blocking various MSN servers, AND we know the dialup user's account. In a free cafe, you know none of that. Having an inbound mirror image of the outbound ACL helped initially, and then a coworker crafted a reflexive access list that really stopped them. Inbound packets had to have matching outbound ones or were tossed. We had visions of their finding a $spam$ friendly ISP that would sell them a SPAM OC-3 as long as he got no spam complaints. It could have served many spam machines running with dynamic IPs from many different ISPs and many user accounts on each - all at once. In the free cyber cafe that does not NAT and that does not know who the users are, there is potential for similar abuse.
Re: Sheilded Cat-5E Ground Loop - Myth or Reality?
None of his is specific to Cat-5e installations but is common to ALL electrical installations. This does NOT apply to telco cables run outside, often run on the same poles parallel to power wires for miles, grounded at many points in a MGN (Multi Grounded Neutral) environment where, like it or not, the earth carries considerable current. Opening a shield ground in this environment is bad. - Original Message - From: Kevin Oberman [EMAIL PROTECTED] To: Christopher K. Neitzert [EMAIL PROTECTED] Cc: [EMAIL PROTECTED] Sent: Wednesday, April 10, 2002 2:17 PM Subject: Re: Sheilded Cat-5E Ground Loop - Myth or Reality? *This message was transferred with a trial version of CommuniGate(tm) Pro* Date: Wed, 10 Apr 2002 17:43:50 + (UTC) From: Christopher K. Neitzert [EMAIL PROTECTED] Sender: [EMAIL PROTECTED] I'm in the process of managing cabling for a large install (500-ish runs) and a vendor came to me with a story about the creation of ground loops in running sheilded+gounded cat-5e in large installations. Does anyone have any experiences they would like to share regarding this? Just follow standard rules for grounding. If the shield is connected to anything, it should only be connected at one end! This is really always true, but is especially true when there is significant physical distance involved as this can result in current flow between the grounds. This will almost certainly create a significant hum field. Due to its excellent common mode rejection, this may not be a real problem, but it always deteriorates S/N margins to some extent. There is also a REAL safety issue! Make sure that ground is NOT exposed at the un-grounded end. A potential of many volts can occur, especially in areas subject to thunder storms. None of his is specific to Cat-5e installations but is common to ALL electrical installations. R. Kevin Oberman, Network Engineer Energy Sciences Network (ESnet) Ernest O. Lawrence Berkeley National Laboratory (Berkeley Lab) E-mail: [EMAIL PROTECTED] Phone: +1 510 486-8634
Re: long distance gigabit ethernet
The Cerent 454 (now cisco 15454) has 2 port Gig-E cards that cost a little more than a PA-GE card. The pair of ports shares OC-12 available to that slot (I'm assuming this is NOT an OC-192 equipped shelf) and the bandwidth can be split in certain multiples of STSes (OC-1), or used totally for one port. You only need as many STSes between boxes as you want to use. You can play some nice games with 802.1Q VLANs and multiple sites, too. There is a newer 4 port Cerent gig-E card I have not seen, but that probably can do a FULL gig-E on at least 2 of the ports (i.e. use a full OC-48 if the box has the OC-192 cross connect matrix installed). This newer card I think is only for point to point and does not understand VLANS, though probably can carry them. Some DWDM boxes have their own gig-E ports. We have Sycamore ones that give us several Boston, NYC, and Reston routers on the same ethernet. Consider also that many switches support what cisco calls ether-channel. If one gig-E isn't enough, add more in parallel. Any router on this ethernet can freely talk to any other. You are not stuck with just one router talking to one at the other end. - Original Message - From: Antony [EMAIL PROTECTED] To: Greg Pendergrass [EMAIL PROTECTED] Cc: Richard A Steenbergen [EMAIL PROTECTED]; 'Nanog@Merit. Edu' [EMAIL PROTECTED] Sent: Friday, March 22, 2002 12:28 PM Subject: Re: long distance gigabit ethernet On Fri, Mar 22, 2002 at 11:54:05AM -0500, Greg Pendergrass wrote: I'm going to take a stab and assume that you're actually more interested in finding a longhaul line with GigE on the ends, and not so much how many miles you can get with whatever optics... Absolutely right, I don't care what's in between as long as I have GigE at the end. Other options include using wave (too expensive), or ethernet over MPLS (worth considering although latency may be too high for longer that 1000 miles). there are solutions of this type. SURFNet line, currently used for test and network research is an example. It is from Amsterdam to Chicago. It is presented as GigE at the ends. So fairly long distance, RTT is 93 msec.Actually it terminate as SONET OC48 that goes too TDM Switch which has GigE interfaces. So there is SONET encapsulation in the middle. In theory we can get upto 2.5Gbps. Line is provided by Teleglobe. End equipements are CISCO, ONS 15454. This don't do any routing. This page may be interesting to browse. http://carol.wins.uva.nl/~delaat/optical/index.html You can probably find different variants of such non standard technology from other carriers. -antony
