FYI - potential network changes for atm this Sunday --- Forwarded mail from "Roth, Quilla (IRM/DIS)" <address@hidden> From: "Roth, Quilla (IRM/DIS)" <address@hidden> To: "'Mike Schmidt'" <address@hidden>, "Jacobs, Clifford A." <address@hidden> Cc: "Cypher, Stephen L." <address@hidden> Subject: RE: Internet2 Connectivity of "Weather Machine" Date: Wed, 19 Jan 2005 14:43:07 -0500 Mike/Cliff, Thanks for your very fast response and willingness to work with us on this. The current plan is to try to block commodity Internet access only from Sprint, so that if the laser is up, commodity Internet traffic can come in thru Qwest. We hope to implement this change during our maintenance period this Sunday morning, 1/23/05, between midnight and 8am). Steve Cypher, one of my staff, is trying to catch up with Raleigh Romine (who has worked with you on other "weather machine" issues in the past) to discuss the implementation plan a bit more. Last week Raleigh mentioned that we might not be able effectively prevent Internet swamping by blocking at the router, so I want to be sure we understand his thoughts before implementing a pointless change. Since I'm out Friday (and we're all out tomorrow, for Inauguration Day), Steve will let you know if our plans change. Also, Mike, should you need to phone Steve, his number is 703-292-8147. /Quilla ********************************* Quilla Roth Branch Chief National Science Foundation 4201 Wilson Blvd, Room 455 Arlington, VA 22230 P: 703-292-4295 F: 703-292-9146 E: address@hidden ********************************* -----Original Message----- From: Mike Schmidt [mailto:address@hidden Sent: Friday, January 14, 2005 4:55 PM To: Roth, Quilla (IRM/DIS) Subject: Re: Internet2 Connectivity of "Weather Machine" Quilla, In retrospect, I'm wondering if we could be allowed some commodity Internet access? I can see two possibilities, the one you mentioned about blocking access only if I2 is down. I don't know how easily one could implement that? The other possibility might be to limit the commodity Internet access to the Qwest link only. If the laser link goes away, we lose all networking at once. Comments? mike On Jan 13, 11:49am, Jacobs, Clifford A. wrote: > Subject: RE: Internet2 Connectivity of "Weather Machine" > Quilla, > > Thank you for your informative message. We (the Unidata folks and myself) > will work with you to find the right configuration for the current set up. > > Cliff > > -----Original Message----- > From: Roth, Quilla (IRM/DIS) > To: Jacobs, Clifford A. > Sent: 1/13/2005 11:34 AM > Subject: Internet2 Connectivity of "Weather Machine" > > Cliff, > > When you get back in town, I'd like to briefly discuss a change we're > planning to make that will occasionally affect external connectivity to > the GEO "Weather Machine". This server currently has a LOT of network > traffic (avg level of 36.0 - 36.5 Mbps) coming in to it from outside > NSF. It is, in fact, the highest external-trafficked device on the > network. The bulk (perhaps all?) of the traffic to this system comes > via the Internet2 (Int2) connection, which is a Qwest-provided OC3 > (155Mbps capacity) connection. Under our current routing scheme, if the > Int2 connection goes down, traffic to this machine automatically > reroutes through our 2 commodity Internet connections (Sprint and > Qwest), which are DS3s (45Mbps capacity each). The most common reason > for the Int2 connection to go down is some sort of problem (ironically, > usually weather-related) with the laser connection it uses. Because > this laser connection is also used by our Qwest Internet connection, > it's therefore very likely that when the Int2 connection goes down, so > does one of our Internet connections. When this occurs, we therefore > end up with 36Mbps worth of traffic swamping the remaining Sprint > 45Mbps-capable Internet connection. > > To prevent swamping Internet connectivity, we plan to implement a change > in our border router so that the Weather machine communicates ONLY with > Internet 2. I wanted to let you know and see what issues this might > cause. As I write this msg, I've realized that there may be some small > level of Internet traffic to the machine already, so we may need to > discuss exactly how we block things, so that Internet access is blocked > only if Int2 is down. > > We've had a couple of laser outages in the past month due to unusually > heavy fog. One occurred during the day, but was not a total outage. > The second, last night, was more solid. I'd like to have us positioned > so that should a solid one occur during the day, we don't end up > swamping the Sprint connection during a major deadline. > > /Quilla >-- End of excerpt from Jacobs, Clifford A. ---End of forwarded mail from "Roth, Quilla (IRM/DIS)" <address@hidden>
NOTE: All email exchanges with Unidata User Support are recorded in the Unidata inquiry tracking system and then made publicly available through the web. If you do not want to have your interactions made available in this way, you must let us know in each email you send to us.