- NANOG - lists.nanog.org

RE: QoS/CoS in the real world?
by Jeff Hancock 25 Jul '02

25 Jul '02

Steve, Hope this info helps answer your questions about QoS, implementations and customers. Forwarded from a product person person in our org... ------------------------------------------------------------------------ Sorry I didn't see this note earlier, but wanted to make you aware that Masergy Communications is actually offering such a service on a native MPLS based IP network. We provide differentiated IP services via customer DSCP marking at the network edge. QoS is supported end to end through the Masergy core via promotion to the MPLS EXP marking. Masergy closely manages its network by service class. This allows each marking to have its own end-to-end SLA, customized to the type of customer traffic sent with each marking. Customers see the need for QoS in two broad categories: 1) Prioritizing business applications for performance reasons 2) Providing guaranteed performance to real-time IP applications such as IP voice and IP videoconferencing Some real examples: A Masergy customer does file backups overnight. When the backups continued into the next morning performance of daily business activities suffered. By lowering priority of the backup traffic, acceptable performance for both the backups and day users can be provided at a lower cost to the customer. Most of our customers have similar stories (the p2p example mentioned previously is another good one). An interesting application is that customers can mark all outbound traffic as priority--this is a simple config and requires little smarts on the part of the edge router. Any traffic that originates and terminates on the Masergy network is prioritized. All traffic from outside, non-business sites (i.e. surfing, p2p, radio etc.) gets best-effort treatment. Note that many of the applications that need priority are not high bandwidth--MS Exchange for example is a low BW app, but notoriously sensitive to network quality issues. QoS in the manner described above can enhance performance even for lower BW applications. Another customer application is video conferencing - specifically replacing current ISDN video architectures with IP equivalents. IP QoS and MPLS allow Masergy to engineer a class of service for voice and video that provides low jitter and 100% guaranteed throughput across our core. MPLS fast fail-over improves application performance in the case of a core network link or hardware failure. Without differentiated QoS, we would not be able to guarantee this level of performance. One of the major issues with properly utilizing QoS is giving the customer the ability to view and manage performance. Masergy customers use the Service Control Center - a secure, web-based interface for managing their service. It provides per QoS level and application statistics on network utilization and performance. Customers can change their access bandwidth and enable additional QoS capabilities in real time. http://www.masergy.com ------------------------------------------------------------------------ -- ----------------------------------------------------------------------- Jeff Hancock P: 703-846-0161 Senior Engineer F: 703-846-0149 Masergy Communications, Inc. C: 2901 Telstar, Ct. E: mailto:jeff@masergy.com Falls Church, VA, 22042 W: http://www.masergy.com ----------------------------------------------------------------------- -----Original Message----- From: Stephen J. Wilcox [mailto:steve@opaltelecom.co.uk] Sent: Thursday, July 18, 2002 5:26 PM To: John Evans Cc: nanog(a)merit.edu Subject: RE: QoS/CoS in the real world? On Thu, 18 Jul 2002, John Evans wrote: > > I realise this is a US-centric list, however, a significant number of > providers in Europe have deployed Diffserv as a means of supporting > (and > selling) differential SLAs. Of these, some have deployed Diffsev at the > edge and some both the edge and core. See Clarence Filsfils presentation at > NANOG 25 for a description of typical core deployments. > > > 2. Hype aside, to what extent do customers actually want this > > Surely end customers want a service with SLAs that will support their > applications, and at low cost? It then becomes a provider cost > consideration as to whether these SLA assurances can most > competitively satisfied with mechanisms such as Diffserv or without. I have to say that the majority of users barely understand how their outlook client works let alone the difference between applications. I'm starting to think theres no demand for these services other than that which the hype says is there. THis is in line with what people said about using qos behind the scenes but customers dont know.. kind of what I thought to begin with STeve > > I conclude either the people doing this are successful and keep > > their secret safe or the world is yet to sell largescale QoS across > > IP. > > or perhaps they are just not on this list. > > cheers > > John > > > -----Original Message----- > > From: owner-nanog(a)merit.edu [mailto:owner-nanog@merit.edu]On Behalf > > Of Stephen J. Wilcox > > Sent: 14 July 2002 00:47 > > To: nanog(a)merit.edu > > Subject: Re: QoS/CoS in the real world? > > > > > > > > Well, end of the week and the responses dried up pretty quickly, I > > think thats a response in itself to my question! > > > > Okay, heres a summary which was requested by a few people: > > > > Other people too are interested in my questions, they dont implement > > QoS in any saleable manner and wonder how it can be done and whats > > actually required. > > > > A number of people think QoS was interesting for a while but that > > its never either found its true use or is dead. > > > > There are unresolved questions from a customer point of view as to > > what they are actually going to get, what difference it will make > > and how they can measure > > their performance and the improvements from QoS. > > > > There is a real demand for guaranteed bandwidth, however this tends > > to be in the form of absolute guarantees rather than improvements > > above "normal" hence ATM remaining a popular solution. > > > > There is a requirement to differentiate voice traffic, however this > > is necessarily done by the network anyway in order to offer the > > service, this being the case the customer doesnt pay extra or gets > > to know much about how all the > > fancy bits are done. > > > > > > On the face of it this is all negative. Nobody has responded saying > > there are genuine requirements for services to be offered to > > customers. Nor has anybody > > responded with any descriptions of implementations. > > > > I conclude either the people doing this are successful and keep > > their secret safe or the world is yet to sell largescale QoS across > > IP. > > > > Steve > > > > > > On Mon, 8 Jul 2002, Stephen J. Wilcox wrote: > > > > > > > > Hi all, > > > I've been looking through the various qos/cos options > > available, my particular > > > area was in how IP (MPLS perhaps) compares and can be a > > substitute for ATM. > > > > > > Well, theres lots of talk and hype out there, from simple IP > > queuing eg cisco > > > priority queuing, rsvp, diffserv, mpls traffic engineering etc > > > > > > But two things are bugging me.. > > > > > > 1. To what extent have providers implemented QoS for their > > > customers > > > > > > 2. Hype aside, to what extent do customers actually want this > > (and by this I > > > dont just mean that they want the latest QoS because its the > > 'latest thing', > > > there has to be a genuine reason for them to want it). And this > > takes me back to > > > my ATM reference where there is a clear major market still out > > there of ATM > > > users and what would it take to migrate them to an IP solution? > > > > > > Also, how are people implementing bandwidth on demand (dynamic > > allocation > > > controlled by the customer) solutions to customers > > > > > > Cheers > > > > > > Steve > > > > > > > > > > > > > > > > > >

2 1

RE: AS286 effectively no more..
by Huopio Kauto 25 Jul '02

25 Jul '02

>> What is the legal position of an IRU deal if the cable owner goes belly >> up? >Unless someone buys the equipment and agrees to theke the IRU:s on - they >are worthless. How about duct IRU:s? --kauto

2 1

RE: AS286 effectively no more..
by Huopio Kauto 25 Jul '02

25 Jul '02

> >Does anyone know what happened to the Ebone/KPNQWEST European-wide DWDM >system? I figure that if it was shut down, we would see more impact. > >Their IP network load I bet was quite easily handled by other operators >considering the huge over-capacity situation we have had the past years. Parts of the Eurorings (the KPNQwest part of the KPNQwest/Ebone network combination) seem to be semi-active - otherwise bits wouldn't flow from Qwest to Leiden in NL. But there are just KPNQwest core AS/KPN Belgium/ Luxembourg prefixes advertised on the IP side. I am worried on the networks that have swapped/IRU:ed fibers from KQ and have transmission gear co-located at KQ amp huts / major nodes, using KQ rectifiers to get DC juice to their racks. It is going to be interesting to watch if/when the still-lit remains of the Eurorings go dark on transmission level - I've understood there were couple of _very utilized_ WDM rings in KQ central european network.. What is the legal position of an IRU deal if the cable owner goes belly up? --kauto

5 5

Re: Re: PASSIVE [D]WDM... Like, Cisco 15216.
by Frank Coluccio 25 Jul '02

25 Jul '02

It would be nice if Transmode spent a little more attention to the detail in their data sheet and less on their art work. I can't discern what their CWDM spacings are either. If you return to the site you'll note that the 1.25 and 2.5 Gxx numbers you cited were actually *bit rates*, not frequency separations on a grid. The ways in which grids are defined for DWDM and CWDM differ, with DWDM grids typically defined on the basis of frequency spacings. E.g., spacings between the center frequencies of lambdas at 200GHz, 100GHz and 50GHz proposed for the ITU grids. The proposed CWDM Grids, on the other hand, will use *wavelength* spacings (effectively the same thing after you do the arithmetic, it's just noted differently). See the following TIA (Telecommunications Industry Association) paper at the url below. E.g., CWDM grid spacings may be at 20 nanometers (20 nm), whereas a DWDM spacing might be 100 GHz. http://ftp.tiaonline.org/fo-2/fo27/Public/2002-01-Kauai/FO27-2002-01-TD12.d… I don't know how much this helps in the search for a mix and match wavelength conversion assembly, but it at least lays out all known and proposed grid formats so that we have a more factually-based starting point. FAC > > On Thu, 25 Jul 2002, Simon Lockhart wrote: > > > I'm currently using the 15454 to wavelength convert OC48 signals, but have > > not to date seen a black-box wavelength convertor - I would also be interested > > to know if such a beast exists. I think if you want to do this, you're stuck > > with the Metro 1500 (or equivalent from someone else), which becomes very > > expensive quickly. > > Transmode <http://www.transmode.com> has both 1.25Ghz and 2.5Ghz units > with where you pretty much can pick and choose your optics from 850nm MM, > 1310, 1510, 1530, 1550 and 1570nm when you order. This is not a complete > ITU grid from what I can find (I don't know exactly what ITU grid is but I > found some specs) but rather for CWDM use. > > There should be quite a few manufacturers making units like these, I know > the MRV people does it as well (or some company within MRV). > > -- > Mikael Abrahamsson email: swmike(a)swm.pp.se > >

1 0

PASSIVE [D]WDM... Like, Cisco 15216.
by Alex Rubenstein 25 Jul '02

25 Jul '02

We're looking to get glass between three buildings, and looking closely at the 15216 (passive WDM, ie, a prism). A couple of rambling questions, that perhaps folks here have experience with. First, has anyone had experience with ITU Grid Optic GBICs? Do they even exist? Second, does anyone know of a 'back-box' that will take standard 1310 or 1550 nm signal, and 'convert' it to an ITU grid? Any pointers, help, or a kick in arse would be appreciated. -- Alex Rubenstein, AR97, K2AHR, alex(a)nac.net, latency, Al Reuben -- -- Net Access Corporation, 800-NET-ME-36, http://www.nac.net --

3 2

RE: password stores?
by Shawn Solomon 25 Jul '02

25 Jul '02

One common solution is a hash based on the cpe site name or some other unique key provided by the cpe information (address, ph #, etc). Changing the hash occasionally provides new passwords, and it is all easily scripted.. -----Original Message----- From: Daniska Tomas [mailto:tomas@tronet.com] Sent: Tuesday, July 23, 2002 2:35 AM To: nanog(a)merit.edu Subject: password stores? hi, i'm wondering how large isps offering managed cpe services manage their password databases. let's say radius/tacacs is used for normal cpe user aaa, but there is some 'backup' local user account created on the cpe for situations when the radius server is unreachable. for security reasons, this backup account (as well as snmp communities, radius key etc.) is unique per cpe to avoid frauds caused by end-users (even if one does password recovery on the cpe, they still don't have the password for other cpe's). if there are hundreds or thousands of these cpe's that could mean storing of tens thousands of password. are there any crypto-based products available or do the people use their own stuff? thanks -- Tomas Daniska systems engineer Tronet Computer Networks Plynarenska 5, 829 75 Bratislava, Slovakia tel: +421 2 58224111, fax: +421 2 58224199 A transistor protected by a fast-acting fuse will protect the fuse by blowing first.

2 1

Sunspot Activity & Radio Blackouts
by Andy Ellifson 24 Jul '02

24 Jul '02

For anyone that operates a wireless network or a copper based network: Official Space Weather Advisory issued by NOAA Space Environment Center Boulder, Colorado, USA SPACE WEATHER ADVISORY BULLETIN #02- 2 2002 July 23 at 12:00 p.m. MDT (2002 July 23 1800 UTC) **** ( CORRECTED ) MAJOR SUNSPOT ACTITVITY **** A major sunspot region has rotated onto the visible face of the sun. This region, designated as Region 39 by NOAA Space Environment Center forecasters, is believed to have been the source of three large coronal mass ejections on the far side of the sun beginning on July 16. This region will rotate across the visible side of the sun over the next two weeks and is expected to produce more solar activity. Since appearing on the visible side yesterday (July 22) this region has already produced a major flare at 6:35 pm Mountain Daylight Time (MDT) on July 22 (0035, July 23 UTC). Radio blackouts reached category R3 (Strong) on the NOAA space weather scales. In response to the major flare, a geomagnetic storm is possible and is expected to begin between 8:00 pm MDT on July 23 and 8 am MDT on July 24 (0200 - 1400, July 24 UTC). The geomagnetic storm may reach category G2 (moderate) levels on the NOAA space weather scales. Category R3 radio blackouts result in widespread HF radio communication outages on the dayside of the Earth and can also degrade low frequency navigation signals. Category G2 geomagnetic storms can lead to minor problems with electrical power systems, spacecraft operations, communications systems, and some navigational systems. Aurora Borealis / Australis (northern / southern lights) may be seen down into the mid latitudes (New York, Madison, Boise, Vladivostok, Rome, Tasmania, Wellington - NZ, Puerto Montt - Chile) Data used to provide space weather services are contributed by NOAA, USAF, NASA, NSF, USGS, the International Space Environment Services and other observatories, universities, and institutions. For more information, including email services, see SEC's Space Weather Advisories Web site http://sec.noaa.gov/advisories or (303) 497-5127. The NOAA Public Affairs contact is Barbara McGehan at Barbara.McGehan(a)noaa.gov or (303) 497-6288.

4 3

Re: IGP metrics on WAN links
by Jennifer Rexford 24 Jul '02

24 Jul '02

> Just curious as to what people are using for metrics in their IGP > and what their reasons are; bandwidth? geographical distance? latency? We have a survey paper on techniques for setting IGP weights http://www.research.att.com/~jrex/papers/ieeecomm02.ps http://www.research.att.com/~jrex/papers/ieeecomm02.pdf and a longer version http://www.research.att.com/~jrex/papers/ieeecomm02.long.ps http://www.research.att.com/~jrex/papers/ieeecomm02.long.pdf that might be of interest. We discuss how to tune the IGP weights in a systematic, automated fashion based on measurement data (of topology and traffic) and an optimization algorithm, and show that good weights settings allow IGPs like OSPF and IS-IS to perform almost as well as optimal routing schemes that have complete flexibility in selecting paths for the traffic. -- Jen

1 0

Boston Qwest Issues?
by Eric Gauthier 24 Jul '02

24 Jul '02

Heya... Anyone know if something is up on Qwest's run from Boston to NYC today? Eric :)

1 0

Juniper security appnote + martians
by Stephen Gill 24 Jul '02

24 Jul '02

Gents, I thought I would pose the martians question here as well... I'm trying to find out additional information on the reasoning behind adding these martians to the Juniper's security appnote found on their website: Prefix Description 19.255.0.0/16 Ford Motor Company 129.156.0.0/16 Sun Microsystems 192.5.0.0/24 no match 192.9.200.0/24 no match 192.9.99.0/24 Sun Microsystems I don't see a single reference to these in Cisco's IOS Essentials www.cisco.com/public/cons/isp/documents/IOSEssentialsPDF.zip , Bill Manning's draft, www.ietf.org/internet-drafts/draft-manning-dsua-08.txt or Rob T's Bogon List. www.cymru.com/Documents/bogon-list.html I base my bogon filtering for the JUNOS Secure Template and JUNOS Secure BGP Template at www.qorbit.net/documents/junos-template.pdf www.qorbit.net/documents/junos-bgp-template.pdf www.qorbit.net/documents/junos-bgp-appnote.pdf on Rob's list. What are your thoughts on filtering the above prefixes? Are some of these worthy of being added to the master bogon list? Now, on to some of Juniper default martians: 128.0.0.0/16 191.255.0.0/16 192.0.0.0/24 223.255.255.0/24 These prefixes seem to be based on www.ietf.org/internet-drafts/draft-iana-special-ipv4-03.txt. I'm curious what the reasoning is behind selecting these prefixes only. Also, given that these may be allocated in the future (per the draft) what are your thoughts on having these in Juniper's default config? Perhaps these would be good additions to a dynamic (up-to-date) bogon list instead of a static placement in JUNOS even though they can be overridden if necessary. Thoughts? -- steve

2 3