- NANOG - lists.nanog.org

Recommendations, Colo Reno, Albuquerque, Phoenix, Las Vegas
by Eric A Louie 21 Jun '16

21 Jun '16

Does anyone have recommendations for Colocation space in any of those 4 cities? thanks Eric

4 3

Measuring the quality of Internet access
by Max Tulyev 21 Jun '16

21 Jun '16

Hi All, I know there are many people from many countries. Do you know something about mandatory measurements of Internet access quality from country telecom regulators? If yes, could you please share that information with me? I found ETSI EG 202 057-4 standard (http://www.etsi.org/deliver/etsi_eg/202000_202099/20205704/01.02.01_60/eg_2…) but in fact it is about measurements inside operator's network, not Internet access itself. Is it possible in general to measure the quality of Internet access? And if yes - how?

9 12

Netflix banning HE tunnels
by Davide Davini 21 Jun '16

21 Jun '16

Today I discovered Netflix flagged my IPv6 IP block as "proxy/VPN" and I can't use it if I don't disable the HE tunnel, which is the only way for me to have IPv6 at the moment. But the fun part has been Netflix tech support: "Oh I see, yeah we have been receiving reports of some other members with ipv6 having this issues, at the moment Netflix is not really designed to work with ipv6 connections, in this case I can recommend you two things, one is to turn off the ipv6 and the other one will be to contact directly with Hurricane Electric, there are some customers that were able to use Netflix with an ipv6 under some specific settings set by Hurricane Electric." I don't obviously expect HE to fix it, I don't pay for shit, it's a free service, why should they? But it's fun to know that " Netflix is not really designed to work with ipv6 connections ". Who did it say on this ML that the best way to solve these issues is Netflix tech support? :) Ciao, Davide Davini

42 99

Sri Lanka options
by Gavin Henry 20 Jun '16

20 Jun '16

Dear all, One of our partners[1] customers is based in Sri Lanka and we're having issues with latency creeping up to them. We're usually good via our tier 1 transits (Level3 and NTT) at around 180ms, but it's now up to 260ms and causing issues for their tunnels into the kit our partner hosts for them in one of our UK racks in Telehouse East. We've reached out directly to SLT.LK who provide their leased line out there and they adjusted routing into Telia, then into Level3 the last time this happened. We've tried again with no response. So I'm looking for options. They are on AMX-IX, but not sure about being on the RS. That may not improve anything as we'd need to get up there first. I'll contact our tier ones and companies we already have NNI's with but wanted to check if anyone can offer me anything? I know this is a US list, but hoping others have done something like this. Not many ideas from UKNOF apart from a kind email from Rod Beck, which I'm pursuing: ======== There is a surprising amount of connectivity into Sri Lanka. http://www.submarinecablemap.com/#/country/sri-lanka SeaMeWe-5 Bay of Bengal Gateway (BBG) SeaMeWe-3 SeaMeWe-4 Bharat Lanka Cable System Dhiraagu-SLT Submarine Cable Network WARF Submarine Cable ======== We need a guaranteed 5mb/s pipe with < 200ms or something similar. Ideas? Thanks for reading, Gavin. [1] http://www.surevoip.co.uk/partners/partner -- Kind Regards, Gavin Henry. Winner of the Best Business ITSP (Medium Enterprise) 2016! http://www.surevoip.co.uk/2016-best-provider OpenPGP (GPG/PGP) Public Key: 0x8CFBA8E6 - Import from hkp://pool.subkeys.pgp.net or http://www.suretecgroup.com/0x8CFBA8E6.gpg

1 0

RPKI implementation
by Jakob Heitz (jheitz) 20 Jun '16

20 Jun '16

During the RPKI presentation there was a question about resilience of the router if the RPKI cache loses connectivity. The IOS-XR implementation allows multiple caches to be configured. When a cache loses connectivity, the entries from that cache are purged after a time interval. Default is 60 seconds and it is configurable. A lookup of a prefix that is not loaded will return not-found. 5 seconds after the latest RPKI database update, a refresh request is sent to each neighbor, provided that the neighbor either: - dropped any received route due to a policy that contains validation-state, or - received a route, the validation state of which changed. If soft reconfiguration inbound is configured, then the refresh is avoided, because the received paths are stored. Thanks, Jakob.

3 7

NOC Contact Megafon AS31133
by Marco Paesani 20 Jun '16

20 Jun '16

Hi, I need a contact person NOC Megafon AS31133, can you help me ? Thanks ! Kind regards, Marco Paesani Skype: mpaesani Mobile: +39 348 6019349 Success depends on the right choice ! Email: marco(a)paesani.it

1 0

Re: cross connects and their pound of flesh
by Brandon Butterworth 19 Jun '16

19 Jun '16

Dave Temkin <dave(a)temk.in> wrote: > And as colo operators get freaked out over margin compression on the > impending 10->100G conversion (which is happening exponentially faster than > 100->1G & 1G->10G) they'll need to move those levers of spend around > regardless. If they've based their model on extracting profit proportional to technology speed then they've misunderstood Moore's law brandon

7 7

Re: cross connects and their pound of flesh
by Brandon Butterworth 19 Jun '16

19 Jun '16

> From: "Patrick W. Gilmore" <patrick(a)ianai.net> > Colo providers are absolutely worried about drops in xconn revenue. > Look at some large colo providers who are public and split out their > numbers. Youll see that the percentage of their profit from xconns is > usually more than double the percentage of their revenue from xconns. Perhaps their investors should have looked closely at those figures and priced according to risk. > Put another way, if xconn revenue drops by 10%, their profit drops by > over 20%. How many public companies can shrug off a 20% drop in EPS? I > submit: Not very many. Tough, revenues can go down as well as up. If you gouge your customers they will try harder to find ways not to give you money. If I have to do an xcon I prefer to do it in Telehouse rather than Telecity as Telecity is 100% more install and 32% more annually (Telehouse also discount for extra cores). Some USA DC are way more expensive than either of these UK DC (for now, UK DC only moved to the recurring fee model after it'd been established elsewhere) They've probably been using xcon revenue to make it look like they are still growing quickly to inflate share price. A few aquisitions can hide reality for a bit longer. Capitalism is great but it does lead to unreasonable expectations such as never ending growth xcon revenue worked for Telecity to get a nice buyout, be interesting to see how their new owners handle it. brandon

1 0

Re: IP and Optical domains?
by Gert Grammel 19 Jun '16

19 Jun '16

Glen, Here's a list of technical issues for whatever it is worth: Transport networks assume a service to be a circuit with two end-points which optionally can be protected by some transport mechanisms. Each of such services is unrelated to others. In a routed network a transport service corresponds to a link adjacency between routers. To design a resilient router network it is desired that links between different pairs of routers are unrelated to each other and don't fail together (are shared-risk-link-group diverse). Hence links are related in being unrelated. Now let's start: 1) transport folks don't disclose their physical topology: no chance to design a reliably resilient network on top. Workaround: ask for protected transport links. This increased availability comes with the cost of duplicating transport resources. Actually it is often more than double since the protecting circuit is a bit longer than the protected one. 2) transport folks do disclose the physical topology including nodes. Considering those nodes as single point of failure, it requires more BW available in Routers to cover failures and tricky configuration. 3) alternatively you could ask to install more transport equipment to avoid SPOF which obviously comes with a price tag too 4) once settled, the Transport network may grow. This means that circuits once in a while get moved to other resources. Typically the result of a transport network re-planning, a protection action or simply a forgotten reversion after a maintenance event. Consequently, whatever was achieved in 2) or 3) may just become void over time. There are basically two ways to deal with the complexity: 1) create an information exchange that let networking and transport controllers take informed decisions: You may want to look at https://datatracker.ietf.org/doc/draft-ietf-teas-interconnected-te-info-exc… 2) forget about switching anything in Transport, nail the structure down in config and move on. You may like then https://meetings.ripe.net/see3/files/Ian_Farrer-The_Terastream_Native_IPv6_… If you now think transport is all the same, irrespective of technology, then I suggest to take another breath: 1) since multiplexing lambdas is passive, Operators often do not count WDM or ROADM nodes as SPOF. Note: multiplexing is indeed passive but followed by Amplifiers which are not. 2) in SONET one could connect any vendor's node to everyone else's e.g. With OC48. This is not true in DWDM. While the Ethernet connecting to a wavelength is a standard, the wavelength itself is proprietary and linked to the component vendor of the transceiver. Transponders are in fact acting as adaptor plugs between standard and proprietary transport. As a result, if you find a left-over Gen-1 transponder and want to hook it up to a Gen-3 transponder of the same system vendor - it may not work because the component supplier changed. It also means you can't simply re-structure a DWDM network in an on-demand manner if such effects are present. I leave it up to you to draw your own conclusions. Gert Sent from my Apple ][)

1 0

Re: cross connects and their pound of flesh
by Dave Temkin 19 Jun '16

19 Jun '16

On Sat, Jun 18, 2016 at 12:54 PM, Brandon Ross <bross(a)pobox.com> wrote: > > > Value based pricing is all the rage these days, which is why they charge > you so much for cross connects. Exactly. Not that I don't like free cross connects (they're the bees knees, in fact), but at the end of the day, an existing colo operator is not going to go from paid->free cross connects without extracting that pound of flesh (read: sweet sweet 100% pure margin) from somewhere else. Your space and/or power prices will go up to backfill that lost profit. That said, those of us that buy a decent amount of colo prefer to trade in the value of the asset leased/purchased - space & power - as we have real world indexes to tie the underlying cost to for negotiation purposes. And as colo operators get freaked out over margin compression on the impending 10->100G conversion (which is happening exponentially faster than 100->1G & 1G->10G) they'll need to move those levers of spend around regardless. -Dave

2 1