Reliability & performance of NZ carrier WAN/Internet services
Hi all One of our customers is looking at migrating away from their current WAN carrier and have nailed down their options to either FX Networks or Gen-i. This is a 3 office site (AKL/WGTN/CHCH) with relatively fat PIP pipes between all offices, and 20/30Mbps to the Internet from the AKL & WGTN locations using carrier managed BGP devices to allow for Internet failover. Obviously the overview of services according to the provider themselves are understandably at least somewhat bias, but I was wondering whether anybody from the NZNOG community has any objective views on each with regards to reliability and performance? For example FX seems to cost significantly less for relatively the same thing, however without personally having a lot of experience in provider topologies I can't justify being able to go back to the customer vouching for either provider either way. Any replies, either on or off-list are appreciated. Regards, Josh Farrelly
Given Chorus has split off from into its own separate company and FX owns
its own fibre I would imagine FX would have less middlemen involved, which
could partially explain the price difference
On Tue, Jun 12, 2012 at 9:55 AM, Josh Farrelly
Hi all****
** **
One of our customers is looking at migrating away from their current WAN carrier and have nailed down their options to either FX Networks or Gen-i. This is a 3 office site (AKL/WGTN/CHCH) with relatively fat PIP pipes between all offices, and 20/30Mbps to the Internet from the AKL & WGTN locations using carrier managed BGP devices to allow for Internet failover. ****
** **
Obviously the overview of services according to the provider themselves are understandably at least somewhat bias, but I was wondering whether anybody from the NZNOG community has any objective views on each with regards to reliability and performance? For example FX seems to cost significantly less for relatively the same thing, however without personally having a lot of experience in provider topologies I can’t justify being able to go back to the customer vouching for either provider either way.****
** **
Any replies, either on or off-list are appreciated.****
** **
Regards,****
** **
Josh Farrelly**
** **
_______________________________________________ NZNOG mailing list NZNOG(a)list.waikato.ac.nz http://list.waikato.ac.nz/mailman/listinfo/nznog
-- Sam Russell Network Operations Research & Education Advanced Network NZ Ltd ddi: +64 4 913 6365 mob: +64 21 750 819 fax: +64 4 916 0064 http://www.karen.net.nz
Hi Josh,
Gen-I is an integrator who will probably be looking to sell you services on
Telecom's backbone, and FX is a network operator. Both Telecom and FX have
their own fibre present in AKL/WLG/CHC, same as the orange provider who
sells a product called PIP. These organizations have different approaches
to doing business.
FX is a Cisco based provider who sells Ethernet, wavelength, and dark
fiber, with their own network in two dozen or so major centres. FX spends
money by digging holes and putting cables in them, then putting moderately
priced equipment (German car price, not expensive house price) on the end
driving a handful of redundant 10gbps Ethernet links.
Gen-I/Telecom have their own fibre *everywhere* and have traffic loadings
orders of magnitude higher than FX. They have an Alcatel-Lucent
multi-service core that does all sorts of neat things, and does them well.
The gear on the ends of their cables (which are now mostly 50/50 owned with
Chorus) is house priced, and does everything. I'm sure if you needed every
13th packet set with the evil bit, it could be done.
The philosophies of network operation (provide Ethernet or provide
Everything) can lead to a pretty big difference in price. Hope this helps.
-JB
On Tue, Jun 12, 2012 at 9:55 AM, Josh Farrelly
Hi all****
** **
One of our customers is looking at migrating away from their current WAN carrier and have nailed down their options to either FX Networks or Gen-i. This is a 3 office site (AKL/WGTN/CHCH) with relatively fat PIP pipes between all offices, and 20/30Mbps to the Internet from the AKL & WGTN locations using carrier managed BGP devices to allow for Internet failover. ****
** **
Obviously the overview of services according to the provider themselves are understandably at least somewhat bias, but I was wondering whether anybody from the NZNOG community has any objective views on each with regards to reliability and performance? For example FX seems to cost significantly less for relatively the same thing, however without personally having a lot of experience in provider topologies I can’t justify being able to go back to the customer vouching for either provider either way.****
** **
Any replies, either on or off-list are appreciated.****
** **
Regards,****
** **
Josh Farrelly**
** **
_______________________________________________ NZNOG mailing list NZNOG(a)list.waikato.ac.nz http://list.waikato.ac.nz/mailman/listinfo/nznog
The philosophies of network operation (provide Ethernet or provide Everything) can lead to a pretty big difference in price. Hope this helps.
-JB
Well summed up Jon; note a slight clarification in that Gen-I sells WAN services that are Telecom Retail defined l2 network product offerings (GWS) that have some added layer3 stuff thrown on top by Gen-i. Chorus(2) at this stage have very little to do with these products other than provide the physical link over them. This of course most likely will change due to separation; time will tell. Also note there are also other players in this game: Kordia and Telstraclear would also be interested to talk to you.
On Tue, Jun 12, 2012 at 9:14 PM, Joel Wiramu Pauling
The philosophies of network operation (provide Ethernet or provide Everything) can lead to a pretty big difference in price. Hope this helps.
-JB
Well summed up Jon;
note a slight clarification in that Gen-I sells WAN services that are Telecom Retail defined l2 network product offerings (GWS) that have some added layer3 stuff thrown on top by Gen-i. Chorus(2) at this stage have very little to do with these products other than provide the physical link over them. This of course most likely will change due to separation; time will tell.
It's worth noting that Chorus are constrained to Layers 1 and 2 only - no routing!
On 12 June 2012 22:11, Andy Linton
It's worth noting that Chorus are constrained to Layers 1 and 2 only - no routing!
Which I personally would argue for various reasons is an untenable and ridiculous technical position to force anyone to take. L2 effectively is a routed service as it is implemented in the back end these days; "This here L2 Pipe goes to that there L2 pipe WAAAAY over there via XYZ intermediate systems" might as well be routing for all intents and purposes IMHO - and if they want to do it as a routed l3 service then why the hell not?
It's worth noting that Chorus are constrained to Layers 1 and 2 only - no routing!
Which I personally would argue for various reasons is an untenable and ridiculous technical position to force anyone to take. L2 effectively is a routed service as it is implemented in the back end these days; "This here L2 Pipe goes to that there L2 pipe WAAAAY over there via XYZ intermediate systems" might as well be routing for all intents and purposes IMHO - and if they want to do it as a routed l3 service then why the hell not? If I remember correctly the point of that distinction is to prevent Chorus(2) or LFC's from becoming a service provider to consumers. If so then let that statement stand, but the hell cares about the technical implementation detail IMNSHO or by sticking to the rules when it makes no sense. For example; How about a Big City Council - Auckland say ... Currently they will need to setup a shell company and become an RSP so they can provide services back to themselves for a multi-site situation. Is this desirable? I am not sure.
On Wed, 13 Jun 2012, Joel Wiramu Pauling wrote:
It's worth noting that Chorus are constrained to Layers 1 and 2 only - no routing!
Which I personally would argue for various reasons is an untenable and ridiculous technical position to force anyone to take.
It sounds ridiculous until you consider how many corner cases it solves. If Chorus and/or the Local Fibre Companies are allowed to offer routed (L3) traffic, then: - whose routing table do you use? - are they allowed to charge a premium (or a discount) for using "their" routing table rather than "yours"? - are they allowed to effectively charge less for a L3 service than for a L2 service? - are they allowed to charge "as if" all data went via a nominal aggregation point? - are they allowed to charge by byte (traffic), rather than by bit-per-second (bandwidth)? - if they're not allowed to charge for traffic, are they allowed to pretend your pipe is sized according to how you actually use it? - are they allowed to pretend you have multiple pipes, depending on route egress points? (e.g. peering / national / international) - if L3, then why not caching at L4? - are they allowed to charge more (or less) for cached data? - if data is routed to an any-cast address, how do you decide how much to charge for it? There are just *so* many levers that could be pulled to make life difficult for a "non preferred" wholesaler in such an environment. Writing a rule-book that would allow L3 routing but disallow the various shenanigans would be a tall ask for a bunch of techos, and pretty much impossible for a bunch of politicos. -Martin
On 13 June 2012 15:59, Martin D Kealey
On Wed, 13 Jun 2012, Joel Wiramu Pauling wrote:
It's worth noting that Chorus are constrained to Layers 1 and 2 only - no routing!
Which I personally would argue for various reasons is an untenable and ridiculous technical position to force anyone to take.
It sounds ridiculous until you consider how many corner cases it solves.
If Chorus and/or the Local Fibre Companies are allowed to offer routed (L3) traffic, then:
I would have to respectively disagree that there is really any more complication than already exists. Why are you invoking metering and caching or multicast for that matter and differentiated pricing it is still a UFB /MEF network underneath it. I think you are perhaps mistaking advocating for a VPLS at the Aggregation switches rather than E-PIPES (thus allowing switching before the traffic gets to the RSP handover upstream).. I would argue here also there are good arguments to at least allow for a Service Provider to have services bundled into an VPLS so that CPE to CPE traffic on the same SVLAN doesn't soak up backhaul... but I digress. All I was describing was a Layer3 routed WAN service, it would still be marked and routed[sic - switched] as the existing UFB constructs. There just happens to be a LFC run IPAM dishing out addressing at an RSP handover that happens to be an LFC. It solves for the cases that are going to transpire where XYZ organisation wants a WAN service between all it's newly connected UFB premises but doesn't want to have to deal with APNIC/Address allocations etc. I get the argument that that is a Market opportunity for an RSP to manage such a service on behalf... but it just seems like it is going to cause problems artificially limiting LFC from providing **ANY** sort of l3 services. I think regardless of the WAN service scenario described that it would be desirable to have a V6 management/monitoring portal. If you do this you can serve up things like Customer Self provisioning portals - which ultimately means more choice and faster provisioning. It would also make management and performance monitoring easier. Jonathan - there is always l2tpv3 .
From: nznog-bounces(a)list.waikato.ac.nz [mailto:nznog- bounces(a)list.waikato.ac.nz] On Behalf Of Joel Wiramu Pauling Sent: Wednesday, 13 June 2012 3:25 p.m. To: Andy Linton Cc: nznog(a)list.waikato.ac.nz Subject: Re: [nznog] Reliability & performance of NZ carrier WAN/Internet services ---8<--- L2 effectively is a routed service as it is implemented in the back end these days; "This here L2 Pipe goes to that there L2 pipe WAAAAY over there via XYZ intermediate systems" might as well be routing for all intents and purposes IMHO
There is an important difference though: L2 forwarding is done based on MAC address and L3 forwarding on IP address. L3 forwarding will only pass IP. L2 forwarding will pass (hopefully) any Ethernet frame. This is useful if you want to run anything other than IP, like 802.1ag fault management. Jonathon This email and attachments: are confidential; may be protected by privilege and copyright; if received in error may not be used, copied, or kept; are not guaranteed to be virus-free; may not express the views of Kordia(R); do not designate an information system; and do not give rise to any liability for Kordia(R).
participants (7)
-
Andy Linton -
Joel Wiramu Pauling -
Jonathan Brewer -
Jonathon Exley -
Josh Farrelly -
Martin D Kealey -
Sam Russell