I came across the idea of the virtual POP , but the website for them have
way too much jargon to me[1][2][3], can someone explain it like i'm five
(:-D)?
Specifically, my question is :
1. Is virtual POP basically a L2VPN? That is, the provider will provide a
port at site A,that is somehow connected to LAN of site B ? What's
difference with vpop and layer 2 transport then?
2. Do such vPOP have guaranteed latency/bandwidth?
3. Is that really useful? If I'm already buying transit bandwidth/announce
my blocks from provider, the site B peers is already going to send traffic
through provider's backbone to site A, then what's the difference?
I came across the idea of the virtual POP , but the website for them have
way too much jargon to me[1][2][3], can someone explain it like i'm five
(:-D)?
A virtual Point Of Presence means that you provide services at a
location via someone else's facilities.
The classic example was extending a PRI for dialup modems inside a
particular local calling area via a point-to-point T1 back to your
modem bank somewhere else that would have been a long distance call
for those customers. If you put a modem bank in their local calling
area, it's a POP. If you extend the circuit from their local calling
area back to your modem bank elsewhere, it's a virtual POP.
Modern examples of virtual POPs are much fancier but it's the same basic idea.
1. Is virtual POP basically a L2VPN?
It can be. Depends on what service you're extending from the "virtual" location.
2. Do such vPOP have guaranteed latency/bandwidth?
Depends on what you're extending and how.
3. Is that really useful?
It can be. It can let you dip your toes in a market without a large
up-front investment in equipment and backhaul.
Yes, except it is done via Switched Ethernet and VLANs. The idea behind virtual peering. Your gear is in Amsterdam and someone gives you VLANs to LINX.
I understand on layer 2 or like william point out (on anything other than
IP) it make total sense.
However on layer 3, with existing transit bandwith with said provider it
would be redudant. (Assume The one you wanted peer at site b is already
peering with your provider).
This could be done with an OPS and diverse fibers or a dual transponder type solution or an Ethernet ring amongst other solutions.
It generally means to router geeks that a router isn't there, but some other technology is in use. Often it is a growing market or declining market for the provider.
The term "virtual PoP" is more commercial than it is technical.
As William mentioned, you are providing services via someone else's
infrastructure. It is between you and that other network to determine
how much of their infrastructure you will depend on.
But ultimately, the objective is for you to reduce your exposure in what
you would consider a new venture that still needs some proofing.
Different providers use the term with different definitions, but this is how we use it:
At Level 3, a VPOP is a POP that we operate under someone else's license. For example, we have VPOPs in a number of markets throughout the Asia Pacific region, including countries like China, Vietnam, Indonesia, and others. We are buying a service from a partner that has an operating license in that country where they provide routers, entrance facilities, colo and other related infrastructure items, but we otherwise operate it as a full POP. It's in our OSS/BSS systems like any other location.
As far as our customers can tell, there is nothing virtual about it. It looks like any other node on our network, so the distinction is purely internal to our company and how we have to manage support for the site.
The key is really that it could mean different things for different providers, although I would agree that the gist is that the location is enabled to look and feel like a POP without the provider installing the full complement of requisite hardware. A provider I worked at in the past, for example, defined a virtual POP as a non-POP location at which POP pricing was offered - the actual method of delivery there being both irrelevant to it being defined that way and unimportant to the concept as a whole. It let the company be price-competitive with others that may have made more extensive investments in hardware at higher-demand locations, and it was purely based on a business justification. There was no specific technical definition (although in reality we were transparent with our customers about methodology anyway) - this contrasts with other providers that are clearly using it in a way that does define a technical approach. It's just an approach specific to that provider.
At Level 3, a VPOP is a POP that we operate under someone else's
license. For example, we have VPOPs in a number of markets throughout
the Asia Pacific region, including countries like China, Vietnam,
Indonesia, and others. We are buying a service from a partner that
has an operating license in that country where they provide routers,
entrance facilities, colo and other related infrastructure items, but
we otherwise operate it as a full POP. It's in our OSS/BSS systems
like any other location.
how does this work for mpls vpn based services across continent/country?
i.e. are there inter-provider mpls vpn issues?
Interestingly, being a relatively young player in the MPLS space, we see
Africa slowly moving away from typical l3vpn services as more
countries/cities/metros get fibre, as a result of businesses moving
their IT infrastructure into some kind of cloud. It's not happening
quickly, but it is happening, and I think what the SD-WAN (yet another
buzz word) boys are doing will only accelerate the process.
That has meant that the only progressive inter-provider MPLS NNI's we
are seeing in/for the region are l2vpn's, and despite all the promises
of BGP-driven l2vpn NNI's, I'd say all the l2vpn NNI's we are setting up
are simple back-to-back VLAN's. Can't possibly mess those up :-).