 |
|
PeterStar data services ensure quality and reliable communication at a rather flexible pricing system providing for reductions depending on the total capacity of subscriber terminals purchased.
|
MPLS Technology
MMPLS means Multiprotocol Label Switching.
MPLS technology consists in simplifying the process of data packets routing
when being transited via the provider’s backbone network. Nothing changes for
a customer and the structure of the customer’s networks; however due to MPLS
technology extra positive properties are added to the customer’s virtual private
network (VPN) (see Section “PeterStar Virtual Private IP Networks”).
Compared favorably with other methods of VPN constructing, for instance VPN
based on ATM/FR or IPSec, MPLS VPN represents high scalability, possibilities
of automatic reconfiguration and natural integration with other IP services,
which are at present compulsorily rendered by every successful provider: Internet
access, Web and mail services, hosting. Using MPLS makes it possible to create
customers’ isolated networks not dependent on each other. ÌÐLS ensures the
network isolation by means of applying tunnels to transmit customer traffic
over the provider’s external network.
This is achieved by route declarations from the customer network skipping
over the provider’s internal network by means of the protocol BGP; thereafter,
having been specially configured (using an enhanced version of MultiProtocol
BGP, MP BGP) they get to the networks of the same customer. As a result, routers
of different customers have no route information about each other, that’s why
they are not able to exchange information, i.e. the required isolation is achieved.

To combine the customer’s geographically scattered networks into a single
VPN a tunnel is used between edge routers of the provider internal network.
Advantages of such MPLS VPN tunnels is that they are installed automatically,
plus those benefits obtained at the expense of applying MPLS technology as
such: - speeded up advance over the provider network, as well as controlling
Classes of Service (CoS) for the tunnels with traffic engineering.
The MPLS VPN network is divided into two areas: customer IP networks and the
provider’s internal (backbone) MPLS network to combine the customer networks.
In the backbone MPLS network of the provider IP packets are advanced based
on local labels rather than IP addresses.
The MPLS network consists of Label Switch Routers (LSR) directing traffic
over preliminarily formed paths with label switching (Label Switching Path,
LSP) in accordance with these labels’ values. The LSR device is a peculiar
hybrid of an IP router and a switch; in this case the IP router capability
applies of identifying network topology using routing protocols and selecting
efficient traffic paths, and the switch capability applies of advancing packets
using labels and local switching tables. The LSR devices are often called for
short just routers; there is a certain reason for that – in case MPLS support
is disconnected, they are able to advance packets based on an IP address as
well.
Among the LSR devices on the provider network edge routers can be marked out
(Provider Edge Router, PE), to which customers’ sites are connected via customer
edge (CE) routers, and internal routers of the provider backbone network (Provider
Router, P). A physical channel on which a channel-level protocol such as PPP,
FR, ATM or Ethernet is run usually interconnects directly the CE and PE routers.
CE and PE communicate on the basis of standard TCP/ IP stack protocols; MPLS
support is required only for internal PE interfaces (and all P interfaces).
On the provider backbone network only edge PE routers are to be configured
for supporting virtual private networks, therefore only they are aware of the
existing VPN’s. If the network is considered from the VPN side, the provider
(P) routers do not directly interact with the customer edge (CE) routers; they
are just located along the tunnel between the input and output PE routers.
The PE routers serve as terminal points of LSP paths between the customer sites,
and the PE routers exactly attach a label to an IP packet for its transiting
via the internal network of P routers.
Let’s imagine that via a PE router an invisible boundary passes splitting
it into the zone of customer sites and the zone of the provider’s network core.
On one side there are interfaces via which the PE router interact with the
P routers; on the other – interfaces the customer sites are connected to. From
one side declarations of backbone network routes come to the PE routers, from
another side – declarations of routes in customer networks.
For each new customer site the PE router creates a separate routing table.
An advantage for the customer is that addressing of the nodes of each separate
site is carried out independently, for a variety of nodes of one site under
no circumstances will get routing information from another variety of nodes
of another site.
Restricting the area of routing information circulation to the limits of separate
VPN’s isolates address spaces of each VPN, allowing applying within its limits
both public Internet addresses and private addresses registered according to
RFC 1819.
To all addresses of the address space of one VPN a prefix is added called
Route Distinguisher (RD), which identifies this VPN as a unique one. As a result
all addresses on the PE router, relating to different VPN will necessarily
differ from each other even in case they include a coinciding part - the IPv4
address.
Routing information exchange among the sites of each separate VPN is controlled
by the protocol MP-BGP (Multiprotocol BGP).
MPLS-Based Virtual Private IP Networks (IP-VPN)
Key MPLS advantage from the user’s standpoint is ÑoS, next one in terms of importance
is the simplification of VPN access protection and procedure.
- Any data can be transmitted, since the content remains unchanged along the
entire path, except for replacing labels. Hence users are able to transmit
SNA, SPX/IPX, IP packets with illegal addresses (RFC 1918 addresses), frames,
ATM cells etc.
- As opposed to a virtual channel, the MPLS fixed path is provided as IP interface
part, therefore a buyer needs to do nothing in order to use it. An MPLS-based
VPN sold to an end user will contain a parameter describing the way to distinguish
traffic of this VPN. For example, an IP packets stream, when arriving at the
ISP’s IP interface, will be analyzed by the MPLS edge device. Those packets
meeting the VPN criterion will be directed over the MPLS path for further processing.
- ÌPLS VPN can be created to support critically important applications on
a round-the-clock basis. In this case an ISP determines the fixed path during
the validity of the user contract.
- If we represent the IP-based service interface as a big circle, than a VPN
created by means of MPLS will add a small logical tunnel toward this common
interface (a small circle inside the big one). Such a tunnel is capable of
transporting IP traffic using a private internal address. The second tunnel
is capable of moving NetWare SPX/IPX traffic, the third – voice within packets.
It should be noted that in this case a user is not limited to voice over IP
only, for MPLS doesn’t depend on the protocol. Each of the VPN tunnels is capable
of providing a unique quality of service.
- At the corporate level, temporary VPN’s should facilitate the organization
of direct connections between the ISP’s fixed MPLS paths and a corporate network.
In this case we can expect appearance of switching products for third-level
local networks, in particular devices for high-class backbones supported by
MPLS. One of MPLS functions consists in what Cisco calls virtual channels integration,
i.e. when several MPLS tunnels are integrated to create a single tunnel. Such
a structure resembles a river tributary system and extends the MPLS-based VPN
on the operator network to the interoffice network and directly up to the server
or customer. In case of such VPN extension the operator might have control
responsibility for ensuring a continuous through control over ÑoS.