The present invention is related to an Ethernet based access network, wherein residential subscribers are coupled via an optional remote unit and an aggregation node and via Ethernet switches to an edge node.
The network access nodes in an Ethernet based access network, i.e. remote unit and aggregation node, connect the residential users to the network of the Network Access Provider (NAP). The remote units are the first gate from the user towards the NAP e.g. VDSL terminators. As a remote unit only connects a few users, it needs to be very cost effective. The aggregation node aggregates a number of remote units and bears more functionality because their cost is shared among more subscribers. The network edge nodes connect the Layer 2 Ethernet network of the NAP to the Layer 2 network of the Network Service Providers NSP, or connects the Layer 2 network via an IP point to the IP-network of the NAP. Ethernet bridges are used to interconnect all these network devices.
Ethernet access networks in a bridged configuration are very attractive for this access architecture because of their management-less self-learning capabilities, flexibility and efficient support for multicast services. A major problem they encounter is that one can never derive from the packet which subscriber line it belongs to. This poses severe problems with respect to security, subscriber management and, indirectly, to cost.
In order to solve the above-mentioned problem bridged Ethernet networks are deployed in a Cross-Connect configuration. On top of the normal Ethernet a pipe model is constructed. All traffic from one subscriber is put inside a pipe. Pipe architectures based on Virtual LANs (VLANs) and MPLSoE (MultiProtocol Label Switching over Ethernet) have been proposed. Standardisation efforts on both types of solutions are ongoing.
Cross-connect networks are very well suited for business deployment but are less flexible, need more management and have problems with multicast traffic. All this is due to the pipe model, which is on top. Therefore they are not suited for residential access.
The present invention aims to disclose a bridged Ethernet network architecture for residential access that overcomes the drawbacks of the state of the art solutions.
The invention relates to a method to forward an Ethernet packet in an Ethernet based access network, comprising the step of receiving said Ethernet packet at a network access node being part of said access network. It further comprises the step of transforming at the network access node a first Ethernet MAC address contained in the Ethernet packet into a second Ethernet MAC address, whereby either the first or the second Ethernet MAC address comprises subscriber or subscriber line identification. Subsequently the step of continuing the communication protocol is performed.
The Ethernet MAC address comprising subscriber or subscriber line identification has preferably a format comprising:
Advantageously the subscriber or subscriber line identification allows the retrieval of information regarding the origin of the Ethernet packet.
In an advantageous embodiment the first Ethernet MAC address is an Ethernet source MAC address of an upstream Ethernet frame to be transformed into the second Ethernet MAC address comprising subscriber or subscriber line identification.
In an alternative embodiment the first Ethernet MAC address comprises subscriber or subscriber line identification to be transformed into a second Ethernet MAC address, whereby the second Ethernet MAC address is an Ethernet unicast destination MAC address of a downstream Ethernet frame.
Optionally the Ethernet MAC address comprising subscriber or subscriber line identification also contains a netmask comprising the netmask length. The netmask splits the Ethernet MAC address in a network related part and a device related part.
Typically the network access node is a remote unit or an aggregation node.
In yet another embodiment the method disclosed in the invention comprises—after the step of receiving the Ethernet packet at a network access node being part of the access network—the further step of transforming a first Ethernet MAC address contained in the Ethernet packet into a second Ethernet MAC address, whereby the first Ethernet MAC address is a broadcast Ethernet destination MAC address of an upstream Ethernet frame and the second Ethernet MAC address is a multicast-group address.
In this embodiment part of the Ethernet packet advantageously is 1-to-1 mapped with the said multicast address. The part of the Ethernet packet to be 1-to-1 mapped with the multicast address preferably comprises protocol information.
In an alternative embodiment the step of transforming is replaced by step of encapsulating the Ethernet packet with the first Ethernet MAC address in a new packet comprising the second Ethernet MAC address. The step of transforming can also be replaced by the step of applying a MAC address translation, wherein the first Ethernet MAC address is overwritten with the second Ethernet MAC address.
As a second object the invention relates to an electrical circuit device arranged to carry out the method as in any of the previous claims.
As a third object the invention relates to a network access node arranged for receiving an Ethernet packet in an Ethernet based access network, wherein the access node is further arranged for transforming a first Ethernet MAC address contained in the Ethernet packet into a second Ethernet MAC address, whereby either the first or the second Ethernet MAC address comprises subscriber or subscriber line identification.
The invention also relates to a 48 bit Ethernet MAC address format, with the following parts:
In a preferred embodiment the metro edge identification starts with a zero bit, indicating an access node. To indicate an edge node the metro edge identification starts with a one bit.
In yet another object the invention relates to an Ethernet packet comprising a MAC address in the following format:
The optimal solution to the state-of-the-art problems would be a bridged Ethernet network without pipes, wherein the Network Access Provider has direct associations between a packet and a subscriber line.
Key to the disclosed invention is therefore the introduction at the network access nodes, i.e. remote units and aggregation nodes, of a new predefined Ethernet MAC address structure comprising an explicit subscriber line identification. The goal to be achieved is to allow the retrieval of all useful information related to a packet from just this MAC address. Packets coming in upstream will undergo a source MAC address transformation, whereas the destination MAC address will not change. In downstream the destination MAC address will undergo the inverse transformation. This enables user-segregation and information retrieval inside every packet, yielding a better security and rendering more cost-effective most user-management deeper in the network.
Wherever Ethernet address translation is mentioned in this description, MAC-in-MAC can also be used: if e.g. a packet is coming from the user, either the old MAC address can be overwritten with the new address structure (MAC address translation), or the packet with the old address can be encapsulated in a new packet with the same destination address but with a new source address.
In a first embodiment of the invention an Ethernet MAC Address Translation (EMAT) is performed in the Network Access node whereby the user's unicast MAC addresses are transformed into a special form of the Ethernet unicast MAC address. A newly defined 48-bit address, split into different parts, will be discussed below. So the source address of the upstream frames are transformed in order to incorporate subscriber line information. With respect to the destination address it must be noted unicast destination addresses are not transformed in upstream. In another embodiment downstream packets are considered. There it is the unicast destination address that undergoes the inverse address transformation, i.e. an Ethernet address in said special form is transformed back into its original format. The source address remains as is.
Broadcast destination addresses however also cause some problems. Therefore these addresses may be transformed too. This constitutes a third embodiment of the invention. In order to reduce broadcast traffic, the invention proposes to translate the upstream destination addresses, which are equal to the broadcast address, to a multicast-group address. These addresses are only broadcast to members, which have registered them to receive it. Note that in downstream nothing happens with destination broadcast addresses.
A Broadcast Ethernet MAC Addresses translation to Ethernet multicast-group addresses is performed at the Network Access node, whereby a control of broadcast storms is reached. Broadcast storms are a potential threat to bridged Ethernet networks because the packets will be duplicated inside every Ethernet bridge on all interfaces, which creates overload inside the aggregation network and potentially a Denial of Service (DoS) attack. By applying a 1-to-1 mapping between a protocol identifier, present in the upstream packet, e.g. Ethernet-Type (or higher layer IP/TCP information) fields, and a multicast-group address, edge points will be able to advertise their interest in receiving broadcast messages of a certain protocol type, via the known Ethernet GMRP protocol. This reduces the broadcasts to only those devices that need them. If no edge device has registered its interest, the multicast packet will be discarded in the first Metro switch.
Optionally, an IP netmask concept is used by the Ethernet switches in order to minimise the learning tables and to reduce drastically the flooding problems. The netmask splits the Ethernet MAC address in a network related part and a device related part. Due to the fact that all upstream packets carry a transformed source address which incorporates information on the origin of the packet (e.g. Access Node, remote Unit, Line, . . . information), this transformed Ethernet MAC address will have a hierarchical structure (which was not the case for a globally unique address). Therefore, all upstream packets coming from a single access node will carry a common field inside the Ethernet address, which will be called the network part of the address. In normal operation Ethernet switches inside the aggregation network will learn all MAC addresses of all subscriber devices. Due to the fact that potentially millions of devices are attached to this network, all switches will need to learn millions of addresses in order to know to which interface packets needs to be forwarded. As all packets of a certain access node with the same common network part will always be forwarded via the same path, only 1 entry could be enough to forward the frames. This entry then will only learn and forward based on the common part of the MAC address. This optional functionality will need a small change inside normal Ethernet switches, but will reduce memory (and accordingly cost) for the switching tables. Additionally, this mechanism will make the learning of the aggregation network converge much faster (reduced flooding) because only one frame from a certain access node needs to be learned, instead of one frame for each attached Ethernet device.
Generally speaking there are 2 types of devices inside the access network:
The general idea is that every source address becomes a trusted MAC address and by just looking at this trusted new addresses one knows where a packet is coming from/going to. In order to quickly detect whether a packet is going upstream (from the access nodes towards the edge nodes) or downstream (from the edge nodes towards the access nodes), a separate range of source MAC addresses is specified uniquely for the access nodes and uniquely for the edge nodes. When this split is made, every intermediate switch will be able to detect by looking at both the source and the destination address, what type of packet it is: e.g. every packet coming from a customer receives a MAC address from a first range of addresses and every edge node is configured with a MAC address from a second range. If an intermediate switch looks at both the source and destination address (SA and DA, respectively) it can conclude the following:
Packets coming from a user device can e.g. have their MAC address translated into an address with fixed two bit part bit48=0 (unicast) and bit47=1 (Locally managed). For edge device (Broadband RAS) with, as already mentioned, always a globally unique address (bit48=0; bit47=0), it is possible to also use the bit48=0 and bit 47=1 fixed part and to configure this address manually in the trusted IP edge devices. Note that for the edge nodes the parts of remote unit, subscriber line and device information have no meaning. So for edge nodes there is no real structure, the address should just be easy to distinguish from the translated addresses. If all subscriber addresses are transformed, all addresses can be used except those already used for the transformation.
Two kinds of new Metro edge addresses are therefore defined:
It will be clear to a person skilled in the art that the methods of the invention as described above, can be implemented in an electrical circuit.
The advantages the invention offers are manifold. The new architecture creates the possibility of using all advantages of a normal bridged network, but makes it possible to deploy these networks for residential access by improving the security, privacy, scalability, stability, manageability and cost of these networks.
Security & Privacy are improved:
Number | Date | Country | Kind |
---|---|---|---|
03291558 | Jun 2003 | EP | regional |
Number | Name | Date | Kind |
---|---|---|---|
6560610 | Eatherton et al. | May 2003 | B1 |
6928485 | Krishnamurthy et al. | Aug 2005 | B1 |
7099295 | Doyle et al. | Aug 2006 | B1 |
20020019933 | Friedman et al. | Feb 2002 | A1 |
20020024964 | Baum et al. | Feb 2002 | A1 |
20030172307 | Henry et al. | Sep 2003 | A1 |
20030236916 | Adcox et al. | Dec 2003 | A1 |
20040004967 | Nakatsugawa et al. | Jan 2004 | A1 |
20040202199 | Fischer et al. | Oct 2004 | A1 |
20040258003 | Kokot et al. | Dec 2004 | A1 |
Number | Date | Country |
---|---|---|
WO 9836608 | Aug 1998 | WO |
WO9836608 | Aug 1998 | WO |
Number | Date | Country | |
---|---|---|---|
20040264458 A1 | Dec 2004 | US |