Tuesday, April 24, 2007

MPLS… Multi-Protocol Label Switching! Not to be Confused with Minneapolis

MPLS… Multi-Protocol Label Switching! Not to be Confused with Minneapolis
MPLS for the speed and efficiency you need.

Multi-Protocol Label Switch, or MPLS, is a standards-approved technology for speeding up network traffic flow and making it easier to manage. It acts the same way as VPN tunneling, in that it encapsulates protocols and data before sending them out in their own tunnels. Sometimes, MPLS is referred to as “Layer 2.5” because it emulates properties of Layer 2 (data link layer i.e. Ethernet) and Layer 3 (networking layer i.e. Internet Protocol, or IP).

MPLS is just a newer way of doing the same work that frame-relay and Asynchronous Transfer Mode (ATM) do. MPLS is becoming more popular because it is better suited for current and future technology needs. In particular, MPLS skips the cell switching and signaling-protocol of ATM. ATM breaks up data into encrypted fixed-sized cells to send out between two end points. MPLS recognizes that small ATM cells are not needed in the core of modern networks, since modern optical networks are so fast (at 10 Gbit/s and well beyond) that even full-length 1500-byte packets don’t suffer any real noticeable real-time delays. Thus, because of the increased bandwidth available, traffic engineering and out-of-band control, which frame relay and ATM became popular for, is still maintained.

MPLS was originally called “Tag Switching” by its developers from Cisco Systems, Inc., and was renamed "Label Switching" when it was handed over to the IETF for open standardization. It was developed to make way for the creation of simple high-speed switches, since it was impossible to forward IP packets in its entirety through hardware for a long period of telecommunications. Recently, advances in VLSI (Very-large-scale integration) have made the hardware possible for such duties, but the systemic advantages of MPLS, such as the ability to support multiple service models, do traffic management, etc., remain.

Instead of encrypting packets, MPLS adds a 32-bit tag to packet headers. MPLS works by pre-pending packets with an MPLS header, containing one or more 'labels'. The packets are called label stacks. Each stack contains four fields: a 20-bit value, a 3-bit field for QoS priority, a 1-bit ‘bottom of stack’ flag (meaning if set, the current label is the last of the stack), and an 8-bit TTYL (time to live) field.

When packets enter an MPLS-based network, Label Edge Routers (LERs) give them a label (identifier). A tag router, the ingress router, will examine the desired destination address, and creates a tag that chooses a virtual circuit or label switch path for that packet. From there on out, tag switches will only look at the tags to determine how to forward the packet. Routers that are performing routing based only on Label Switching are called Label Switch Routers (LSR). There may be multiple routes available for each label switch path so that the tag switches can manage outages, congestion and differentiated services. At the egress point, the exit router, the MPLS tag is removed before sending the packets on their way.

Processing small tags is faster than having to deal with larger headers at each router, creating choke points for data flow. Another advantage of MPLS networks is that it can be designed to provide more bandwidth, or shorter latency paths for voice packets in VoIP telephone systems. Video packets are extremely heavy on bandwidth so it would be best not to funnel them into paths where computers are backing up large databases. Through MPLS networks, voice and video can have the bandwidth needed to maintain quality of service.

When properly designed, deployed, and maintained, MPLS in a private network is a powerful tool to increase business efficiency while reducing costs and improving performance. MPLS networks are now spreading to include access networks.

To set up MPLS or other networking services for your business, check out T1 Stop Shop.