P
packet
A
protocol data unit (PDU) at
Layer 3 (L3) of the
Open Systems Interconnection (OSI) Reference Model. A packet contains source and destination addresses, user data, and control information such as the length of the packet, the header checksum, and flags indicating whether the packet has been fragmented. The user data in a packet is often referred to as the payload. The actual format of a packet depends on the protocol that creates the packet.
A packet sent through a
connectionless protocol such as
User Datagram Protocol (UDP) is sometimes called a datagram.
packet switching
A data-transmission method that transmits information over one of several routes. Information is sent to the destination through the best route, determined by a routing algorithm.
A packet switched network breaks information to be transmitted into discrete packets. Related packets might not all follow the same path to their destination. Packet sequence numbers are used to reassemble the original message at the destination.
A packet-switched network is
connectionless because each packet contains its destination address and does not require a dedicated path to reach that destination. Multiple users may transmit packets over the same connection at the same time, independent of one another.
The Internet is an example of a packet-switched network.
paravirtualized
A software component that is aware that it is running in a
virtual machine (VM). For example, a paravirtualized virtual device driver runs in a VM that communicates with the underlying host OS. Typically, a paravirtualized driver is optimized to share queues, buffers, or other data items with the underlying host OS to improve throughput and reduce latency.
path computation element (PCE)
An entity (component, application, or server) that can compute a network path or route based on a network graph and constraints (see RFC 4655).
path-vector routing
A routing technique that advertises a network as a destination address and a complete path to reach that destination. Each entry in the
Routing Information Base (RIB) contains the destination network, the next router, and the path to reach the destination.
A path vector protocol guarantees loop-free paths by recording each hop the routing advertisement traverses through the network. A node can easily detect a loop by looking for its own node identifier in the path.
This technique is sometimes used in
Bellman-Ford algorithm to avoid
count-to-infinity problems.
Border Gateway Protocol (BGP) is an example of a prefix-based path-vector protocol where the
Routing Information Base (RIB) maintains the autonomous systems to traverse to reach a destination.
peer
Immediately adjacent device with which a protocol relationship has been established. Also called neighbor.
penultimate hop popping (PHP)
A technique where the outermost label of an
Multi-Protocol Label Switching (MPLS) packet is removed by a
label switch router (LSR) before the packet is passed to an adjacent
label edge router (LER).
physical layer
ping (packet internet groper)
A command used to test network connectivity by transmitting an
Internet Control Message Protocol (ICMP) diagnostic packet to a specific node on the network, forcing the node to acknowledge that the packet reached the correct destination. If the node responds, the link is operating; if not, something is wrong.
The word ping is often used as a verb, as in “ping that workstation to see if it is alive.”
policing
Applying rate limits on bandwidth and burst size for traffic for a particular interface.
policy-based routing (PBR)
Classifying packets to determine their forwarding path within a device. PBR is used to redirect traffic for analysis. Also called filter-based forwarding (FBF).
port
The point at which a communications circuit terminates on a network. A port can be logical, physical or both. Examples include:
• The physical interface between a device and a communications circuit, usually identified by a number or name.
• The logical interface between a TCP/IP applications and a communications facility which use well-known port numbers such as FTP: 20, HTTP: 80, and NFS: 2049.
• The logical interface between a process and a communications facility that allows more than one logical port to be associated with one physical port. For example,
Ethernet uses multiple MAC addresses to distinguish between separate logical channels connecting two ports on the same physical transport network interface.
Precision Time Protocol (PTP)
A protocol that synchronizes clocks throughout a computer network. On a LAN, PTP achieves clock accuracy in the sub-microsecond range, making it suitable for measurement and control systems. The time synchronization is achieved through packets that are transmitted and received in a session between a master clock and a slave clock. Defined by IEEE
1588v2.
Priority-based Flow Control (PFC)
A flow control mechanism that can be set independently for each frame priority on full-duplex links. Defined by IEEE
802.1Qbb.
private VLAN (PVLAN)
A switch with ports that cannot communicate with each other, but can access other networks. A PVLAN has at least one private port and a trunk port. All traffic received on a private port is forwarded out the trunk port. All traffic received on a trunk port is handled as normal switch traffic. No traffic communication occurs between the private ports.
protocol
A set of rules that end points in a network connection must follow when they communicate. A protocol includes data representation, data item ordering, message formats, message and response sequencing rules, block data transmission conventions, and timing requirements.
The
Open Systems Interconnection (OSI) Reference Model and
Transmission Control Protocol/Internet Protocol (TCP/IP) are both used as a model for many protocols. There are one or more protocols at each layer in the models that both ends of the connection must recognize and observe.
protocol data unit (PDU)
A unit of data transmitted as a composite by a protocol.
In the
Open Systems Interconnection (OSI) Reference Model, the actual name used for a PDU depends on the layer:
See also
bridge protocol data unit (BPDU). Sometimes called datagram.
Protocol Independent Multicast (PIM)
A method to determine the best paths for distributing a multicast transmission. PIM uses unicast routing tables (such as those used by
Open Shortest Path First (OSPF) and
Border Gateway Protocol (BGP)) and static routes to perform multicasting. Each host must be registered using IGMP to receive the transmission.
PIM has these variations:
• PIM dense mode (PIM-DM: RFC 3973) uses a push model. PIM-DM initially floods multicast traffic throughout the network. Routers that have no downstream neighbors prune back the unwanted traffic. This process repeats periodically.
• PIM sparse mode (PIM-SM: RFC 4601) uses a pull model. PIM-SM uses a
shortest path tree (SPT) where sources forward multicast packets to a designated router which unicasts the packets to an assigned rendezvous point router, which then forwards the packets to members of multicast groups.
• PIM source-specific multicast (PIM-SSM: RFC 3569) uses PIM-SM functionality to create a SPT between the client and the source without using a rendezvous point.
• Bidirectional PIM (Bidir-PIM: RFC 5015) uses PIM-SM functionality to route traffic only along a bidirectional SPT that is rooted at the rendezvous point for a group.
protocol stack
The layers of software used in network communications.
Provider Backbone Bridge-Traffic Engineering (PBB-TE)
An extension to
Provider Backbone Bridging (PBB) that removes features such as flooding, dynamically created forwarding tables, and spanning tree protocols. PBB-TE also covers
Connectivity Fault Management (CFM) and
Ethernet Linear Protection Switching (ELPS).
In PBB-TE, a network administrator configures the forwarding tables in the backbone switches with static routes to ensure that frames take predetermined paths within the network. Frames with destination MAC addresses not in a forwarding table are dropped. Broadcast frames are not supported and are also dropped by backbone switches.
Defined in IEEE
802.1Qay.
Provider Backbone Bridging (PBB)
A technique to create
Ethernet backbones for service access networks. Defined in IEEE 802.1ah, PBB extends
Provider Bridging (PB) defined in 802.1ad in these ways:
• The 802.1ah header adds an
I-SID (Service Instance Identifier) which is a label that maps to a customer VLAN identifier. An I-SID virtualizes VLANs across a network. VLANs are mapped into I-SIDs by configuring only the edge of the network at a
backbone edge bridge (BEB). This makes the maximum number of service instances 16 million.
• The 802.1ah header encapsulates backbone source and destination MAC addresses (
B-MAC) along with the customer source and destination MAC addresses (
C-MAC). The B-MAC contains MAC addresses of the service provider's PBB edge switches. The 802.1ah format is sometimes called “MAC-in-MAC” because of this MAC address encapsulation. The encapsulation of customer MAC addresses in backbone MAC addresses means that the backbone does not need to learn customer MAC addresses. Customer MAC addresses are learned at BEB ports only.
Provider Bridging (PB)
A technique that enables a service provider to use the architecture and protocols of 802.Q to offer the equivalent of separate LANs, bridged LANs, or VLANs to multiple customers. Provider bridging requires no active cooperation between customers and requires minimal cooperation between an individual customer and the service provider.
When VLANs were originally defined in 802.1Q, the number of unique VLAN identifiers was limited to 4096. In large provider networks, each subscriber needs a separate address, so this limit could prevent a provider from having more than 4096 subscribers.
To overcome this limit, 802.1ad inserts an additional VLAN tag into a single 802.1Q
Ethernet frame. Frames passing through the provider network are doubly tagged with:
With two VLAN identifiers in combination for each provider-customer pair, it is possible to define up to 16,777,216 VLANs.
The frame format for 802.1ad is also called Q-in-Q, double tagged, stacked VLANs, or VLAN stacking.
provider edge (PE)
A device at the edge of an enterprise or service provider core network. A PE offers an initial, first level of network traffic aggregation for many
customer edge (CE) devices.
pseudowire (PW)
An emulation of a point-to-point connection over a packet-switching network. A pseudowire is a way to transport legacy services such as TDM over a packet-switched network:
• Structure-aware TDM circuit emulation service over packet-switched network (CESoPSN)
• Structure-agnostic TDM over packet (SAToP)
A pseudowire that both originates and terminates on the edge of a single packet-switched network (autonomous system or carrier network) is called a single-segment pseudowire (SS-PW). A pseudowire that extends through multiple autonomous systems or carrier networks is called a multi-segment pseudowire (MS-PW).