8-Port E1/T1 TDM over IP "Pseudowire" chassis

Product Overview

The PacketBand-TDM8 is a highly-featured 2U chassis supporting 8 E1 and/or T1 TDM over IP Pseudowire circuits delivered across different types of packet networks.

• Highly accurate and stable clock recovery processes maintains end-to-end synchronisation – the absolute foundation of making this TDM over IP technology work.
• G.823 Synchronisation Mask levels of accuracy are achievable
• Dual load-sharing hot-swappable AC/DC PSUs for resilience
• Inter-works with smaller PacketBand units and the 16/32 port chassis
• Support for:
  • Clear-channel and channelised E1 T1 circuits (frames and unframed)
  • G.704 Grooming
  • External clock sources
  • RSTP
  • LACP
  • UTP and fibre (SFPs)
  • “Jumbo” packets
• Enhanced Multicast clocking mode for larger networks
• “Tuneable” algorithms to match network characteristics
• Various clocking options to suit different network types
• Compact 2U chassis mounts 300mm deep
• All interfaces front presenting
• Robust, reliable and professional quality
• Excellent management, network visibility, diagnostic and statistics
• PacketBand excels in delivering stable and accurately clocked E1 and T1 “leased lines” over IP and Ethernet networks with varying size and performance.

Connectivity Overview

• A single PacketBand-TDM8 compact resilient rack-mountable chassis supporting up to 8 E1/T1 over IP circuits in G.704 or G.703 mode. These TDM over IP/Ethernet Pseudowire circuits can be connected to other chassis and/or to stand-alone smaller PacketBands such as the various single and 4 port units.
• E1/T1 circuits can be individually configured with no constraints. Reconfiguring one E1 does not impact upon any other port or services running across it.
• The system also supports “grooming” whereby a single E1 can support multiple remote low-speed fractional G.704 devices. Grooming also means remote PacketBand-TDM-VXs, which deliver V.35 and V.11/X.21 services, can also be presented on a central E1/T1.
• PacketBand-TDM8 supports up to 8 links as standard but can, when running multiple links into a single channelised port (grooming) support up to 128.

Cards and Interfaces

• All cards are front insertable The chassis has all I/O front-facing and supports:

One or two PSUs

• AC and DC options are available
• “Hot-Swappable” according to the ATCA PICMG 3.0 specification
• One PSU supports a full chassis

Ethernet Card

• Standard Card1 x SFP cage and 2 x RJ45 UTP
• Optional Card 2 x SFP cages and 2 x RJ45 UTP

Terminal Card

• Standard Card1 x RJ12 control port and 1 x RJ12 alarm relay
• Optional Card 1 1 x RJ12 control port, 1 x RJ12 alarm relay and 2 x RJ45 external clock inputs 2.048Mbps HDB3
• Optional Card 2 1 x RJ12 control port and 1 x RJ12 alarm relay with 2 x RJ45 external clock inputs 2.048Mbps HDB3 encoded and 2 x 50 Ohms BNC 10MHz external clock sources

E1/T1 over Ethernet or IP Card

• Each card supports 16 user-switchable E1 T1 interfaces
• 8 port E1/T1 cards are available with Telecoms Approvals for direct connection to carrier leased lines
• The chassis supports two cards in any combination
• RJ45 connectors
• 120 Ohm and user-switchable to 75 Ohm (I/O conversion cable required)
• Full E1/T1 or full/fractional G.704

Clocking

The whole PacketBand is designed and optimised for very accurate clock recovery across a wide range of different network types with different performance characteristics.

• Clock recovery, accuracy and stability is key to many TDM Pseudowire applications and the clock recovery performance must be maintained when migrating to a packet network solution.
• Leased lines which are generally referenced to the G.823 Synchronous Interface mask and PacketBand can meet and exceed the G.823 requirements, providing accurate and stable clocks to the same quality as traditional leased lines.
• The clock recovery methods use a variety of mechanisms. These include sophisticated algorithms which allow users to “tune” the performance to match the network characteristics.

Clock Algorithms

The TDM8 supports three advanced TDM over IP algorithms as standard. These are designed and optimised for different network types;

• one for high quality managed networks with low jitter (PDV) and packet loss;
• one for networks with lower performance characteristics;
• and one for applications where stability is key but the G.823 mask is not specifically targeted.
• Within these options are two further settings allowing tighter optimisation if required.

This means PacketBand can be configured to extract the best possible service for any given network.

Clock stability

Clock frequency stability performance can exceed AT&T TR-62411, T1.403, G.824 and G.823 on quality networks for both the Traffic and the much more demanding Synchronisation requirements.

• This is a vital strength as many applications will either require very accurate synchronisation to run correctly (for example mobile backhaul), or there could be multiple clock sources within the network which all require aligning accurately to ensure error-free and reliable services.
• Failure to provide excellent clocks will results in slips and hits – data loss
• PacketBand can deliver reliable TDM over IP services that clock as well as traditionally delivered leased lines.
• Should the recovered clock’s stability move outside of configured historic thresholds and averages, PacketBand can alarm to the DbManager giving operators immediate information on which to investigate the causes (usually network changes/issues)

Clock Sources and Clocking

Although easy to use and configure, PacketBand offers customers a number of solutions for different clocking scenarios as briefly described below.

The PacketBand’s advantage is not only very accurate clocking and reliability, but also the flexibility to work in different modes to suit a variety of situations and applications.

• TDM port
  Clock can be extracted from any of the attached E1 T1 user ports. This would normally be used at the “Master” end of a circuit with the remote “Slave” unit recovering clock across the packet network and clocking its attached equipment as described in “Adaptive” below.
• Adaptive
  Using sophisticated recovery algorithms, clock can be derived from any of the incoming packet streams from a remote “Master” TDM8 or any other PacketBand. The remote “Master” would generally have a high quality clock source presented on one of its interfaces. This allows the “Slave” TDM8 to adaptively recover this clock very accurately across the packet network, in effect synchronising both devices to a common clock.
• External
  The Terminal Card optionally provides 2 x RJ45 connectors which accept either a 2.048 HDB3-encoded clock. The card is also available with two BNC connectors with 10MHz clock input ports.
• Multicast
  Larger networks with many “hops” pose a particular challenge to recovering the clocks down to the very low parts per billion. This is due to the compounded tiny amounts of high frequency jitter added by each switch.
  To circumvent this situation, Patapsco provide a unique “Multicast” system where the clock recovery traffic is separated from the user traffic. This means the clocks can be sourced more locally, perhaps on a regional basis, and transported in a low-speed high-priority efficient Multicast service.
  A significant additional advantage is that the user traffic, the bulk of the capacity, is no longer transporting clock information; it can therefore be assigned a lower priority than would be usual, making network provisioning easier and more efficient.
  The PacketBand-TDM8 can act as either a generator of the Multicast clock for use by other remote devices, or join a Multicast feed using IGMP and thereby source its clock.
  More than one Multicast can be configured to provide enhanced resilience.
  See also the PacketBand-TDM-3MC and the TDM-1MC Application Overviews which have further information on the Multicast feature.
• Internal
  In the unlikely event of a PacketBand losing all external clock sources, including local E1s, it can fall back to its internal oscillator. These are very accurate with good stability characteristics and various options are available – see the data sheets.
• Clock Hold-Over
  Problems in the packet network, for example a sudden increase in jitter or an interruption in packet delivery, can cause an unwanted movement in the recovered clock. To avoid this PacketBand enters a “hold-over” state, maintaining the recovered clock at the last value prior to the problem.
• Loop-Timing
  In instances where both ends of the circuit have a good quality common clock source (for example ISDN clocks into PABXs) both PacketBands can be set to clock from this local source.

Oscillators

If the PacketBand is recovering clock, the quality of the oscillator used as part of the recovery process is very important.

• The receive PacketBand running “Adaptive” clocking uses a variety of different information and many calculations to ascertain how to modify its on-board oscillator’s output to match the clock of the remote or “Master” end.
• It can be seen that the more stable the on-board oscillator, the more stable the recovered clock.
• TDM8 is fitted as standard with a temperature stabilised stratum 3 Enhanced TCXO oscillator which is ideal for all but the very exacting applications. Details are available in the data sheets.

PDV (Packet Delay Variation or jitter)

• The PacketBand-TDM8 has the ability to support large amounts of network jitter. The absolute maximum buffering available which depends on a number of variables including the number of channels, circuit emulation protocol and frames per packet (Patapsco have a spread sheet available).
• As an example, 8 E1s running UDP/ RTP and VLANs with IPv4 (a protocol combination with a high packet overhead) and 8 frames per packet, gives a maximum configurable jitter of 64msec (+/-32msecs). An optional upgrade to 512msecs is available.
• The jitter buffer can be set in msec granularity and adjusted manually or automatically whilst the circuit is in place, overcoming ‘skew’ at start up time caused by the first packet in the buffer being slower or faster than average.
• The buffer re-orders out of sequence packets and if packets are lost by the network the data to be transmitted to the attached device by PacketBand is configurable.
• Statistics are available to provide information on usage (see below).

Ethernet and the Packet Switch

• Support for 10/100/1GE.
• Support for “Jumbo-packets” up to 10,240 bytes in size.
• Single MAC and IP address, Default Gateway and SubNet Mask, support for DHCP. Disordered packets are reordered automatically.
• IPv4 (IPv6 Q2 ’09).

Prioritisation

• TDM packets can be assigned IP Diffserv ( DSCP ) or ToS and 802.1p CoS values.
• PacketBand supports full 802.1q tagging and the associated 802.1p CoS prioritisation levels.
• All egress packets including TDM links can be prioritised across four output queues using CoS (802.1p) or Diffserv/ToS values.

VLAN Handling

• PacketBand’s latest-generation on-board packet switch offers advanced 802.1Q VLAN facilities such as multiple TAGing, TAG insertion/removal, port routing based on default TAG or a Global TAG table. Egress packets can be TAGed, have multiple TAGs or be stripped of TAGs according to configuration.
• PacketBand also supports Provider Mode whereby customer packets are TAGed for transport across the network with the TAGs being removed before passing back to the customer at the far end.

Link Aggregation Control Protocol (LACP)

• This powerful feature enables two or more Ethernet ports to be connected between PacketBand and the network switch and for the links to be aggregated together as defined in LACP IEEE 802.3-
• 2005. This aggregation makes it appear as if the multiple links are acting as a single high capacity circuit. Furthermore, it adds an important level of redundancy with automatic rerouting. See the
• TDM-4E and TDM-1MC Application Overviews for further information.

Rapid Spanning Tree Protocol (RSTP)

• RSTP (defined in IEEE 802.1D-2004) identifies the means to build an Ethernet network which contains physical loops between bridges. This optional facility enables PacketBands to be connected to more than one network switch via different circuits and to provide automatic fall-back in the case of a link failure. This is covered in more detail in the PacketBand-TDM-4E and TDM-3MC Application Overview.

Rate Limiting

• Packet Ports can individually have the traffic capacity restricted in various ways, even though the access is 10/100/1GE. This is particularly useful on Ethernet ports connected to users LANs where the main link to the network could be “swamped” by data from attached devices.

Overheads

• In order to transport TDM data over the packet network there is some overhead caused by encapsulating the data inside the packet protocol.

The Protocol

• PacketBand supports a number of different packet network protocols. The user’s choice of a particular network will affect the overall size of packet headers.

Size of Packet

• PacketBand supports a configurable packet size per Logical Link. There is a trade off between transmitting small packets at a fast rate (low latency, larger overhead due to protocol headers) and transmitting larger packets at a slower rate (bigger latency, smaller overhead).
• Typical overheads are in the 5% to 10% range. Patapsco have a spreadsheet available which identifies overheads based on a number of different parameters. Contact Patapsco or your supplier.

Latency

The total end-to-end latency experienced between two devices connected to TDM over IP PacketBands is made up of four elements:

• Processing Delay
  The latency or processing delay through each PacketBand is optimised to be as low as possible. Typical processing delay is less than 1msec.
• Transmit Delay
  This is the time necessary to wait for sufficient incoming data to arrive from the attached TDM device so a packet of the configured size can be built and transmitted over the network. This is typically around 1msec.
• Jitter
  Packet network networks differ in how consistently packets pass though them; some packets take more or less time than the average. PacketBand provides a synchronous clocked circuit to the attached devices and therefore has to have data ready and available for the relevant clock pulse. PacketBand buffers the fast packets to ensure the slow ones can arrive in time to be used. The amount of buffering is user-configurable and will depend upon the performance of the network. Note that this buffering is only required on the PacketBand receive data path and the amount of buffering needed (which equates to latency) is a result of the network, not PacketBand.
• Transit Delay
  All IP networks have different average transit delays. These vary depending upon a large number of criteria, including the number of “hops” and whether satellites are involved. Typically, domestic links are very fast, inter-continental around 60msec and a satellite can add 250msecs. Please consult your network supplier.
• Summary:
  Between any pair of PacketBands on a terrestrial network, the most significant element contributing to latency is size of the Jitter Buffer (which is user configurable). This is directly dependant on the performance of the network and outside the control of PacketBand.

CESoP Modes

• PacketBand supports both “Structure Aware” and “Structure Agnostic” TDM over IP modes.
• Complies with ITU-T recommendation Y.1413, IETF PWE3 draft standards CESoPSN, SAToP and CES draft IAs from MEF and MFA .

Management Overview

• PacketBand can be locally or remotely configured using Patapsco’s easy-to-use high functionality DbManager GUI software.
• DbLite is supplied free with each unit.
• Optionally available are different versions to support requirements for larger or more integrated networks. It is sophisticated but simple to use via an intuitive Graphical User Interface (GUI) which controls, configures and monitors individual Patapsco units and complete networks.
• Used by various organisations with different network sizes - up to and including carriers - versions of DbManager deliver a network-wide view of all PacketBands and links via a 4-layer “tree-structured” GUI. The status of all PacketBands and links are easy to identify with Alarms being colour coded and passed up the tree. Separate windows provide Event and Alarm information with the ability for operators to add comments etc. Different access levels and passwords provide operators with appropriate capabilities within the program.
• Demonstration software is available which illustrates both the DbManager and the PacketBand features. Please ask for information.

Configuration Changes

• Configuration changes on PacketBand are via the DbManager. All configurations can be stored on DbManager. Installations require little or no expertise in the field as most configurations (other than IP
• address) can be performed remotely and DHCP is supported.

Management Tools

• A wide number of statistics are available for the E1/T1 circuits and Ethernet ports.

Alarms/Events

• All Alarms can be reported back to the DbManager and presented in a dedicated window with descriptor.
• Events and Alarms are held locally in the TDM8 in NV memory for access via DbManager.

Graphs

DbManager and PacketBand provide several network monitoring tools available with graphical out-put:

• Maximum, minimum and average jitter buffer usage over time. This screen shows how the network and link are performing. Particularly useful information includes Lost and Late packets and jitter buffer usage figures.
• Average network jitter over time

A graphical representation showing Minimum, Maximum and Average buffer usage which provides important information on network performance.

• Recovered clock movement over time

This graph shows frequency stability and the status of the acquiring clock.

These are invaluable tools for optimising PacketBand and for acquiring information on network performance.

Together these statistics, which are updated for each Logical Link every 10 seconds, gives accurate and invaluable information on the performance of the network and are vital when commissioning. The information also identifies whether the jitter buffer settings are correct and if the jitter buffer can be reduced to remove any unnecessary latency.

Software/Firmware Updates

• New software can be loaded remotely via the DbManager to PacketBand-TDM8. Multiple remote devices can be uploaded simultaneously. New software is loaded to the off-line sector of Flash and is confirmed via a CRC . Users can switch to the new software at any time.

Further details and specifications are available in the PDFs at the top of this page and when registered.

See also the single port, 4 port and 16/32 port PacketBand versions.