turboIP-G2 Performance Enhancement Proxy

With the convergence of voice, data and video over satellite becoming more common; networks becoming more fragmented and spread over larger geographical areas; and the need to shift large amounts of data from place to place growing rapidly; some organizations are encountering performance limitations. At the forefront of the challenge is user experience coupled against the cost of running and maintaining the link. Users need more access to remote data, and if user experience degrades, users are less likely to use remote services. Historically, administrators have no choice but to throw more money at the problem, either through more bandwidth or very expensive, high-end technology that provides value over long payback periods.

The Comtech EF Data turboIP-G2 Performance Enhancement Proxy is designed to tackle these intricate wide area network (WAN) complications. The turboIP-G2 is designed with the most effective WAN optimization features, targeted specifically for the unique challenges of satellite communication. Most notably, acceleration and caching are used to improve network responsiveness, keeping user experience optimized. In addition, advanced compression techniques are used to minimize data traversing the satellite link, allowing the reduction of monthly WAN bills or avoiding the purchase of additional bandwidth. Finally, the turboIP-G2 is priced well below other vendors’ offerings because the turboIP-G2 concentrates on the features that provide the biggest return on investment, and is not riddled with over designed and complicated-to-configure features. Simple, to the point, and priced to provide high value for the money invested, the moment it is turned on.

Transparent TCP Acceleration

Typical satellite links exhibit both high latency and bit error rates (impaired links), which can be challenging for the transmission of TCP. With this connection-oriented protocol, a number of factors contribute to its performance degradation over impaired links, including:

• The time required for an acknowledgement can severely limit the ramp up in transmission rate
• Sender’s small window size reduces throughput
• Delay that is interpreted as network congestion versus propagation causes reduced transmission rates
• Packet loss that is interpreted as network congestion versus corruption causes reduced transmission rates

Acceleration at speeds up to 155 Mbps, turboIP-G2 provides transparent acceleration of TCP sessions, increasing throughput over satellite links while requiring minimal topology changes. The Performance Enhancement Proxy is standards-based, supporting the Space Communications Protocol Standard Transport Protocol (SCPS-TP). The turboIP-G2 provides reliable connection-oriented, end-to-end data transfer for user applications. This powerful platform also overcomes the deficiencies that exist with TCP, including slow start and congestion control. Since they interoperate with TCP/IP networks and devices, turboIP-G2 can be seamlessly integrated into existing networks in a staged manner, avoiding the need for network-wide upgrades.

WAN Optimization

The turboIP-G2 delivers additional value through a powerful add-on WAN Optimization package. The WAN Optimization package was designed to make more efficient use of WAN links through advanced compression techniques as well as practical caching.

CIFS Acceleration provides benefit for remote file access that allows most applications to open and share file across IP-based networks. Data Redundancy Elimination (DRE) takes advantage of file objects with identical content-based data that can be supressed for delivery and completely reconstructed on the receiving end. Any block of data remaining is then sent through a standard compression engine, for additional WAN savings.

Persistent-LZ is a standards-based compression that takes advantage of long-lived history for TCP flows. By taking advantage of data patterns previously seen, the system does not need to resend large chunks of data, only a pointer to the history location. WebCache is standard HTTP caching engine which keeps local copies of any HTTP document passing through. Successive requests for the same content can be served locally versus traversing the link for the same file.

Security Aware

turboIP-G2 meets strong security requirements for cached data by providing an optional secure erase hard drive technology, allowing for deeper storage and history buffers, while maintaining guaranteed security of stored data.

turboIP-G2 Key Features

Implements Open Standards

• SCPS-TP May 1999
• ISO standard (15893)
• CC SDS standard (714.0-B-1)
• MIL-STD (MIL-STD-2045-44000)
• RFCs 768, 793, 1122 & 1323

IPv4/IPv6 Acceleration and Management

• IPv6 traffic acceleration and box management will be necessary as networks around the world are transitioned to Internet Protocol, Version 6.

Selective Acceleration (U.S. patent 6,937,560)

• This powerful feature provides a method of Quality of Service (QoS) for IPv4 datagrams that are received on the local area network (LAN) interface and forwarded to the wide area network (WAN) interface.
• Rules are established to control the processing, including acceleration, compression and filtering for all IP packets. Up to 255 rules can be established. Rule parameters can include source and destination IP address and mask, protocol (TCP, UDP or any), and TCP or UDP source and destination ports. Each rule is assigned 1 of 8 priority levels plus a maximum data rate.

Session Support

• Up to 10,000 simultaneous accelerated sessions.
• Up to 40,000 memorized sessions, for idle session support, when sessions are started and left idle.
• Unlimited session bypass beyond 10,000. If 10,000 sessions are under acceleration, additional sessions will not be stopped, they simply bypass unaccelerated

Data and Header Compression

• Data and header compression functionality is applicable to accelerated TCP traffic, and is enabled/disabled on a session-bysession basis. The compressibility of each segment payload is evaluated individually and only those payloads where the impacts would be beneficial are compressed. Enabling data and header compression on turboIP-G2 can reduce both bandwidth and transmission time over wide area network links. If disabled, no sessions are compressed.

VLAN (802.1Q) and GRE Tunnel (RFC 2784) Support

• Two widely used network protocol configuration and traffic types, VLAN and GRE tunnel acceleration support, provide more network usage efficiency increases.

Intelligent Congestion Control

• Optimized for real-world, mixed-loss environments; distinguishes data corruption from congestion-induced data loss, prevents unnecessary activation of congestion control mechanisms

Automatic Window Scaling (RFC 1323)

• A larger window size is recommended to improve TCP performance in network paths with large bandwidth, and large latency characteristics. By immediately opening the window size of a TCP transfer, TCP slow startup can be mitigated.

Rate Pacing versus Dynamic Bandwidth Support

• Rate Pacing mode meters out bursty traffic based on rate configured for WAN link, preventing channel congestion OR
• Dynamic Bandwidth or per connection mode ensures accelerated traffic optimization when multiple paths exist and those paths contain different bandwidths or delay characteristics

Black Hole Detection & Recovery

• Supports detection and forwarding of non-accelerated traffic to users that do not support the SCPS option

Path Maximum Transmission Unit (MTU) Discovery

• Adjusts MTU per connection based on receipt of ICMP messages

Tolerates packet reordering & Large Queues

• Out-of-order packets not interpreted as lost packet
• Adjusts retransmission timeout (RTO) based on the round-trip time (RTT) to support large queues

Asymmetrical Route Management

• Enables accelerated traffic to pass thru a different pair of turboIP-G2 units in each direction (forward and return link)

Dual OR Single Interface Mode with Policy-Based Routing

• turboIP-G2 can now be placed as network device hanging on edge of network, as opposed to in-line on uplink and/or downlink side, as shown below: