Transport settings manage Transmission Control Protocol (TCP).

Transport settings are under Optimization > Network Services: Transport Settings. Skipware (SCPS) settings will be available in a later version.

After applying your settings, you can verify whether the changes have had the desired effect by reviewing the Current Connections report. Additionally, under the Congestion Control section, the report displays the specific congestion control method that the connection is currently using.

Identifies the optimal TCP configuration by using the same mode as the peer appliance for inner connections. It defaults to SkipWare when negotiated, or standard TCP for all other cases.

If your environment consists of a mix of different network types connecting to a hub or server-side SteelHead, you can enable this setting on your hub appliance. This allows the hub appliance to reflect the various transport optimization mechanisms used by your remote site appliances.

Appliances advertise their automatic TCP detection capabilities to peers through their out-of-band (OOB) connections.

Applies data and transport streamlining to non-SCPS TCP connections. When enabled, this setting forces peers to use standard TCP as well. It also clears any previously set advanced bandwidth congestion control settings.

Enables high-speed TCP optimization, designed to make more complete use of long fat pipes (high-bandwidth, high-delay networks). However, it should not be enabled for satellite networks.

We recommend enabling high-speed TCP optimization only after carefully evaluating whether it will benefit your specific network environment. Additionally, this feature can be enabled using the tcp highspeed enable command.

Uses an intelligent algorithm along with a modified slow-start algorithm to optimize performance in long, lossy networks like satellite, cellular, microwave, or Wi-Max networks. This feature is a sender-side modification of TCP and is compatible with other TCP stacks.

Estimation is based on the analysis of ACKs and latency measurements. The modified slow-start enables faster ramp-up in high-latency environments compared to traditional TCP. The intelligent bandwidth estimation algorithm allows it to learn effective transmission rates for use during modified slow start. Additionally, it can differentiate between bit error rate (BER) loss and congestion-derived loss, managing both types appropriately.

Bandwidth estimation offers good fairness and friendliness toward other traffic on the same network path.

Applies TCP congestion control to each SCPS-capable connection. The congestion control uses:

This transport setting uses a modified slow-start algorithm and a modified congestion-avoidance approach. When enabled, SCPS per-connection ramps up flows faster in high-latency environments and handles lossy scenarios while remaining reasonably fair and friendly to other traffic. SCPS per-connection is especially effective in satellite networks, offering high performance.

We recommend enabling per-connection if the error rate in the link is less than approximately 1 percent.

Accelerates SCPS and provides error-rate detection and recovery. This setting allows per-connection congestion control to tolerate some loss due to corrupted packets (bit errors), without reducing throughput, using a modified slow-start algorithm and a modified congestion-avoidance approach. It requires significantly more retransmitted packets to trigger this congestion-avoidance algorithm than the SkipWare per-connection setting. Error-tolerant TCP acceleration assumes that the environment has a high BER and that most retransmissions are due to poor signal quality instead of congestion. This method maximizes performance in high-loss environments, without incurring the additional per-packet overhead of a forward error correction (FEC) algorithm at the transport layer.

Enables the Cubic congestion control algorithm, which offers better performance and faster recovery after congestion events than NewReno, the previous local default.

Sets a global data-transmit limit on the link rate for all SCPS connections or when an appliance is paired with a third-party device running TCP-PEP (Performance Enhancing Proxy). By default, rate pacing is disabled, and it does not support IPv6.

Rate pacing works by combining MX-TCP with your chosen congestion-control method. It monitors the actual link rate, adjusting the MX-TCP rate accordingly to prevent congestion, packet bursts, and congestion loss as it exits the slow-start phase. The slow-start phase increases the transmission window size gradually as the network throughput is determined.

Without congestion, the slow-start phase ramps up to the MX-TCP rate and stabilizes. When congestion is detected, whether from other traffic, a bottleneck, or a variable modem rate, the congestion-control method activates to avoid issues and allows for faster exit from the slow-start phase.

To use rate pacing, the client-side appliance informs the server-side appliance, which should have its TCP optimization set to auto-detect for automatic configuration negotiation. You should also configure an MX-TCP QoS rule to apply the correct rate limit. If the QoS rule is missing, rate pacing will not be applied, and the congestion-control method will take over.

You cannot delete the MX-TCP QoS rule when rate pacing is enabled.

The buffer settings support high-speed TCP and are also used in data protection scenarios to improve performance.

Specifies size used to send data out of the LAN. The default is 81920.

Specifies size used to receive data from the LAN. The default is 32768.

Specifies size used to send data out of the WAN. The default is 262140.

Specifies size used to receive data from the WAN. The default is 262140.

High-speed TCP (HS-TCP) is designed for high-bandwidth, high-latency networks (also known as long fat networks, or LFNs). It optimizes connections by accelerating traffic over WAN links with large bandwidth but high latency. HS-TCP is automatically activated when the Bandwidth Delay Product (BDP) exceeds 100 packets.

To calculate the BDP and determine if HS-TCP should be enabled, use this formula to calculate the WAN buffer size:

If the calculated number is greater than the default (256 KB), enable HS-TCP with the correct buffer size. For a link of 155 Mbps and 100 ms round-trip delay:

To configure high-speed TCP, enable the high speed congestion control algorithm, increase the WAN and LAN buffers, and then enable in-path support.

Activates or deactivates all rules in the Single-Ended Interception (SEI) rules table. By default, it is disabled. When disabled, you can still add, move, or remove rules, but they won't take effect until the table is enabled.

The appliance uses the rules to determine whether to enable or pass through SCPS connections. When receiving an SCPS connection on the WAN, the appliance only evaluates the SEI rules table. To pass through an SCPS connection, we recommend setting both an in-path rule and a single-ended connection rule.

Two default rules appear in the list. Both bypass acceleration and pass through matched connections. The first rule matches connections destined to ports listed in the Interactive port label. The second matches connections destined to ports listed in the RBT-Proto port label.

You can also enforce rate pacing for SEI connections where no peer appliance or SCPS device is involved, and SCPS is not negotiated. To enforce rate pacing for a connection, create a rule enabling TCP proxy, select a congestion method for the rule, and then configure a QoS rule (with the same client/server subnet) to use MX-TCP. The appliance then accelerates WAN or LAN-originated proxied connections using MX-TCP.

Specifies the rule’s location in the list.

Specifies IP address and subnet mask for the traffic source or destination.

Specifies the destination port number, port label, or all ports.

Specifies a numerical ID from 0 to 4094 that identifies a specific VLAN. Enter all to match all VLANs, or untagged to match untagged connections. Passed-through traffic maintains existing VLAN tags. The default is 0.

Specifies whether matching connections are sent to a web proxy. Requires that the appliance has web proxy enabled.

These settings determine the action that the rule takes on matching SCPS and TCP connections.

Specifies the action that the rule takes on an SCPS connection. When an option is enabled, web proxy settings become available. Multiple selections are allowed. SCPS discover and TCP proxy are disabled by default, which passes through traffic unoptimized.

Accelerates matching SCPS connections.

Accelerates matching TCP connections.

The TCP optimization settings for single-ended connection rules are similar to the global settings.