About SteelHead

Riverbed’s leading network optimization solutions, with innovative acceleration for Cloud, SaaS, and premise-based applications and services, improve application performance and move large amounts of data. Prepared for the era of artificial intelligence, our cutting-edge acceleration solutions drive high-performance resilient networking that delivers:

• Performance that efficiently moves large amounts of data at scale.

• Security that protects your data against emerging cyber threats

• Agility to futureproof your network investment as your needs change.

Riverbed appliances ensure peak performance for every digital experience and AI application by securely delivering data at scale over every network. Our novel Flex licensing model provides the agility to quickly adapt your network acceleration to needs of the business.

Appliances intercept client-server connections without interfering with normal interactions, file semantics, or protocols. All client requests are passed through to the server normally, while relevant traffic is optimized to improve performance.

Using the SteelHead product family can reduce WAN bandwidth utilization by 65% to 98% for TCP-based applications. They do this using a method called scalable data referencing (SDR), in addition to standard data compression.

SDR breaks data into small, unique chunks and stores them. Each chunk is given a unique label (a reference number). Instead of sending the same data over the network repeatedly, the system just sends the reference number. This saves bandwidth; one small reference can replace megabytes of actual data. The receiving appliance uses the reference to retrieve and rebuild the original data.

This process significantly cuts down on the amount of data sent over the network, boosting efficiency and speed.

The SteelHead family of appliances use a method called transport streamlining to reduce delays and improve the performance of TCP traffic between devices. This method combines standard and specialized techniques to:

• use efficient ways to resend lost data like TCP selective acknowledgments.

• adjust TCP window sizes for better data flow and reduced delay.

• boost data transfer speeds over different types of WAN connections.

In addition to data and transport streamlining, appliances also accelerate a rich set of application protocols that includes but is not limited to Microsoft Exchange, NFS, SMB, and HTTP/HTTPS.

For 4090 appliances, slot 0 paths (0-4) are non-bypass paths. Fail-to-wire, therefore, is not supported on those paths.

Most SteelHead models and in-path NICs support fail-to-wire (or bypass) mode. If there’s a failure or loss of power, the appliance switches to bypass mode, allowing traffic to pass through without acceleration. By default, this is the failure mode.

Some NICs also support fail-to-block mode, where the LAN and WAN interfaces shut down during a failure, stopping traffic.

Fail-to-wire mode prevents a single point of failure by letting traffic flow even if the appliance has issues or isn’t powered on. You’ll be notified in the following ways when the appliance enters bypass mode:

• The Intercept/Bypass status light on the bypass card

• A Critical status in the appliance Management Console

• SNMP traps (if configured)

• System log entries (syslogs)

• Email notifications (if configured)

Once the issue is fixed, new connections are accelerated automatically, but connections made during the failure are not. You can use the kickoff feature to reset those connections and optimize them, though this is usually not needed since most connections are short-lived.

In fail-to-block (or disconnect) mode, the appliance shuts down both LAN and WAN interfaces during a failure, blocking traffic. This forces traffic to be rerouted to other paths if other SteelHeads are deployed.

This mode is useful if your network can automatically reroute traffic when the appliance blocks it. You can use this mode with connection-forwarding, the allow-failure command, and another SteelHead on a different path to ensure redundancy.

Riverbed simplifies appliance deployment and management through these key methods:

• Autodiscovery allows SteelHeads and SteelHead Mobile to automatically detect and connect with remote counterparts, enabling traffic acceleration without needing manual network configuration. Autodiscovery helps administrators control connections, secure traffic, and choose peers for optimization.

• SCC provides centralized configuration and monitoring for multiple SteelHeads. It offers a single view of the overall network performance and health, making management easier. The intuitive interface lets you manage appliances using high-level categories like applications, sites, and networks. With SCC, you can streamline configuration and management tasks, making hybrid networking more efficient at scale.

• SteelHead Mobile Controller monitors the health and performance of each software client installed on endpoints. It also manages licenses and certificates (where applicable). The SteelHead Mobile Controller helps you track which users are connected, view their data reduction stats, and perform support tasks like resetting connections, pulling logs, and troubleshooting, all without needing user input.

Peering rules, typically configured on server-side appliances, let you manually define how appliances connect with each other or enable automatic connections through autodiscovery. In-path rules, typically configured on client-side appliances, allow fine-grained control over optimization methods the appliance uses to accelerate matching traffic.

To manage appliances with a controller, the controller must be running the same software version or a later version than the appliances. You cannot manage appliances running a version later than the controller’s version.

We recommend placing appliances behind your firewall. To ensure proper communication through firewalls, the following ports must be open:

• 7881 and 7882 for TLS optimization

• For the management plane, controllers communicate with managed appliances on TCP ports 9443 and 22.

• For the control plane, controllers communicate with managed appliances on TCP port 9443.

• The encryption service uses encapsulating security payload (ESP), also known as IP protocol 50 and is part of the IPsec suite of security protocols. If the network is public, the service uses UDP on port 4500.