Stacking is the process of connecting multiple physical network switches together, so they function as a single, logical switch. This is achieved by using stacking-capable switches which have dedicated ports and use dedicated cables to connect to other switches in the stack.

Up to eight stacking-capable switches can be connected through their StackWise ports, using StackWise cables, to form a data stack. This document is focused entirely on this kind of stack. A data stack is alsoreferred to as a switch stack or a device stack.

You can stack up to four switches together through the StackPower ports, using the StackPower cables to form a power stack. So a full eight-switch stack, forms two separate power pools, each sharing power among their four members.

The following are the benefits of stacking:

• Simplified Management: A stack of switches can be managed as a single entity with a single IP address, reducing administrative overhead compared to managing individual switches.
• Increased Performance and Bandwidth: Stacking aggregates the backplane bandwidth of individual switches, providing higher inter-switch communication speeds and overall network throughput. Cisco C9350 Smart Series Switches can offer up to 1.6 Tbps of stacking bandwidth.
• Enhanced Redundancy and High Availability: If one switch in the stack fails, other members can seamlessly take over its functions, minimizing network downtime.
• Scalability: Network capacity can be easily expanded by adding more switches to the stack as needed, up to the maximum supported members.

A switch stack always includes an active switch, and a standby switch. Each individual stacking-capable switch that is part of a stack is known as a stack member. Each stacking-capable switch in a stack assumes one of three roles:

• Active
  • Controls and manages the entire stack.
  • Handles all configuration, management, and Layer 2/Layer 3 protocol operations.
  • Maintains and updates the stack’s configuration files, which include both system and interface settings for each stack member. All configuration is managed through the active switch, with backup copies stored on each member.
  • Relays console access to other stack members as needed.
• Standby
  • Acts as a backup to the active switch and automatically takes over as the active switch if the current active switch fails.
  • Continuously synchronizes with the active switch, maintaining an up-to-date copy of stack configurations.
• Member
  • Forwards data and participates in stack operations as directed by the active switch.
  • Maintains a synchronized copy of configuration files for redundancy purposes.
  • Does not participate in stack management or decision-making processes. You can use show switch stack-member-number

This identifier distinguishes each physical switch within the logical device (stack) for management and configuration purposes.

A standalone device is a device stack with one stack member that also operates as the active switch. You can connect one standalone device to another to create a device stack containing two stack members, with one of them as the active switch. You can connect standalone devices to an existing device stack to increase the stack membership.

You can display stack switch members and their switch numbers by using the EXEC command show switch along with show module .

These identifiers are used for network-level identification of the entire stack. For example, during spanning tree operations or when routing protocols are in-use. In the network, a switch stack is identified by its bridge ID; and if it is operating as a Layer 3 device, its stack MAC address. The bridge ID and stack MAC address is determined by the MAC address of the active switch.

If the active switch changes, the MAC address of the new active switch determines the new bridge ID and stack MAC address. If the entire switch stack reloads, the switch stack uses the stack MAC address of the active switch.

You can use the persistent MAC address feature to delay changing the stack MAC address when the active switch changes. During this delay period, if the previous active switch rejoins the stack—even as a member and not the active switch—the stack continues to use its original MAC address. If the previous active switch does not return within this time, the stack adopts the MAC address of the new active switch.

By default, the stack MAC address stays with the original switch indefinitely. If the original active switch does not rejoin within this period, the stack MAC address changes to that of the new active switch.

To ensure the stack MAC address never changes, you can configure stack MAC persistency with the stack-mac persistent timer 0 command in global configuration mode.

Hello messages are exchanged among all stack members to monitor their status. If any member (including the active or standby device) doesn’t respond, it is either removed from the stack or replaced, with new standby or active devices elected as needed. Additionally, keepalive messages between the active and standby devices ensure redundancy. If either device fails to respond, the appropriate role is reassigned to maintain stack operation.

This section provides information on how modifications to the composition of a switch stack impact its operation. A change in stack membership occurs when physical switches are added to, removed from, or replaced within an existing switch stack. The behavior of the stack during such changes depends on the role of the specific stack member at that time.

A stack member priority value is a configurable number ranging from 1 to 15, assigned to each switch within a stack. This value plays a crucial role in determining which switch will be elected as the active switch during stack formation; a higher priority increases the likelihood of a switch becoming the active switch. By default, every member in a stack has a priority of 1.

To change the priority value for a stack member, use the switch stack-member-number priority new priority-value EXEC command. Any change to a switch's priority is saved to the configuration database but only takes effect after the switch reboots.

You can display the currently active stack member priority value ((the priority applied during its last bootup) by using the show switch in EXEC command. To verify the priority that has been configured and will be applied on the next boot, use the show platform software bp content config command.

Configuration files define how a switch stack operates. These files contain two types of settings that are applied after stack membership changes:

• System-level (global) settings: These settings (such as IP addressing, STP, VLANs, and SNMP) always apply to the entire stack. When a stack member is added or removed, the system-level settings of the current stack are applied automatically to all members, including any new switches that join.
• Interface-specific settings: These settings (such as port descriptions, VLAN tasks on specific interfaces) are tied to each stack member’s unique member number. When a switch joins the stack or replaces an existing member, the following applies:
  • The switch automatically applies any interface-specific settings that exist for its member number.
  • If no interface configuration exists for the stack member number, the switch uses default interface settings.

Offline Provisioning is a feature that enables network administrators to pre-configure a new switch—specifying its stack number, switch type, and interface settings—before physically adding it to an existing switch stack. These settings, known as the provisioned configuration, are stored on the active switch. The switch intended to receive this configuration is called the provisioned switch.

Offline Provisioning consists of two main parts:

• Offline (Pre) Provisioning: Administrators can pre-configure a switch on the active device before physically adding it to the stack. All interfaces and settings are created in advance. Once the switch is connected, the pre-provisioned configurations are automatically applied.
• Auto Provisioning :This process requires no user intervention. When a new switch is added to the stack, interface configurations are automatically generated, and the active device creates the appropriate switch provision command for the new switch.

While the stack will display interface configurations for a provisioned switch in its running configuration, these settings will remain non-operational—and may not appear in feature-specific show commands such as show interface status until the provisioned switch becomes active within the stack. The provisioned configuration is kept in the running configuration even if the provisioned switch is not currently present in the stack, and can be saved to the startup configuration for persistence across reloads.

1:1 Redundancy allows you to designate specific switches in a stack as the active and standby members. This mode enhances stack resiliency by ensuring a dedicated standby switch is always ready to take over if the active switch fails. The existing member switches remain in the same state.

The following are prerequisites for 1:1 redundancy:

• All the switches in the stack must be running the same license level as the active switch.
• All the switches in the stack must be running compatible software versions.

To verify the current stack mode on a switch, enter the show switch stack-mode command in priviledged EXEC mode.

Device# show switch stack-mode
Switch   Role    Mac Address    Version   Mode    Configured   State
---------------------------------------------------------------
1       Member 3c5e.c357.c880             1+1'     Active'     Ready
*2      Active 547c.69de.cd00    V05      1+1'     Standby'    Ready
3       Member 547c.6965.cf80    V05      1+1'     Member'     Ready

The Mode field indicates the current stack mode.

The Configured field refers to the switch state expected after a reboot.

Single quotation marks ( ' ) indicate that the stack mode has been changed.

Stacking connects multiple physical network switches, enabling them to function as a single logical switch by using stacking-capable devices with dedicated ports and cables for interconnection.

• No software configuration is required for stacking.
• Before you install a Cisco 9350 Series Smart Switch, decide on your data stack and StackPower requirements. For a data stack, you need either 0.5, 1.5 meter or 3 meter StackWise cables.

For more information, see Cisco 9350 Series Switches - Preparing for Installation

The entire switch stack (including all member interfaces), are managed through the active switch, which is the single point of control. Network administrators can perform management operations using the CLI, SNMP, or any other supported network management applications. Individual stack members cannot be managed separately. All management must be conducted via the active switch. SNMP can be used to manage network features on the stack, but only for those features that the switch’s supported MIBs include. There are no supported MIBs for stacking-specific functions like, switch stack membership and active switch election.

While the basic operation of a switch stack requires no specific global software configuration beyond physical connection, the active switch serves as the central point for configuring both global features that apply to the entire stack and interface-level settings for individual stack members.

Global settings affect the behavior of the entire switch stack. An example of a global configuration is setting the persistent MAC address timer, which controls how long the stack retains the MAC address of a previous active switch after a switchover.

Although individual stack members cannot be managed separately in terms of their own control plane, the active switch allows for specific configurations or operations to be applied to individual members. This includes setting properties like a member's priority or provisioning a new switch before it joins the stack.