What are the switch stack options available in the network market

  • Find the switch stacking options for your network.
  • Cisco stacking variants per switch models.
  • HP Intelligent Resilient Fabric.
  • DELL Stacking variant per OS version.
  • Summary and final considirations.

Find the switch stacking option for your network

Stacking a Network switch becomes more and more popular having more variations and version coming from different switch brands manufacturers such as Cisco, HP and Dell. In a way it becomes more acceptable over the traditional spanning-tree variants. Network Professionals considers switch stack specially on the campus distribution, core and datacenter aggregation layer, though not all network design are same but this is where is mostly fit in.

Stacking a switch can be describe as grouping a multiple switch to act as a single forwarding switch. Though the term stacking can be use brand agnostic, it has different specification from different switch models and brands.

Stacking a switch gives the network an understanding that the group switch are a single logical switch even though physically they are separate switches connected by cables and protocols. The stacking setup provides the benefits of link failover where in if one link fail the other link still be active to support the traffic. Stack switch itself is also provided with a high-availability pair so as it can be redundant in case of downtime or failure. Part of the features of switch stack is to aggregate a link to allow a scalable bigger throughput, say each 1Gbps cable link be logically group into four will makes a 4Gbps link. Or a 10Gbps pair with another 10Gbps line makes total of 20Gbps link giving more room for network traffic most especially for the uplinks.


Switch stack logocal setup
Switch stack Physical and logical setup


While we describe the fundamental characteristics of stacking a switch, it is important to understand that not all stacking techniques available on the market are the same. Technical specification differ from different switch manufacturer and models. It can differ mostly from the number of switch it can join in a group. While some can do only two in a group others can max up to more or less 9 in group. Management of switches may or may not be as one switch as well.There are option that will require management for each switch individually. Also to put into consideration are the supported switch ports and cables that are supported for a stack. It is also best to stack switches of same brand as mix switch brands would not most of the time be compatible given that stacking techniques comes with proprietary protocols that joins them.


There are probably a lot of stacking options out in the market from different manufacturers, what is going to be discuss on this article do not emphasises inclination to popularity nor a manufacturer’s stacking advantages over another but rather what the author have encounter so far. let us discuses those coming from vendors such as Cisco’s Stacking versions, the VSS, VPc and HPe’s IRF and Dell’s MLAG, VLT.


bus to pology or chain stack
Switch stack in Bus to topology or also called chain stack


Ring topoly Swicth stack
Switch stack in Ring topology


Cisco stacking variants per switch models

Cisco have a lot of stacking variants define by different switch series. The lowest model that support stacking in a SMB (Small-Medium Business) switch category is a basic L3 switch, the Cisco 500 series which support up to 8 units stack forming ether ring or chain topology connected with 10Gmbps fiber interface.

On the Cisco Catalyst series enterprise switch, lowest model that support stacking is the 2960 model series that utilize the Cisco “Flexstack” and “Flexstack Plus” options. Setup can be a Ring topology with max four member stack (Cisco 2960-S) or eight member (Cisco 2960-XR and 2960-X). The 2960 “Flexstack” cable specifically design to be use between 2960 switches for stacking and do not use regular port interface and cable.

Higher model catalyst, the Cisco 3850 switch model have the “stackwise-480” as its stacking options where the “480” digit on its name represent the maximum throughput this stack technology can accommodate on a Cisco 3850 switch stack in a ring topology. Also note that lesser 3xxx model switches, the Cisco 3750 and Cisco 3650 both have similar technology on a lesser throughput which is with “Stackwise Plus” (64Gbps) and “Stackwise-160” (160mbps) respectively. This 3xxx switch model also use stacking cable for stacking purpose only.

Cisco VSS (Virtual Switching System) is a technology suitable for the core layer of the network providing High-Availability and link failover, hence Cisco switches that supports these are those normally position on the core such as the Cisco 6500. Unlike the previously discuss stacking variant above, cisco VSS do not offer centralize management between the two core wherein each switch on a VSS are manage separately. The maximum on a VSS are just two switch appliance connected using a two to eight combination of 10Gbps interface. Protocols between the two core appliance is called VLT and its downlink multi-chassis system is called MEC (Multi Chassis Etherchannel) supporting an active-active line by either PagP(Cisco proprietary) or  LACP(widely use and Open standard)

For datacenter switches, Cisco’s Nexus’s switches support vPC (Virtual Port Ethernet). vPC have similar characteristics as the Cisco’s VSS  expect that VSS is for catalyst while vPC is for Nexus switch series.


HP Intelligent Resilient Fabric

While Cisco have its VSS, HP’s on the other hand have IRF (Intelligent Resilient Fabric). HP IRF have similar characteristics as VSS where in HP switch in IRF are logically become as one single unit called as IRF domain.  On a single site IRF domain, specification states that the failover can be effective even up to 70km of distance.

IRF Supports up to four chassis core switches or nine access layer switch model. No additional license is required as well. IRF runs on standard hardware interface/cable of 10, 40 and 100Gbps ports to form an HP IRF domain and has a failover time of 50ms. IRF also support the standard LACP protocol for link aggregation

Setup Support single master and the rest as slave switch in an ether Bus or ring fabric topology like the other stacking version, in addition IRF also supports Spine-and-Leaf topology which is more common deployment setup found on datacenters environment where   in IRF support two CB (Controlling Bridges) in a domain and each CB support up to thirty PE (Port Extenders) as per HP’s initial release of Spine-and-leaf on IRF.


DELL Stacking variant per OS version

Dell switches come from its acquisition of the Force10 Network. Today Dell have variety of network switch series that support campus and datacenter network. Two of the Dells OS system type are the Dell OS6 and Dell OS9. The Dell OS6 support MLAG protocol while the DELL OS9 support VLT. Dell Switches that support OS6 are the Dell N-Series, on the other hand OS9 is on the S-Series switches. The MLAG and VLT both have similar features just that they are supported on different switch Models/OS system. While the N-series are normally use for campus distribution and core, the S-Series are typically use for datacenter though can be implemented on large campus network as well. It can be describe that Dell’s MLAG competes with the VSS, while the VLT competes with vPC.

Dell MLAG allow multiple chassis switches to interpret the uplink switch as one logical switch. Similar with the VSS and IRF, Dell MLAG also support the LACP protocol for link aggregation. In a MLAG topology, connecting two switch as core in a High-Availability setup is called Peer-Link, where one of the pair act as the primary and the other as secondary. Dell MLAG support two peer switches per MLAG domain and eight interfaces per Peer-Link

VLT (Virtual Link Trunk), offered on Dells S-series switches with Dell  OS 9. VLT Is design as Dual Active Control Plane Multi-System LAG. VLT support up to 2 peers. While on Stacking, can accommodate up to 6 units. VLT Stacking is manage on single control plane or single logical fabric, while the data plane remains distinct on each switches.


Multi chassis LAG switch setup
Multi-chassis LAG Setup showing Physical and logical representation



Stacking switch have different specification and approach base on different switch brand implemented and network use case. On purchasing a network switch with the intent of implementing a link aggregation and failover in mind, it is important to put in consideration the network switch brand as each switches manufacturers have different proprietary technologies that is interoperable with same brand given that the all requirement is complied. For a mix brand of switches the nearest open standard technologies that provide link failover are the Spanning-Tree-Protocol (STP) and LACP.

Other consideration to keep in mind is the distance of physical switches from each other. If this group of switches needs are to be group in a stack check on the specification of the switch stacking technology if it would support a certain distance. Will the stacking technology of your choice be using standard network cable and interface port or would require a proprietary stacking cable with limitation on the length. Finally, part of the planning and designing for stacking switches is to utilize a stacking technology base on what role the switch stack would be applied to, will this be for campus access, distribution, core or datacenter.

Leave a Reply