Author Archives: Serenity Huang

What Is a Core Switch and Why Do We Need It?

Network switches are categorized into different types according to different principles, such as fixed switch and modular switch based if you can add expansion module to it, and managed switch, smart switch and unmanaged/dumb switch depending on whether you can configure it and the complexity of the configuration. Another way to classify the type of a network switch is by the role it plays in a local area network (LAN). In this case, one switch is considered to be an access switch, an aggregation/distribution switch or a core switch. In small networks we do not see core switch. So many people are having questions about what core switches are. Do you know what is core switch? Is there only one core switch in a network? What are the differences between core switch and aggregation/access switch?

What Is Core Switch?

If we spend some time looking up dictionaries for the meaning of core switch, we will find a definition similar to “A core switch is a high-capacity switch generally positioned within the backbone or physical core of a network. Core switches serve as the gateway to a wide area network (WAN) or the Internet—they provide the final aggregation point for the network and allow multiple aggregation modules to work together (An excerpt from Techpedia).” The definition explains its high-capacity feature, the physical location and its function of connecting multiple aggregation devices in network.

What Are the Differences Between Core Switch and Other Switches?

The biggest difference between core switch and other switches is that, core switch is required to always be fast, highly available and fault tolerant since it connects all the aggregation switches. Therefore, a core switch should be a fully-managed switch. But if it is a switch not used in the core layer, it can be a smart switch or an unmanaged switch.

Another difference is that, the core switch is not always needed in a LAN while we may often have the aggregation switch and the access switch. Because in small networks that have only a couple of servers and a few clients, there’s no actual demand for a core switch vs aggregation switch. In the scenario where we don’t need the core layer, we often call it a collapsed core or collapsed backbone since the core layer and the aggregation layer are combined.

The third difference is that there’s generally only one (or two for redundancy) core switch used in a small/midsize network, but the aggregation layer and the access layer might have multiple switches. The figure below shows where the core switch locates in a network.

Core switch in the core layer

What Should Be Kept in Mind When Using Core Switch?

The first thing we should keep in mind is that core switch is urgently required in two occasions. One occasion is when the access switches are located in different places and there is a aggregation switch in each place, then we need a core switch to optimize the network. Another occasion is when the number of the access switches connecting to a single aggregation switch exceeds the performance of it, and we need to use multiple aggregation switches in a single location, then the use of core switch can reduce the complexity of the network.

With core switch and without core switch

As for specific type and number of core switch that we should adopt in a network, that depends on the scale and budget of our network, including how many servers, clients or lower layers switches we have. For example, say that a small network has 100 users and has 6 48-port Gigabit aggregation switches, a suitable core switch will be like Juniper EX2200, Cisco SG300, or FS.COM S5800-8TF12S.

The second thing is that a core switch should be fully-managed, which means it should support different method of management, such as web-based management, command line interface and SNMP management. Also it should have some advanced features like support for IPv6, built-in Quality of Service (QoS) controls, Access Control Lists (ACLs) for network security.

And generally the connections to the core layer should be the highest possible bandwidth. In addition, since the core switch act as the center of a LAN, it should be able to reach any devices in the network, not directly but within the routing table. A core switch is usually connected to the WAN router.


In the design of a network, there might be access layer, aggregation layer and core layer. Though the core layer is not required in smaller networks, it is indispensable in medium/large networks. And the high-capacity core switch plays an important role in delivering frames/packets as fast as possible in the center of the network. Its contribution can not be underestimated especially in networks where speed, scalability and reliability are key to users.

10G Ethernet: 10GBASE-T or 10G SFP+?

10GBASE-T has been available as an add-in card in servers, switches and network interface cards (NICs) since 2008, and it has been widely adopted since 2012. It is highly praised for its advantages which include lower cost than 10G fiber, cost-efficiency of using existing MAC (Media Access Control), easier migration from 1GBASE-T to 10GBASE-T, and the ability to deliver PoE (Power over Ethernet). Does that mean we should all turn to use 10GBASE-T now? And what are the 10GBASE-T cable requirements? Every application differs, let’s see some specific cases in short-reach applications.

10G copper or fiber

Where Can 10GBASE-T Be Used?

When building a 10G network, the link can be either copper or fiber. If using 10GBASE-T cable, the places are required to be in the Data Center or Horizontal areas (in building, including wiring closet). But it is not suited for Vertical (riser links) applications within building, or campus & metro applications.

Cases for 10G Ethernet Connections

Case 1: Connecting a switch with only SFP+ ports to a switch with only 10GBASE-T ports.

10GBASE-T cable 1

When the distance of these two switches are less than 30 m, which is the max. link distance for 10GBASE-T copper SFP+ module, the desired connection for them can be made by using a 10GBASE-T module and a Cat6a cable. It’s the simplest solution for this case.

Case 2: Connecting two switches with only 10GBASE-T ports.

10GBASE-T cable 2

Connecting two switches with all 10GBASE-T ports are as simple as placing the plug into its mating socket. One Cat6a Ethernet cable is born for such a connection and that is why it is called the standard 10GBASE-T cable. By using a Cat6a cable for 10GBASE-T, it can reach up to 100m distance.

Case 3: Connecting two switches with only SFP+ ports.

10GBASE-T cable 3

There are three choices for connecting two complete SFP+ switches. For distances between 30 m to 400 m, it is recommended to get two 10GBASE-SR SFP+ modules for each switch and connect them with a OM3/4 LC duplex multimode fiber patch cable. The second is to use two 10GBASE-T SFP+ modules and Cat6a cable. If the link is as short as 7 m, it is suggested to use a low cost 10G SFP+ direct attach copper (DAC) cable.

Case 4: Connecting switches with both SFP+ and 10GBASE-T ports.

10GBASE-T cabling 4

When the two switches both have SFP+ and 10GBASE-T ports, you will be free to use methods from Case 1 to Case 3 above. But in my experience, it would be better to use the 10GBASE-T copper ports first, and save the SFP+ ports for possible future connections to an optical network for longer transmission distance.

Words in the End

10GBASE-T is taking its way to being more extensively used on network gears without a doubt, and cost for deploying 10GBASE-T equipment will be lowered with its wide spreading. Know the requirements for 10GBASE-T cabling is necessary for correctly choosing between 10GBASE-T or 10G SFP+ in practical usage. After all, cost-efficiency is very important in large-scale deployment.