I would like to know the meaning of MAC interface and Line interface in detail.

The MAC interface is a digital interface that does not incorporate physical layer PHY functions, and refers to interfaces (SGMII, RGMII, USXGMII,​ ​etc.) that can be connected to other devices including CPUs and physical layer PHY devices.

Line interface refers to an interface with built-in physical layer PHY functions (Copper: 1000Base-T, Fiber: 1000Base-X, etc.).

Will it be flooding if there are some DAs missing in the MAC address table due to multi cast?

If there is no DA in the MAC address table, the corresponding packet will be flooded.

Regarding QoS, is it simply sent in order of priority, or is the allocation ratio of the ring buffer changed?

Depends on scheduling settings. With strict scheduling, packets are sent in the order of high priority ⇒ low priority.

If your switch supports a feature called Weighted round robin, you can also adjust the priority ratio.

Is the MAC address table created for each L2 switch? Will it be shared between switches?

Although it depends on the device specifications, it is possible to recognize whether to have an individual table or to share it with other switches depending on the switch settings.

Devices connected to L2 switches don't know IP addresses, how do they know other devices? Is it the MAC address?

Even in a LAN built with L2 switches, connected devices and equipment can finally know the MAC address of the destination using ARP (Address Resolution Protocol).

The L2 switch refers to the MAC address table that links MAC addresses and port numbers to switch packets.

Is it correct to understand that the TOS field in the IP packet is naturally not referenced by the L2 switch?

Depending on the switch settings, the TOS/COS fields can be referenced.

Is the MAC interface compatible? For example, is it possible to connect an RGMII compatible device and an MII compatible device?

Cannot connect. Connect with the same MAC interface. However, if the device has an RGMII interface, there are many cases where the MII interface is also supported.

Regarding QoS, how should Priority be set for each service?

L2 switches have various settings for referencing Priority, such as per port, per MAC address, and per packet (TOS/COS/PCP).

Please select the most suitable method for the application to be developed as to which priority should be referred to.

Will the MTU change if I add a VLAN tag?

Yes, as you know. The maximum size of general Ethernet frame is 1518 bytes, but it becomes 1522 bytes when VLAN tag is inserted.

How should the L2 switch support TSN?

Since TSN is a generic term that summarizes multiple standards, please check which standard you should meet first.

Depending on the standard, an external CPU or software may be required.

I would like you to consider which device to implement which function, including L2 switches, CPUs, and software.

Is the data in the FCS a simple checksum or something?

Yes, as you know.

In Store and Forward, when an FCS error occurs, it is discarded by the L2 switch, but is it possible to forward it without discarding it?

A typical switch checks for FCS errors at the MAC layer, so packets with FCS errors are discarded upon receipt.

Some switches have a function to remove the FCS error check.

Why can't FIFO transfer between different speed links? Does it mean that reorders are not allowed?

In general, FIFO does not secure a large capacity area, so it cannot handle cases where there is a speed difference such as 100M ⇔ 1G.

Are Store and Forward and Cut through settings generally changeable?

Yes, many switches have been released that support Cut through as it is a common feature.

I think most of the switches in the world are store and forward, but what kind of applications are cut through switches used for?

It is used for applications that require real-time performance where differences in latency between packets can cause problems.

Why are daisy chains popular in industrial applications?

It is often used in factory lines, and the lines often span hundreds of meters, so in such cases, a daisy chain network is often constructed.

When did TCAM come to be implemented in L2 switches?

L2 switches became popular around 1990, when TCAM was used for the MAC table.
After that, a method that can be realized based on inexpensive SRAM was developed, and in networks such as enterprises and SOHO, there is a history of shifting to SRAM-based L2 switches instead of TCAM.

​TCAM continues to be used in large L2 switches used in carrier networks and data centers.
However, in recent years, even in the enterprise and industrial markets, there has been an increase in user-defined types or special required specifications.

I would like to know the prospects for further speeding up of the L2 switch interface (25G compatibility, etc.).

​Devices for data centers and carriers that require high-capacity communication already support interfaces such as 25G, 100G, and 400G.

You introduced a switch IC with a built-in microcomputer, but is there a model with a built-in PHY IC as well? (Something that can be connected with "switch IC (built-in microcomputer) → transformer → RJ-45")

Yes. Many of Marvell 's L2 switches have a built-in microcontroller and PHY.

Are there cases where packets get corrupted even though there are no FCS errors?

An FCS error will always occur if the packet has garbled bits on the transmission line.