Layer 4 of the OSI Model Handles Transport Protocols

Like TCP and UDP

Layer 4 of the OSI model, also known as the transport layer, manages network traffic
between hosts and end systems to ensure complete data transfers. Transport-layer
protocols such as TCP, UDP, DCCP, and SCTP are used to control the volume of data,
where it is sent, and at what rate.

By providing standardized access to communication services such as connection-

oriented communication, reliability, flow control, and multiplexing, layer 4 eliminates the
need for the application-oriented layers 5 – 7 to consider the characteristics of the
communications network itself. Layer 4 is also responsible for end-to-end error
recovery.

Because layer 4 coordinates data transfer without visibility into message content, layer
4 load balancing can make routing decisions without the need to decrypt or inspect
network traffic. This makes layer 4 load balancing a quick and efficient approach
for packet-level load balancing based on simple algorithms such as round-robin routing.

To route traffic based on media type, localization rules, application, or other criteria
where message inspection is required, layer 7 load balancing must be used.

Layer 3 of the OSI : Networking layer

The term layer 3 function effectively means routing. The term routing describes the

specification of paths (routes) for communication between different networks. This means:

how a data packet travels from subnet A to subnet B.

The third-lowest layer of the OSI Reference Model is the network layer. If the
data link layer is the one that basically defines the boundaries of what is
considered a network, the network layer is the one that defines
how internetworks (interconnected networks) function. The network layer is
the lowest one in the OSI model that is concerned with getting data from one
computer to another even if it is on a remote network; in contrast, the data link
layer only deals with devices that are local to each other.

Logical Addressing: Every device that communicates over a TCP/IP network

has a logical address assigned to it. This address is known as an IP address.
This address is not to be confused with a NIC’s physical address which is the
MAC addressed discussed in the previous article. IP addresses are unique
within the network.
Routing: IP packets may need to be routed between subnets before the data
reaches the target system. It is the network layer that handles incoming
packets from various sources. Routers are a good example of systems
that depend on the network layer to forward traffic to the correct destination.
Data Encapsulation: The network layer encapsulates data received from
transport layer into datagrams, also known as packets and adds a network
layer header.
Fragmentation and Reassembly: The network layer fragments the data if the
packets are too large for the data link layer. Packets are reassembled at the
target node.

PROFINET vs EtherNet/IP: The

Similarities and Differences
PROFINET and EtherNet/IP are currently the most popular Industrial Ethernet
protocols in automated industrial environments. So why choose one protocol
over the other? After all, both offer real-time communication, use the same
Ethernet technology and they can even operate on the same physical network.

Let’s compare these two platforms to see how distinctive they really are. We’ll start by
giving a brief introduction to both, then we’ll move on to their similarities and differences.

What is PROFINET?
Like PROFIBUS, the classic fieldbus solution first introduced in 1989, PROFINET was
developed by a team of major automation companies and several German institutes,
and it is managed by the PROFIBUS & PROFINET International group (PI).

PROFINET is an application, so it sits on Layer 7 (application layer) of the OSI model

delivering data through several communication channels. For non-time-critical tasks
such as configuration and parameterization, it can use TCP/IP or UDP/IP
communication. But for time-critical tasks like RT, IRT or TSN, it skips this and
communicates directly through the Ethernet layer instead.
Introduced in the early 2000s as an open standard, PROFINET is a full duplex system
and runs at high transmission speeds. It sends I/O data between controllers and
devices in standard Ethernet frames, whose sequence is prioritized by a virtual local
area network (VLAN). Because these frames have little overhead, the deviation of the
cycle time, or jitter, is low.

Identification is mainly done using device names that comprise letters and numbers,
and IP addresses are assigned automatically.

Startup description

In the startup of an Profinet IO device (before the IP address has been set) the DCP
protocol is used. It is like DHCP, but without using a central server.

The PLC sends a DCP broadcast message, and all IO Devices on the subnet answer
with their MAC addresses. The PLC sends a DCP message to the IO Device with a
specific MAC address, containing which IP address and station name that the IO Device
should use. The IO Device sets its IP address and station name accordingly.

Then the PLC starts the actual configuration of the IO device, using the DCE/RPC
protocol that runs on UDP over IP.
After the configuration is done, cyclic data is continuously exchanged between the IO-
device and the PLC. This communication runs on Ethernet layer 2, i.e. the MAC
addresses are used for routing frames (the IP protocol is not used in those frames).

Communication

Profinet uses three protocol levels:

Protocol Typical cycle time Ethertype Supported by this software

0x0800 =
TCP/IP (UDP actually) 100 ms Yes
IPv4

RT (Real Time) 10 ms 0x8892 Yes

IRT (Isochronous Real-

1 ms 0x8892? No
Time)

Profinet uses IPv4 only (not IPv6).

Overview of all protocols used in a Profinet

Application

A typical Profinet application needs to handle a multitude of protocols. The P-Net stack
implements some of them, while others are optional. For example HTTP is used by some IO-
devices for additional configuration via a web page, but should be implemented outside the
Profinet stack itself.

Content
Layer Protocol Addressing Protocol Notes
description

Remote login on
SSH = 22
Linux
4 TCP Port
HTTP = 80 Web server for
+ 443 configuration

4 UDP Port DHCP = 67 IP address

+ 68 assignment

SNMP = Network
161 + 162 topology query

PNIO-CM Profinet start-up,

 Content
Layer Protocol Addressing Protocol Notes
description

= 34964 uses DCE/RPC

ICMP = 1 Used by ping

Transmission
TCP = 6
3 IPv4 IP address Protocol number control protocol

User Datagram
UDP = 17
Protocol

LLDP = Link layer

0x88CC discovery

ARP =
IP address lookup
0x0806

IPv4 =
Internet protocol
0x0800

MAC PN-DCP =
2 Ethernet Ethertype Profinet start-up
address 0x8892

Profinet RT PNIO_PS =
= 0x8892 Cyclic IO data

(VLAN = Not really a

0x8100) protocol

(VLAN = For double

0x9100) tagged frames

PROFINET uses TCP/IP where it makes sense (that is, where the data is not time critical). For example:

 Configuration
 Parameterization
 Diagnostics
 Therefore, an IP address is needed for these functions.
There are also other benefits. Many devices have embedded web servers. This allows you to type in their IP
address into a web browser and access them like a webpage. One of the biggest reasons people are moving
to PROFINET is DATA ACCESS. By using IP addresses, it helps integrate device-level information vertically
into higher level MES or cloud storage systems.

Real Time
Now PROFINET does not use the IP address for real-time data; it just uses the physical address, the MAC
address. This means it can skip TCP/IP for faster, more deterministic performance.

What is EtherNet/IP?
EtherNet/IP was developed by ControlNet International (CI) and Open DeviceNet
Vendors Association (ODVA). Introduced in the early 2000s and now managed by
ODVA, Ethernet/IP is based on the Common Industrial Protocol (CIP).

Like PROFINET, EtherNet/IP sits at the application layer and is backed by defined
standards and extensive communities. Unlike PROFINET, it transfers data to the
network via the TCP/IP and/or UDP/IP layers. These packages of data are packed and
unpacked, and headers and info are added and deleted as they move through the
layers.

Two types of messages are carried by EtherNet/IP: “Implicit” messages for sending I/O
data and “Explicit” messages for configuration and diagnostics. Implicit messages are
sent from pre-set memory locations to a controller through a UDP port at prescheduled
intervals. Explicit messages are sent to a client through a TCP port in response to a
specific request for that data.

Identification of EtherNet/IP devices is done through IP addresses.

Layer 1: what the signal looks like on the wire.

Layer 2: what happens where the wire connects to a device. How the signal on the wire
gets created.
Layer 3: where messages longer than a standard Ethernet frame are disassembled
going out or reassembled coming in. (Think your outgoing and incoming email
messages which are probably longer than the Ethernet payload length of about 1500
bytes.)
Layer 4: creates a connection between two devices. (Think open telephone line.)
Layers 5 and 6: do pretty much what’s described.
Layer 7: the actual programs that create and consume what’s transmitted.
With PROFIBUS (and many other fieldbuses), layers 1, 2, and 7 are used. In the
Ethernet world, the model collapses to four layers: