I’ve often been asked to explain what I do so that someone without the background knowledge can understand what I do. So, I’ve spent some time to think about analogies for the different types of network related terminologies that I’ve worked with.
There are many types of industrial networking medium out there – Ethernet, ControlNet, Modbus, etc. This is not a comprehensive list but a list of what I’ve worked with. The different medium provides different speeds of how fast data can move. The faster it is, the faster you get your data, just like when you’re watching a movie on YouTube – faster connection = faster load times. In industrial networking, you can think of serial, DF1, DH+ as the dirt road that is bumpy and dusty – it’s old and it serves its purpose. Then you have RS422, RS485 – slightly faster but still slow. Then you have ControlNet, a slightly faster medium. It’s like the highway that you drive on. Next we have Ethernet, the freeway. Ethernet can exist in multiple physical media – CAT5, CAT5E, CAT6, Fiber. The difference in all these types of roads is how much data can pass through at any time. Think about how many cars can drive on the dirt road and how fast they can go… Then compare that with the country road or highway. You can fit more cars and the roads are smoother so you can drive a little faster.
Ethernet, like the freeway, exists in many versions – the 2-lane freeway is like 100MB, some call it Fast Ethernet. Then you have 1GB Ethernet, that’s like your 3-lane freeway. Today, we have 10G Ethernet. That’s like the autobahn.
I’ve spent quite a lot of years working with Ethernet. Using the freeway analogy, think about the different freeways that you drive on. Some have potholes, some are straight and go for hundreds of miles, some go in circles and some take you to one place and you need to take another freeway to get to your end destination. Well, Ethernet can be designed in similar notion – you can have a flat Ethernet network, where the freeway will take you everywhere. Flat network is a quick way to make things work – anyone can get to anywhere by taking any path because there’s more than one freeway to your destination. Well, while it works, at some point, you’ll find yourself living in a place where freeways are stacked on top of one another. The functionality works right? You can get from point A to point B if you took any path. But it’s not the most efficient way of doing things. Why is this important?
Well, if you’re a city planner, you want the city to look nice, multiple freeways stacked on top of one another is a sign of over congestion. And it’s a patch. Stacked freeways may work today but at some point in time, you’re going to run out of room to grow.
This is where proper planning comes to play. A network needs to be designed in such a way that it allows for growth and it handles your day to day traffic. Quite literally, network traffic is like city traffic. There are times of the day when the network is very congested – everyone is going to work at the same time and everyone needs to get there at the same time. When this happens, things slow down. Imagine the data traffic – this is something you want to avoid in industrial network because the data is time sensitive (latency). Late data in industrial network is useless data. It can also be wrong data. Low latency indicates that it takes very little time for your data to get to its destination.
Now that you understand the concept of latency in a network, the next thing that needs to be figured out is how to handle the traffic at those peak times? Well, we’ve talked about having multiple freeways that lead us to the city. Okay, industrial network is not a democratic nation… It’s more like Simcity – a game where you build a city and designate zones where you want residential areas where people live and commercial areas where people shopped and industrial areas where people worked. In the ideal situation, you don’t live far from where you work and you don’t go far from where you shop. This way, people can get to their end destination quickly (low latency). In Ethernet, this is where you have a layered network.
So, if people lived close to where they work, they don’t have to go very far. Their commute is shortened and therefore, their time on the road is reduced, even during peak hours. If their commute is short, we don’t really need a big road do we? So, instead of spending on Gigabit network (1G), we can get away with Fast Ethernet (100MB). This translates to cost savings.
Now that we’ve talked about layered network and latency, we need to find a way for the traffic to move from one point to another. You need to have road signs that properly direct traffic. Your road signs would show you how to get from point A to point B. When you get to point B, you’ll get instructions on how to get to point C. This is the concept of network routing. Routing is done by routers or Layer 3 routing switches, where these Layer 3 switches know how to control traffic from Point A to Point B, and where to move the data. Sometimes, you have multiple road signs. This is like when you come to a fork on a freeway, you can choose to go to point B or point C. Point C can also lead you to point D and this might be seen on the road sign in front of you. Or if you’re driving, you know that point B is on your way to point E, so you need to take that route to E. Network routers know a finite number of routes. You can tell a router all the different routes in every possible combination but what you end up with is a very big road sign, that is confusing and hard to understand. If you come across a road sign with 10 options, you have to slow down such that your brain can process what your next path is… You don’t want your network to slow down for something like that. You want it to keep moving such that it gets there faster.
What are some of the other important things in industrial networks? Well, there’s firewall. These are like security checkpoints. Here, everyone stops such that only the appropriate traffic passes through. With this, you typically have firewall rules, where you can allow only red cars to pass at certain times of the day and you can allow only certain types of trucks to carry certain types of goods through. Think about it – do you want to let sensitive data pass through to the wrong hands? Another thing a firewall does is let appropriate traffic come in. You want to block out the bad guys but you want to make sure that your people get what they need so every traffic is inspected. This is not like the “random sampling” that goes on at an airport. Everyone is inspected at these checkpoints, regardless of who they are and how important they are. This definitely slows down your industrial network – you will take a hit on latency because you want to make sure you have enough personnel that does the security checks for the right amount of expected traffic. If you have a lot of traffic and not enough personnel, then things will slow down because these searches are thorough and they take time.
A couple of basic Ethernet terminologies – packets. A packet is like a mail that you’re going to send. There’s the overhead in mailing something – you need to write it on something and you need to put stamps on it. In a data packet, there’s this overhead information. Every packet knows where it came from and where it is going. There are two waysof sending your data – TCP and UDP. TCP is like putting the To address and your return address on the envelope (packet). With this, the packet knows where to go and how to get back to the sender if it becomes undeliverable. UDP is like a mass mail that is sent from one source to anyone with a legitimate mailing address. It can be so broad that everyone gets it. Well, like every junk mail that you ever received – you have to do something about it – read it, or dump it… UDP is a quick way to get your information across – like you’re having a one day sale and everything is 50% off. If you’re writing an application and you’re trying to decide what to use, UDP is good for data that you don’t mind losing. Not every UDP packet will get to its destination. Some are lost and you won’t know about it. So, if your information is important, you need to use TCP. TCP knows if the letter is undeliverable – you can then do something about it.
Next we talk about managed vs. unmanaged switches. Well, a managed switch is like a freeway with an electronic road sign, equipped with cameras that monitor the traffic real time. The road sign can change the direction of traffic if needed. The cameras can be used to monitor the traffic and they can report to a centralized traffic control room.
In Ethernet network, SNMP can be used to monitor the traffic. SNMP is not limited to networks, it can be used for servers and computers as well. But that’s a different topic. So, when you’re thinking about managed vs. unmanaged, you need to ask yourself if you ever want to know what’s going on in your network. The cost is definitely a lot higher but it has its benefits for sure.
With managed switches, you can do port mirroring, where the traffic going into one port is copied to another, typically a port with packet capture. I don’t have a good analogy for this but it’s important if you ever want to keep traffic flowing accordingly while giving yourself an opportunity to watch what’s going on. Packet capture is a means of collecting data to be analyzed later.
With managed switches, you can do VLAN tagging, where traffic of certain type is only allowed to pass through. It’s like allowing only red cars go through path A but not path B. This is another way for making sure you control unnecessary traffic from taking the wrong path to their destination. VLAN is virtual LAN, it’s a way of re-using your same physical switch for multiple functionality. You may have one VLAN for data traffic, one VLAN for backup and one VLAN for management only. Instead of putting 3 different switches in, you put one in, and then segment the switch appropriately. This is another form of network management.
Then there’s network Prioritization. I don’t know much about this quite yet because I have never needed to do this yet. Most switches that I’ve used do this well enough that I don’t ever configure it. It’s useful for voice data, where you want to make sure that when people talk, they hear everything at the same exact time. Think about a call that you have made where what you say is only heard one second later. You get your point across but it’s annoying as hell – you say something, wait a little bit such that the other person hears it and responds. Annoying isn’t it? Network prioritization, conceptually, is a way for controlling what traffic has priority over another.
Some other things you can do with an Ethernet network (and managed network) is the use of jumbo frames. Standard packet size is 1500 bytes. This is like a passenger vehicle. It can only carry so any passengers before you need to have a vehicle to carry additional passengers. You can expand this standard packet size to something like 9000 bytes, where now you can carry more passengers in one trip. You will have to use less buses to move the same number of passengers? Well, you also don’t want to use a bus if you’re carrying 1 or 2 passengers for 23 hours of the day and 10 passengers for that one hour. Well, you can – no one is stopping you but it’s just not an efficient way of doing things.
All these networking concepts are intended to provide some basic understanding of Ethernet network, which is one of the better networking technologies out there. It can be problematic if you don’t design the network correctly. Imagine yourself being a city planner. Use the information that you have to design the correct network for what you need. If you need help, ask – I’m here to help.
