Invented by Israeli; Etgar, Naor; Elad, Caspary; Guy, Tsai; Mel

Welcome! Today, we are going to learn about a new way to balance traffic in computer networks. This invention deals with how to send data smoothly, so no network device gets too busy. Let’s break it down step by step so you can understand how it works, why it matters, and what makes it different from things that came before.

Background and Market Context

In the digital age, our lives are filled with messages, pictures, and videos moving across networks. Think of these networks like busy highways for information. When many cars (data) try to use the same road at once, traffic jams happen. That’s why companies use special tools, called load balancers, to direct cars to different roads, so every road is used well and nothing gets stuck.

Now, let’s imagine a giant data center. Inside, there are many computers and switches, all connected by wires. Each switch or computer is like a toll booth or intersection on the highway. When a message enters the network, it needs to get to its destination as fast as possible. If too many messages go to the same place, that spot gets crowded, and everything slows down. So, we want to spread out the messages evenly.

This problem is not new. Big web companies like Google, Amazon, and Facebook face it every day. They use load balancing to make sure their websites are always fast and never crash because of too many visitors. The challenge is even bigger now, as more people use cloud services and as networks become more complex. If the load balancer doesn’t do its job well, you might see slow videos, dropped calls, or web pages that don’t load.

Over the years, many methods have been tried. Early load balancers used simple math to split up the traffic. Later, more advanced tools, called hash functions, helped decide where each message should go. One of the most popular hash functions is called CRC (Cyclic Redundancy Check). CRC is fast and easy to use, but it comes with a big problem: sometimes, it puts too many messages on the same path, creating a jam. This is called “traffic polarization.”

In modern networks, where everything needs to be quick and reliable, polarization is a big headache. If one server or switch gets overloaded, users get upset. Companies lose money. That’s why new and better ways of load balancing are needed. This invention steps in to solve this problem by making the balancing much smarter and fairer.

Scientific Rationale and Prior Art

To understand why this invention is special, let’s look at how things were done before.

First, most load balancers would look at the information in a data packet — things like where it’s coming from, where it’s going, and what kind of message it is. The balancer would use this info to make a “load balancing vector,” which is just a string of numbers that represents the packet’s main details.

Then, the balancer would take this vector and run it through a hash function like CRC. The CRC scrambles up the numbers, turning them into a smaller number called a “key.” This key tells the balancer which road or path to send the message down. The idea is that if the hash function is good, the messages will be spread out evenly, and no road will get too crowded.

But here’s the problem: CRC and other hash functions are not perfect. If two packets look a lot alike — say, they have similar addresses or patterns — the CRC might give them the same key. That sends them both down the same road, causing a jam. Even if you change the “seed” (a secret number used to scramble the hash), CRC can still make the same mistake. This is especially true in networks with lots of similar packets, like data centers.

Some smart people tried to fix this by changing the fields in the vector for each device, but this made the system very complex and slow. Others tried adding new kinds of hash functions or mixing up the input data in clever ways. For example, “color recombining” and “coprime-based” balancers tried to make the keys less predictable, but they only worked in special network setups.

Hardware solutions like “murmur hash,” “shuffle-CRC,” and “CRC-RC5” tried to make things better by using more logic or combining different tools. Murmur hash is strong but slow and needs a lot of hardware. Shuffle-CRC tries to mix up the bits before hashing, but if you want lots of variety, you need even more hardware. CRC-RC5 combines two operations, but if the first step (CRC) gives the same result for two vectors, the second step (RC5) can’t fix it.

All these old methods face the same problem: they either don’t solve the traffic jam, or they make the system too big, too slow, or too expensive. There’s always a trade-off between fairness, speed, and cost. The market needs a solution that is fast, fair, easy to build, and works in any network setup.

That’s where this new invention comes in. It introduces a smart way to change the load balancing vector before hashing it, making the final keys more spread out and less likely to create jams. The secret is in using “multi-stage substitution” — a trick borrowed from cryptography — which scrambles the vector in a clever way before hashing.

Invention Description and Key Innovations

Let’s walk through how this invention works, step by step, and see what makes it stand out.

Imagine a device — this could be a switch, router, or server — that wants to balance its incoming network traffic. Here’s what it does:

Step 1: Receiving the Load Balancing Vector
The device gathers key fields from each data packet it receives. These fields might include source and destination addresses, protocol type, session info, or anything else that helps identify the packet. These fields are combined into a “load balancing vector.” Think of this as a special code that describes the packet.

Step 2: Determining a Unique Seed Value
Every device in the network has its own “seed” value. This seed might be based on the device’s name, number, or some other unique marker. The seed is like a secret code, and it’s different for every device. This makes sure that even if two devices see the same packet, they’ll treat it a little differently.

Step 3: Encoding the Vector
The device uses its seed to “encode” the vector. Usually, this means it takes the vector and runs it through an exclusive-OR (XOR) operation with the seed. XOR is a simple but powerful way to scramble numbers. After this step, the vector is changed just enough to make it unique to this device.

Step 4: Multi-Stage Substitution
Here’s where the magic happens. The encoded vector is now run through a process called “multi-stage substitution.” This is borrowed from how secret codes (ciphers) work. The vector is chopped into little blocks, and each block is sent into a tiny gadget called an S-Box (Substitution-Box). The S-Box takes a small group of bits and swaps them for new bits, using a rulebook. You can think of it like a secret decoder ring: you put in a number, and you get out a different number.

The invention uses two stages of these S-Boxes. First, the vector is split up and sent through the first group of S-Boxes. The results are then mixed up and sent through a second group of S-Boxes. Each S-Box can have its own rulebook, and the way the boxes are connected can be different for each device. This double layer of substitution makes it very hard for two similar vectors to come out the same at the end.

Step 5: Generating a Modified Vector
After the S-Boxes are done, the device has a new, scrambled vector. This is called the “modified load balancing vector.” Even if two packets looked very similar at the start, after these steps, they will likely be very different now.

Step 6: Hashing the Modified Vector
Now, the device takes the modified vector and runs it through a hash function, like CRC. Because the vector has already been scrambled in a special way, the hash function is much less likely to give the same result for similar packets. This means the keys that come out — which decide where the packets go — are more spread out and fair.

Step 7: Sending Packets to the Right Destination
The key produced by the hash function is used to pick which path, server, or port to send the packet to. Since the keys are now more random and less likely to clump together, the traffic is spread out evenly. No single road gets too busy, and the network runs smoothly.

Why is this Better?
This invention stands out for several reasons:

– It needs very little hardware — just a couple of S-Box stages are enough, so it’s cheap and easy to build.
– The scrambling is fast and can all happen in one clock cycle, so there’s no extra delay.
– It works with any network setup — you don’t need to change your whole network or use special topologies.
– It doesn’t need to copy groups or duplicate devices, which would waste resources.
– It makes sure traffic is spread out as evenly as possible, even if many packets look alike.

Technical Details Made Simple
Let’s make the technical details easy to digest:

– The S-Boxes are just small lookup tables or circuits that swap input bits for output bits. They can be built into chips with little effort.
– The XOR operation is a very fast way to scramble numbers. It’s used all the time in computers.
– The two-stage substitution acts like a double shuffle, making it almost impossible for similar inputs to turn into the same output.
– By using a unique seed for each device, the system makes sure that each device scrambles the packets in its own way, helping avoid any patterns that could jam the network.
– The hash function at the end is the same as before, but now it works better because the input is much more random.

Putting It All Together
The device can be a server, switch, or even a virtual machine. The method can be used in big data centers or small office networks. It can handle lots of traffic without slowing down, and it keeps the network fair for everyone.

This new way of balancing is like having a traffic cop who changes the color and size of every car, so no two cars look alike. When the cars reach each intersection, the cop can easily send them down different roads, keeping traffic moving smoothly.

Conclusion

In today’s fast-moving digital world, keeping data flowing smoothly is more important than ever. This new invention offers a simple, fast, and highly effective way to balance network traffic. By scrambling packet information in a smart, two-step process before hashing, it ends the problem of traffic jams caused by old-fashioned load balancers. It also stays easy to build and use, without needing lots of new hardware or changes to the network.

If you run a data center, manage cloud services, or just want your network to work better, this new load balancing device and method could be the key to smoother, faster, and more reliable data traffic. The approach is innovative, practical, and ready for the networks of tomorrow.

Click here https://ppubs.uspto.gov/pubwebapp/ and search 20250219945.