Load Balancer Explained

What is a load balancer? So that is the topic of this video. Now a load balancer is a piece of hardware or software that is used to evenly distribute data activity across a network so that no single server or computer becomes overwhelmed with the workload. So as an example, when a relatively small company has an application such as a website, they will have that website running on a server for their customers. So when those customers want to view the website, their computers will make a request and connect to the server to retrieve the website and view it on their computer. But as the company grows larger and attracts more customers, the server may not be able to handle the amount of traffic from incoming connections from customers. So to handle the increase in traffic, the company will put their website on multiple servers to distribute the workload. But even if the website is now on multiple servers, there still could be a problem if one of the servers is getting all or most of the traffic compared to the others. So to solve this problem, this is where a load balancer comes in. A load balancer is used to evenly distribute data activity across the servers so that no single server becomes overwhelmed with the workload. So now as you can see all the servers are getting an equal amount of traffic because of the load balancer.

So the next question is well how does a load balancer work? Well load balancers use algorithms or rules when deciding on how and where to spread out the network traffic to the servers. These algorithms could be either static or dynamic. A

static algorithm uses fixed rules that don't factor in the current state of the servers. In other words, it doesn't really care if a server is busy or not or what it's doing. It only follows its fixed set of rules and nothing else. And one static algorithm is called roundroin. Roundrobin load balancing is when the network traffic is distributed to the servers in sequential order. So for example, when the first request or connection comes in, the load balancer will assign it to server one. And then when a second request comes in, it will assign it to server two. And then when a third request comes in, it'll assign it to server three. And then once all the servers receive a request, it'll start all over again in the same order. Now this is a simple algorithm and it's best used when all the servers have the same capacity. And another version of roundroin is weighted roundroin. In weighted roundroin each server is given a weight based on how much work the server can do. Servers with more weights such as server one will be given more incoming traffic than servers with less weights. That way the servers with more power can handle more requests. This algorithm is best used when the servers have different capacities than the others. Another static algorithm is called IP hash. IP hash assigns a customer's IP address to a number and then this number is assigned to a certain server. So as a customer request comes in, the load balancer assigns a number to the customer's IP address and then the load balancer will assign that number to a specific server. Now this method is essential for applications such as shopping carts because shopping carts require the same ongoing sessions so that the cart doesn't get emptied if the session gets transferred to another server. So it's important that the customer gets routed to the same server that they were using.

Now let's talk about dynamic algorithms. So unlike static algorithms where they don't factor in the current state of the servers, dynamic algorithms do. Dynamic algorithms use real time monitoring. In other words, they first look at the current state of the servers such as how busy the servers are before they distribute network traffic. And one type of dynamic algorithm is called least connection. Now, this type will direct traffic to a server that has the least or fewest number of current connections. So for example, when a new customer request comes in, the load balancer first looks at the current connections of the servers and the server with the fewest current connections will be assigned the new request. So in this example, since server 2 has the fewest, the load balancer will direct the new request to this server. And just like roundroin, the least connection algorithm can also be assigned weights. and the server that can handle the most amount of connections will be given the most weight. Another dynamic algorithm is called least response time. Now, the response time is the amount of time a server takes to process a request and send back a response. So, the least response time means it's the fastest server. So when the load balancer is using this algorithm, it'll check the status of the servers and send the next request to the server that can process it the fastest. So in this case, server 3 has the least response time. So the load balancer will direct the next request to this server. And another dynamic algorithm is called resource-based. And this is an algorithm that uses agents that are installed on each server. These agents send real-time data to the load balancer. This data is the server's resources such as CPU load and RAM usage. The load balancer then looks at this resource data and determines which server is using the fewest amount of resources. In other words, the server that's doing the least amount of work, and the server that's using the fewest resources will be chosen to handle the next request. So in this case, it'll be server one.

So what are the benefits of load balancing? Well, the first benefit is performance and it does this by spreading the network traffic across multiple servers which reduces lag and response time. Another benefit is high availability because if a server fails, the load balancer will direct traffic to other servers which will prevent an outage. Load balancing also provides scalability and flexibility because you can easily add or remove servers depending on the demand

demand and load balancing can also provide security. So for example, it can help defend against a DDoS attack. So if a DDoS attack were to happen, the load balancer can monitor the traffic and block the attack. Or if the DOS attack gets through the load balancer, the load balancer can spread the attack to multiple servers, preventing a single server from getting overwhelmed with traffic. Hey guys, are you tired of keeping track of your passwords? Does your computer look like this? Or are you tired of forgetting your passwords like me and constantly clicking the forgot password link? Well, so was I. So, for the past 3 years, I've been using RoboForm as my password manager. RoboForm is the number one ranked password manager that makes your life easier by remembering your passwords and logging you into websites automatically. Just remember your one master password and RoboForm remembers the rest. It's just that easy. So say goodbye to searching through text files, notepads, or especially clicking the forgot password links because RoboForm automatically logs you into your favorite websites with one click. So there's no more typing. RoboForm uses industry standard encryption for all your data. So you can be absolutely sure that your information is secure. RoboForm is the perfect solution to your password problems, especially in my case. And if you're interested in getting RoboForm, just click my link in the description below. So guys, I want to thank you for watching this video on load balancers. Please subscribe, leave a comment, and I will see you in the next video.