Welcome back. In this episode, we're gonna start talking about application load balancing with an introduction to load balancers.
The objectives include understanding load balancers, identifying load balancer features and then understanding load balancer Excuse that are available inside of azure.
So first, what is a load? Balancer? Load Balancers and azure are responsible for traffic distribution that allows scaling your applications and virtual machines to support millions of network streams. Inbound and outbound. A load balancer is the front end for inbound network flow and is responsible for distributing that traffic to back in pool instances
like a group of virtual machines, all running the same application or website
load balancers air not just for inbound traffic flow to its back end Resource is a load. Balancer can also be used for outbound connectivity for virtual machines, allowing those virtual machines to access the Internet Using a load balancers Public i p address Load balancers can be public or internal.
A public load bouncer will have a radical public I P address assigned to it that will receive request while distributing traffic to Internal Resource is with private I P addresses
an internal load balancer will have a private i P address attached to its front end and will distribute traffic to virtual machines in an internal private network. A scenario for a public load balancer would be using one to distribute Web traffic from external resource is to a pool of back and servers hosting the website.
Load balancers responsible for distributing traffic evenly to the back end server So one server is not overloaded with request while others sit idle.
An example of an internal load balancer would be putting one in front of a group of database servers, since these database servers would not need to be accessible on the Internet. But a Web application could make calls to a database server.
An internal load balancer would distribute this traffic from the application across multiple database servers.
Let's discuss a couple of features of our load balancer.
The first is how the load balancer determines how to distribute traffic. This is done three hash based distribution system.
The load balancer looks at five different pieces of information when it receives an inbound network stream. The source i p. The source port, the Destination I p. The destination port and the protocol being used. This is used to direct traffic to the available servers and allows the external in point to keep that same session to the back and server.
If the external in point creates a new session, the source board will change and cause the traffic to go to a different server in the pool.
You can use all five pieces of information or only used two or three, which will cause better affinity meaning. If the port changes, the endpoint would still connect to the same backend server. You can also use a load balancer for port fording. This allows creating a rule to take incoming traffic on a specific port and friend and i p address and send that
to a specific port on the back and instance in the pool.
A common scenario is taking an incoming RTP request over an unconventional port, say 50,000 and translating that to port 3389 to a specific virtual machine in the back in pool. This would allow excess in the internal virtual machine, while it doesn't have a public I P address assigned from the Internet.
Finally, we have health probes, which allows the load balancer to determine if the back and server in its pools are healthy and available to accept incoming traffic.
If a health probe fails, the load balancers stops any traffic to the failed back and server until it's healthy again.
Health probes can be configured using TCP HTTP or https
as your load balancer comes in two different varieties or skews basic and standard.
A standard load balancer can do anything that a basic load balancer can do. However, the to do have some distinct differences that will cover in this table. The first is the backend pool size. A basic load balancer only supports 100 pool size, while a standard can support up to 1000 2nd is for the health probes.
Basic doesn't support https, while the standard load bouncer skew does
the next is availabilities owns the standard load balancer has an availability zone, which means it is automatically configured across multiple availability zones inside the azure region.
As long as the zone remains healthy inside the region, the low bouncer can still process traffic to the back and servers, and you can see in the table the basic load bouncer does not have this option.
Next is default security
by default the basic load balancer will accept all incoming traffic to it and is not restricted
while in the standard load balancer, all ports are closed until they're loud Through security rules.
The next is S L. A. The basic load balancer actually doesn't have an S l a. While the standard load balancer does of a 99.99% up time as long as there are two healthy virtual machines and the back in pool. Finally, there's pricing.
The basic load balancer is free and no charge. While the standard charges based on the number of rules in the amount of inbound and outbound data process through it.
Next, let's talk about back end pulls. This is a group of servers providing the same function that the load balancers Since traffic, too,
the back end pool can be made up of a couple of different options. It can be an availability set, a single virtual machine or a virtual machine scale set
so you can see some of the concepts we learned back in the virtual machine. Episodes are coming back into play when configuring our load balancers On the right hand side is a little screen shot of just showing how to create this back and pool and the options to associate it with. And we'll see more of this in our demo later.
That does it for some of the basics of our load balancers. Want to follow it up with a quick quiz question. What does the load bouncer used to determine traffic distribution.
And again, it goes back to our five to pull hash that it creates using a source I p Sore sport Destination I, P. Destination port and the protocol
coming up next. We're gonna take some of these concepts and see what they look like in our load. Balancer Demo. See you in the next episode.