Overview

In distributed systems at the end level, a server or machine and a NIC (Network Interface Card) have a defined physical IP address. A physical IP address is an address that is assigned to one machine or device. It cannot be assigned to another physical device. In other terms, a physical IP address is tied to a specific machine.

A virtual IP address is not tied to a specific machine or device. A virtual IP address can move from one device to another device. It is like a floating IP address.

When we say a Virtual IP address belongs to a server at a point in time then that doesn’t mean that it is tied to the network interface card of the server. It is assigned to that server at one moment of time while the same virtual IP address could be assigned to another server at a different moment of time.

The typical use case of a virtual IP address is when there is a particular set of physical devices or servers acting as a single application or a service.

The virtual IP address acts as an entry point for that application or service and provides load balancing and failover mechanisms and helps them in achieving high availability. How it provides these things we will see later in this tutorial

A Virtual IP address can be assigned –

• Multiple Servers

• Multiple Domain Names

• Multiple Applications residing on a single server

• Single Server itself
• …

How Virtual IP Helps in high availability (Load Balancing and Failover)

Let’s see examples of both

Example of Failover

One example where a virtual IP address helps in high availability and provides failover is the case of databases. In databases generally, we have a primary and a couple of secondary nodes. The primary node is behind a particular virtual IP address. Now the client always communicates with the virtual IP address through which it connects with the primary node.

Now if behind the scenes, the primary goes down then its virtual IP address will now be given to the machine which becomes the new primary. This is possible because the Virtual IP address is not to a particular physical machine as compared to the physical IP address which is tied to a specific machine. The client would still communicate to the same virtual IP address. It wouldn’t get to know that the primary has changed and it is now a different machine.

What we described above is a failover mechanism provided by Virtual IP

Example of Load Balancing

Another example of High Availability by providing Load Balancing using Virtual IP is as below

Since the Virtual IP address is not tied to a specific machine, we can have a load balancer built to send messages to a set of Virtual IP addresses behind it. Each of the machines behind the load balancer is given a Virtual IP address. When any machine goes away due to some failure or some other reason then its virtual IP address could be assigned to a new machine that gets spun up.

We can think of having a set of free virtual IP addresses behind a load balancer. Whenever a new instance comes up it takes one Virtual IP from that free pool.

Conclusion

This is all about virtual IP addresses. Hope you have liked this article. Please share feedback in the comments

Note: Check out our system design tutorial series System Design Questions