A practitioner's guide to building scalable systems

Building Scalable Systems: Distributed Systems basics to System Design Interviews

We want to give you the tools to compare multiple approaches and pick the one that works for your scenario.

Grokking the Distributed and Scalable Systems

# What is DNS Load Balancing?
As we've seen, DNS is the phonebook for the which returns an IP address for the website. It turns out that you can specify multiple IP addresses against the same website name. e.g.

2606:4700:10::6814:7b7 educative.io

2606:4700:10::6814:6b7 educative.io

When the DNS is queried, it sends the list of IP addresses in a Round-Robin manner. Most clients (operating systems) simply pick the first IP address in the list and hence every client will be randomly going to a different server. By applying this technique, you can use the DNS to act as a load balancer for your website. 

## Problems with DNS Load Balancing
DNS Load Balancing us used in some cases but is not recommended for Load Balancing. Here are the problems associated with Round-Robin DNS for load balancing.

### 1. Too many Servers
If you are operating at scale, you might be using hundreds or thousands of web servers. DNS is not designed to map 100s or 1000s of Servers to a single IP address.

### 2. Caching makes it difficult to replace servers
DNS is designed in a way such that each Node in the DNS hierarchy Caches the name to IP mappings for a certain duration (typically 6-12 hours). Now if a server has crashed, clients won't be able to access that server and would have to wait for the cache to expire to fetch the fresh list of IP addresses to choose from.

Based on the above two issues, Round-Robin DNS can cause a lot of hassle in failure recovery, error direction and server replacement. Hence, don't use it.

# DNS Load Balancing to route traffic to different Data Centers

As you grow, your service will be inevitably hosted in multiple data centers and you would be able to serve your users. DNS Load Balancing can be useful in that case.

In a typical setup, each Data Center will have its own Load Balancer in front of the application servers and the DNS will contain the list of the IP addresses of the load balancers.

Unlike using DNS for server load balancing, this is a great use case for DNS load balancing. Load Balancers have known IP addresses and are less likely to fail.

# Geo DNS

When you have multiple data centers, it makes sense to route the client to the data center closed to their physical location. This improves latency and response time for each client. That's where Geo DNS is used.

When a client queries the DNS, Geo DNS looks for the IP address of the Load Balancer that's closed to the physical location of the client and then sends that IP address to the client.





What and Why

Node in a Distributed System

Performance and Numbers

Fault Tolerance and Availability

Understanding TIME in Distributed Systems

Data Centers for software engineers

Scaling Distributed Systems

Load Balancers

Structured Storage at Scale

Sharding in Storage Systems

SQL vs NO-SQL

Blob Storage at Scale

Caching in Distributed Systems

Queues and Asynchronous Processing

Data Analytics and Batch Processing

Logging and Monitoring

Searching at Scale

Conclusion

DNS Load Balancing and Geo DNS

What is DNS Load Balancing?