WTF is Scaling?
consider you're designing a system for some web or mobile application, so when you're building modern applications — especially those that have to live in the Cloud — one of the first challenges you’ll face is how to handle growing traffic and data; this is where the use of Scaling comes to play.
to put it in simple words, Scaling refers to - the ability of your system to handle increased load (or site traffic), whether it's from more users, more requests, or more data.
now, Scaling usually involves 2 Primary Strategies and that being:
- Vertical Scaling (Increasing or Up Scaling)
- Horizontal Scaling (Branched or Wide Scaling)
and just like any other system in this world - these 2 Primary Scaling Systems also have some pros and cons, and knowing when and how to use them is key for designing reliable and efficient systems (so that your system doesn't end-up crashing).
WTF is Vertical Scaling?
I'll be straight-forward here, Vertical Scaling means increasing the power of a single machine or designing a single system with more efficient performance than multiple systems combined — that is possible by adding more CPU, RAM, or storage to an existing server and make it more capable.
think of it like upgrading your laptop: more RAM, faster processor — same device, just better performance.(basically you can consider the example of a GPU too - not to mention NVIDIA's RTX Series)
using Vertical Scaling is a common starting point for small applications because it’s:
- simple to implement as it's all a one-man game
- cost-effective in the early stages because of-course you don't have to handle or look after multiple machines or servers, a single-server system is handling the whole business here
- requires no changes in your application architecture, like using multiple servers or machines may require constant updates in the system architecture
Example:
if your application is hosted on a Virtual Machine (VM for short) with 2 vCPUs and 4GB RAM, scaling vertically would mean switching to 4 vCPUs and 8GB RAM.
WTF is Horizontal Scaling?
Horizontal Scaling means adding more machines or servers to your system and distributing the load among them.
just think of it as hiring more team members to divide the work, instead of making one person work overtime.
Horizontal Scaling is only needful when:
- you expect high traffic or fast growth on your system
- you need high availability and redundancy to avoid system crashouts
- you want to scale without hitting hardware limits - free space is needed to avoid latency
Example:
instead of upgrading one server, you spin up 3–5 servers and place them behind a load balancer which distributes user requests evenly.
Comparison - Vertical v/s Horizontal Scaling
let’s look at how these two approaches compare:
| Feature | Vertical Scaling | Horizontal Scaling |
|---|---|---|
| Better Known As | Increasing or Up Scaling | Branched or Wide Scaling |
| How it Works | Upgrade the Existing Machine | Add More Machines - Widen Out |
| Cost | Initially Cheaper | More Expensive |
| Complexity | Easier to Manage | Requires Load Balancing, Networking |
| Scalability | Limited by Hardware | Virtually Unlimited |
| Fault Tolerance | Single Point of Failure | High Availability |
| Use Case | Small to Medium Apps | Large-Scale, Distributed Systems |
Advantages of Horizontal Scaling
- High Availability if one server fails, others can continue serving users — there’s no single point of failure.
- Flexible and Scalable you can add or remove servers on-demand depending on traffic or load and while this happens - other servers continue to perform without any interference.
- Better Performance at Scale each server handles a portion of the load, making response times faster and more reliable.
Advantages of Vertical Scaling
- Cost Effective (Initially) upgrading one server is cheaper than managing multiple servers.
- Simple to Implement no need for load balancers or complex networking as it only involves a single-server.
- Easier Maintenance managing one powerful server is often easier than managing a cluster of servers.
Disadvantages of Horizontal Scaling
- Complex Architecture requires load balancers, service discovery, and sometimes distributed databases.
- Higher Costs infrastructure costs increase with more machines, along with higher DevOps overhead.
- Synchronization Overhead keeping data consistent across nodes becomes challenging.
Disadvantages of Vertical Scaling
- Limited Growth there’s a hard limit to how powerful one server can become.
- Single Point of Failure if the server crashes, the entire application gets fucked up.
- Downtime During Upgrades hardware upgrades may require system downtime or restarts.
In Summary
understanding the difference between horizontal and vertical scaling is critical when designing systems that need to grow - especially in the case of social media apps.
Key Understandings -
- use Vertical Scaling when starting small, for quick wins, or when your system doesn’t need to handle a massive user base.
- use Horizontal Scaling when your application needs to be highly available, fault-tolerant, and infinitely scalable.
a well-designed system often starts with vertical scaling and later evolves to use horizontal scaling as demand increases, so knowing when to scale and how to do it properly is what separates good system design from a great system design.
Thank You for Reading - This Blog has been written by Aayush Mishra for - Blogspot!