Scalability² - the stepping stone to a...

It’s no secret that Polkadot is the most scalable technology for building and hosting any kind of Web3 application. In the past year I’ve been working with an awesome team at Parity to implement and deploy Elastic Scaling, a game changer for rollup scalability and the Web3 cloud.

In this article I aim to explain the scalability story of Polkadot, the value proposition of Elastic Scaling and how the Web3 Cloud compares to traditional Web2 Clouds.

Before I dive into it, I want to thank Joyce Siqueira, Natalie Tillack, and Oliver Brett for their additional input and help.

In the context of blockchains, it has been frequently argued how the technology behind Polkadot offers a compelling solution to the scalability problem. In Web3 this problem refers to the limitations of networks in handling a growing number of transactions. Without a proper solution, this can lead to network congestion, slower confirmation times, and higher fees. Several factors contribute to these issues, including block size, network bandwidth, storage requirements, and consensus mechanisms.

To solve them, Polkadot splits load across many interoperable rollups (previously known as "parachains") which are secured by the Relay Chain. In the past few years we’ve continued to innovate and improve this amazing technology. We have incrementally pushed the evolution of Polkadot by scaling up in all dimensions and introducing layers of abstraction to ultimately shift the perspective towards a first-of-its-kind Web3 Cloud.

In this post we will explore Polkadot's scalability, discussing the scaling techniques that have been implemented, those still to come, including Elastic Scaling, and the general scalability challenges of Web3 applications.

Our journey starts by going back in time to understand how Polkadot evolved to its current state.

The Polkadot decentralized multicore computer

Launched in 2021 and completed two years later, Polkadot 1.0 was the first step towards a multicore "world computer". It represented a groundbreaking advancement in blockchain technology: a fully decentralized and scalable platform supporting a multitude of interconnected rollups.

Well ahead of its time, this innovative architecture allowed for a vast array of use cases, from decentralized finance and supply chain management to identity verification and gaming: Centrifuge , Hydration , Kilt are just some of these examples. This architecture allowed Polkadot to address the scalability problem at the L1 level from day one. We now see others considering similar solutions: for example the Ethereum Foundation has recently proposed to build native rollups.

Polkadot introduced the concept of “a core” as a fundamental unit that provided compute and bandwidth for rollups. Note that a core's resources can be consumed by rollups every 12 seconds. A total of 50 cores were available at the time, each one providing the following resources:

500ms of execution
5MB of bandwidth

What is a core in Polkadot? It is important to make a distinction on how Polkadot cores are fundamentally different from regular CPU cores. The actual computation is not performed once, but multiple times to ensure that the result is verified in a fully trustless and decentralized way. At the lower levels it is implemented via specific protocols that leverage cryptography and economic incentives to ensure that the outcome of the computation is correct. No malicious actor can prevent the computation from happening or alter its result.

How does a Polkadot core provide bandwidth? Due to its decentralized nature, Polkadot doesn’t make any assumptions about the input data storage location in relation to the compute resources location. The Polkadot DA (data availability) layer uses erasure coding to securely store the data in smaller redundant fragments on as many different nodes as possible. Each core provides the necessary bandwidth for this process. This is a necessity for implementing a fully trustless and permissionless system that can securely store these inputs. The bandwidth of the core bounds the size of the input available for computation.

Polkadot has made a few early simplifying assumptions that had significant impact on both vertical and horizontal scalability in the short and medium term:

core resources can be used only once every 12s
each rollup is only able to use a single core

This resulted in a hard cap of maximum 50 rollups running on the global computer. Rollups are able to connect to Polkadot in a fully permissionless and trustless way after acquiring a core by participating in and winning auctions.

However, these simplifying assumptions are not to be considered one-way doors. In the next section you will learn how Polkadot has dramatically improved the rollup scalability story.

Polkadot 2.0 - faster and more accessible

Horizontal scaling

We’ve incrementally optimized and improved the consensus technology of Polkadot so the validator count can be raised from 200 validators (2021) to 500 validators (2024) with a goal of 1,000 validators (2025).

This increased the number of usable cores and how many rollups can run on Polkadot. It also increased the resilience and security of the network, skyrocketing Polkadot to the top of the pile as the PoS network with the highest Nakamoto coefficient . Learn more about why decentralization (measured by this coefficient) is important when comparing different networks in this short video.

In December 2024, Polkadot demonstrated what performance can be achieved with just 15 cores (out of 100) during “The Spammening”. It already outpaces top competitors like Solana and Aptos, proving its technical edge and, while impressive, these results only scratch the surface of Polkadot’s true potential.

Polkadot is not only the fastest in terms of throughput, but is also the leader in terms of data availability bandwidth by far.

Check out the full December live test results here.

Vertical scaling

In 2024, Polkadot launched a major protocol upgrade to further scale the overall network capacity vertically as well as improving individual rollup performance characteristics. This feature reduced the rollup latency from 12s to 6s, effectively allowing one core’s worth of resources to be consumed every 6s. This doubled the individual rollup throughput. At the same time the compute resources a core provides was increased from 500ms to 2s which represents an additional 4x increase in throughput.

Following that upgrade, Agile Coretime was launched. It rendered the auction model obsolete and lowered the entry barrier for builders. It enabled simpler and more efficient management of compute and data availability bandwidth resources. What was once a very static model of resource consumption is now dynamic, allowing two distinct ways for rollups to acquire and use cores:

Bulk: A fixed duration of continuous core allocation represented by an NFT that can be purchased through coretime sales with DOT. It can be split, shared, or resold, enabling rollups to share the same cores, effectively doubling the cap on how many rollups can run in the system.
On-demand: As the name suggests, a single core can be acquired and used immediately when needed.

Elastic Scaling - more vertical scalability

Scheduled to be launched in Q2 2025 on Polkadot, Elastic Scaling provides a proper solution to the scalability challenges that are currently affecting successful Web3 applications.

High Transaction Fees:

Network congestion on popular dapp platforms like Ethereum often leads to a bidding war driving up transaction fees (gas chart), as reported here. The focal point of the problem in most cases is the architecture which has inherent limitations in transaction throughput (transactions per second).

The effect is of great consequence, hindering the large-scale adoption of Web3 technology in critical sectors like financial services, for example. We can look at DeFi applications and reason that compared to their Web2 counterparts they suffer from:

Throughput bottlenecks, leading to much lower performance and usability
Higher cost of microtransactions and frequent interactions

Slow Transaction Times:

Network congestion can slow down transaction confirmation times, which negatively impacts the user experience and creates significant limitations for real-time Web3 applications.

Web2 has incrementally raised the bar in terms of UI and UX, conditioning users to expect applications that are both fast and responsive. Slow-loading applications are simply unacceptable in today's digital landscape, and we must strive to bridge the gap and meet these heightened expectations.

The Benefits of Elastic Scaling - Use Cases and Examples

Polkadot’s transition as a first mover into the Web3 Cloud space means that it has become inherently scalable. It is able to satisfy the most demanding use cases - something it could not do before. As such, we are pushing the industry scalability gap between Web2 and Web3 to grow ever shorter.

The value proposition of Elastic Scaling is built around key use cases that were not possible before due to the aforementioned unsolved challenges. All rollups running on Polkadot can adjust their usage of core resources on the fly to match any of the following application profiles:

High throughput, low latency

Popular games with millions of players are a prime example of CPU-intensive applications. Gamers anticipate that their actions will have immediate results. Minimizing transaction latency is crucial for enhancing the Web3 gaming experience and bringing it closer to the engagement level of Web2 games.

High throughput and bandwidth

Elastic scaling can significantly increase the speed of CPU and bandwidth-intensive applications, such as airdrops, account migrations, and storage migrations, potentially up to six times faster.

Low throughput, high bandwidth

Solutions for L2 availability, decentralized storage, IoT networks, or other DePIN use-cases, might require significant bandwidth but less CPU. These applications can scale their bandwidth usage up to 12MB/s when the need arises.

Low latency

Applications using 12 cores can confirm transactions in 500ms, improving user experience and efficiency where speed is critical, such as in high-frequency trading.

Elastic Scaling gives Web3 real cloud capabilities

One of the key properties of any Web2 cloud is that the applications they host can scale their resource usage up and down to achieve lower costs or better end user experience. Polkadot cannot truly become the first Web3 cloud unless it is able to exhibit the same property.

To achieve the Web3 Cloud vision, we have improved the scalability story of rollups in Polkadot. And we have done so without compromising the long term value of security and decentralization for short term gains in performance or scalability.

The table below shows how Polkadot further improved rollup scalability with the introduction of Elastic Scaling.

Polkadot 1.0	Polkadot 2.0	Polkadot 2.0 + Elastic Scaling
Execution time (per 6 seconds)	500ms	2s	Up to 12s**
DA bandwidth/storage	0.9MB/s	0.9MB/s	Up to 12MB/s
Latency (confirmation time)	12s	6s	500ms
Number of cores	1	1	Up to 12 cores

** The execution time and maximum number of cores that can be fully used depend on how fast the sequencer hardware is.

Elastic Scaling builds on the existing capabilities of Polkadot without introducing new concepts or requirements for rollups. It simply allows rollups to dynamically provision and scale the resources needed to improve the end user experience.

Sequencers will submit the state transitions as they are built, but their verification happens in parallel on the Relay Chain, enabling the multi-core architecture for a single rollup. While the validation is a parallelizable process, the state transitions still need to form a chain. This indicates future performance improvements should be possible by improving the speed of sequencers.

Can rollups be the VMs of Web3?

In a Web2 cloud environment, the physical hardware resources are partitioned into virtual machines of arbitrary size. For example, one 64-core bare metal machine can be partitioned into 32 dual-core VMs, or 16 quad-core VMs. Then each of these VMs are used to host one or many applications running in parallel.

The introduction of Elastic Scaling enables Polkadot to partition its resources dynamically in a similar fashion. Each rollup becomes a VM with an arbitrary number of Polkadot cores.

A VM using a single core can consume two seconds of execution every six seconds.

A VM can use Elastic Scaling to increase its computational power three times.

Polkadot vs Web2 cloud - similar but different ?

Web2 clouds (like AWS, GCP, etc) have introduced a few key concepts that are widely understood and accepted:

Scalability: The ability of a system to handle growth, achieved through horizontal scaling (adding more machines) and vertical scaling (adding more power to an existing machine).
Compute: The ability to run applications and workloads on-demand, without the need to own the physical infrastructure.
Storage: The ability to store data in a reliable and durable way, with easy access from anywhere.
Networking: The ability to connect applications and services to each other with high performance and reliability.
Security: The ability to protect data and applications from unauthorized access, use, or disclosure.
Management: The ability to provision, monitor, and manage cloud resources.

By definition, the Web3 Cloud doesn’t stray away from these. Instead, it strongly aligns all of them with the Web3 mission. What we get is a fully permissionless, trustless, decentralized and transparent Cloud that empowers builders to securely own and manage compute, storage, bandwidth and networking resources.

Developers can leverage the technology stack of Polkadot to build unstoppable applications that are highly reliable, performant, and secure but at the same time can scale to meet the ever increasing demands of end users.

To gain a clearer understanding of how Web2 and Web3 cloud compare and contrast, the table below provides a side-by-side comparison:

Any Web2 Cloud	Polkadot
Compute	Centralized, Heterogenous CPUs	Decentralized, Standardized Cores
Storage	Disks	DA layer, rollup state
Networking	Trust-based, VPC, Subnets	Trustless networking, DMP, UMP, XCMP
Security	Trust-based security,IAM, security groups, etc	Trustless, crypto-economic security
Management	Permissioned, CloudFormation, Terraform, etc	Permissionless, RPC
Scalability	Horizontal and vertical scaling	Horizontal and vertical scaling

The world needs the Web3 cloud

Web3 has demonstrated solutions to solve problems that have been ignored or exacerbated for so long by Web2. It should not be viewed as a drop-in replacement, but rather as an augmentation.

Polkadot can host any kind of Web3 application and provides the scalability properties required for a seamless integration of Web2 and Web3 technologies into a unified end user experience. It is now possible to reduce or completely remove the trust users need to have in current centralized Web2 applications and systems.

When implemented properly, Web3 services tremendously improve resilience, censorship resistance, ownership, privacy, security and transparency on behalf of end users:

Financial services, offering lending, borrowing, and trading without intermediaries
Enhanced transparency and traceability by recording every step of a product's journey on an immutable ledger.
Representing ownership of art, collectibles, and other digital items.
Creating in-game economies that enables true ownership of digital assets by players.
Secure and self-sovereign digital identities, reducing fraud and enhancing privacy.
Transparent and tamper-proof voting, increasing trust and participation.
Streamlined property transactions, reduced paperwork, and increased transparency.

Finally, we can conclude that Polkadot isn’t just making Web3 better, it's changing the game entirely. With Elastic Scaling, rollups can adjust their throughput and latency on the fly, meaning we can finally have awesome performance without sacrificing security or decentralization.

The future of the internet is in the Web3 cloud, built on the foundation of less trust and more truth. Polkadot is where this future is being created right now. Join us and let’s build it together!

Github: https://github.com/paritytech/polkadot-sdk

Forum: https://forum.polkadot.network/

Wiki: https://wiki.polkadot.network/

X: https://x.com/Polkadot