Arweave versus IPFS & Filecoin

Introduction

Technology has been improving at an exponential rate and it has become quite difficult keeping up with all these new developments. In order to not be overwhelmed by these advancements, we have a predisposition to put them into the same category.

This can be seen in the data storage industry where projects such as Arweave, Filecoin (built on IPFS), Sia, Storj and the like are treated as the same: data storage through blockchain. However, they are solving completely different problems.

In this article, we will discuss what IPFS and Arweave are, their similarities and differences and what they ultimately aim to accomplish.

What is Arweave?

Arweave is a decentralized blockchain project that goes a step further: it ensures data permanence. Users pay a one-time upfront fee in order to permanently store their data on Arweave. Once uploaded, it will never be deleted and there are no hidden charges for when the user would like to retrieve their data.

This is made possible through Arweave’s unique blockweave and storage endowment.

The blockweave is Arweave’s blockchain-like data structure that contains all the data that is stored on the network. Arweave uses a consensus mechanism called Succinct Random Proofs of Access (SPoRA), an improvement on the previous Proof of Access (PoA) mechanism, to verify whether miners are correctly storing the blockweave. To understand how SPoRA works, it is first important to understand the structure of the blockweave. In the blockweave, each block of information can be linked to the previous and the following block, as well as a random chunk of data. Hence the name: Blockweave.

In order to be able to add more data to the blockweave, it is required for miners to show that they have access to a randomly selected block. Therefore, the SPoRA mechanism incentivizes miners to store the entire blockweave in order to mine further blocks and earn rewards.

Paying to store data for 200 years seems like it would be expensive but in reality it’s actually quite inexpensive. At the time of writing, storing a gigabyte of data for 200 years only costs ~$7.4. It is also important to understand how the data storage endowment functions and enables users to only pay a one-time fee when uploading their data.

The protocol charges users to store a minimum of 15 replicas for 200 years with very pessimistic assumptions about the future. On average, the cost of storage has declined by 30.5% per year over the past 50 years. To be on the safe side, Arweave makes an extremely cautious assumption that the decline will be 0.5% each year when calculating the cost of uploading data. Any time the cost of storage declines > 0.5% a year, it extends out the lifetime that the protocol can pay to store your data.

Currently, it is estimated that there are more than 200+ replicas of the blockweave across Arweave’s nodes. The endowment will increase block rewards if the number of replicas begins to decline and approaches 15. This would make mining more profitable, encouraging existing miners to add more replicas or entice new miners to join the network. In fact, since genesis no tokens have left the endowment.

What is IPFS?

The InterPlanetary File System (IPFS) developed by Protocol Labs is, as stated on their website, “a distributed system for storing and accessing files, websites, applications and data”. It aims to make the distributed web more efficient, decentralized and resilient. In order to understand how it does so, it is vital to first understand how the traditional web works.

When you search for anything on the traditional web, it uses location addressing to bring up the results. Your computer will essentially request the information from the computer(s) that is hosting that information. However, this method is slow, inefficient, centralized and can be censored quite easily.

To overcome these problems, IPFS uses content addressing where the file is identified based on its contents. Through IPFS, your computer requests this information from many other computers all around the globe. Instead of requesting it from a centralized entity, you are now able to access that information from anyone that hosts it. Additionally, through IPFS your computer is also used to distribute information to other users of IPFS that request it. Data is available on the network as long as there is at least one individual storing the data. 

Users of IPFS must be aware of garbage collection – a mechanism used by the protocol to free disk space on your IPFS node by deleting data that it thinks is no longer needed. To avoid unwanted deletion, data must be 'pinned' in order for the system to not clear it during garbage collection. Therefore, users usually use tools like the IPFS desktop app to create their own IPFS node and pin their data for it to persist on the network. Still, your own device and node may go offline making data inaccessible, or get damaged resulting in your data being lost.

In order to make data on the network more durable and resilient, users can utilize third-party pinning services such as Infura or Pinata. These services will pin data for you across their own nodes. This comes with its own set of challenges, as each service has different policies for the data to remain pinned on IPFS. Overall, these services prioritise a user-friendly service which comes at a relatively expensive cost, for example, Pinata charges $100 ever month just for 250 GB storage. Further to this, there is no guarantee that these services will continue to maintain their services, as the IPFS documentation itself warns. Finally, persistence of data is dependent on the continued payment of these services.

What is Filecoin?

As there is no incentive on IPFS to hold the data of others, which is why Protocol Labs also built Filecoin, a complementary protocol, on IPFS.

Filecoin has been described as the Airbnb of data storage. It enables users to essentially rent out the extra space on their hard drives and creates a decentralized storage marketplace. It's a peer-to-peer network designed with cryptographic measures and economic incentives to provide for long-term storage of data.

Clients make deals with storage providers (SPs) on the amount of data to be stored, the number of replicas to be made, the duration, and the price for storage and retrieval. The network then verifies whether these SPs are correctly storing the data and allows users to release payments for the duration of the deal. SPs that do not uphold their end of the deal by correctly storing the client’s data are penalized in several ways: a loss of collateral they must provide in FIL to agree a deal, a loss of block rewards and reputation. It should also be noted that the minimum file size accepted by a Filecoin storage provider can often be several GiB.

To retrieve data, clients must pay a fee, agreed in the deal. The retrieval process is not normally as quick as a pinning service, so clients can pay a higher fee to utilize specialized nodes that allow quicker retrieval of data. They can normally retrieve data directly from SPs, but can also retrieve data from specific retrieval providers (RPs). This has a benefit of potentially optimizing access: popular files will get picked up by other RPs, which bring the files closer to the user. Data moves to areas of demand.

Filecoin uses Proof of Replication (PoRep) and Proof of Spacetime (PoSt). In order to verify that storage miners are correctly storing the client’s data, they are required to prove that the data is present in their physical storage (PoRep). As time passes, miners are randomly selected to provide proof. In turn, this information is stored in a PoSt proof which shows that the storage miners have been correctly storing the data over time.

It's important to understand that data can’t be unpinned until the end of the storage deal. Even then you can’t ensure data is ever deleted on IPFS at that moment. As a decentralised peer-to-peer network, your data could be spread across many other providers.

Similarities and Differences

Contract-based vs. permanent storage

Similar to centralized cloud services, Filecoin’s economic model for storage is contract-based. Essentially there’s an ongoing contract and payment agreement, which once broken will mean the service will stop storing your data.

As discussed in previous sections, Arweave offers permanent storage – a completely new economic model. This was once a clear differentiator between Arweave and all other storage providers, but recently the Filecoin Virtual Machine (FVM), a runtime environment for smart contracts on the network, introduced their own ‘perpetual storage’.

In reality, Filecoin’s version of perpetual storage is still contract based. With this new offering, they are allowing users of the network to set an indefinite expiration date on the storage deal. But as described in their documentation, “As long as the uploader still has funds (FIL) in the contract account, the contract will keep incentivizing storage providers to create deals to meet the goal of replication.

The storage is still contract-based and dependent on the uploader maintaining the funds. Therefore, permanence cannot be guaranteed and subjects the uploader to an unknown total cost of ownership, as funds to maintain the contract indefinitely cannot be known.

How much would it cost to store data for 200 years on Filecoin? Unlike Arweave, this cannot be known, especially considering their token is currently being used to massively subsidize storage on the network.

Tokenomics

Both projects have a native token which functions as a blockchain’s currency for transaction fees and rewards. Filecoin has the FIL token while Arweave has the AR token. At the time of writing, FIL is valued at ~$6.3 while AR is valued at ~$9.2.

Just looking at the value of a token provides an incomplete picture. That’s why it’s crucial to understand the tokenomics of each protocol. Tokenomics essentially determine the economic value of each project based on the supply and demand of the protocol’s token.

When it comes to the supply side, it’s important to consider three factors: circulating supply, maximum supply and whether the protocol has a burn mechanism.

Currently, Filecoin has 454 million FIL in circulation with a maximum supply of 2 billion FIL that can be issued. The rest of these tokens will be minted over the next 30 years according to the Filecoin Protocol.

Additionally, Filecoin also has a burn mechanism in which a portion of tokens are burnt through transaction fees on the network or through penalties. To date, around 36 million FIL have been burnt.

Arweave has around 50 million AR tokens in circulation with a maximum supply of 66 million AR that can be issued. Although Arweave does not have a built-in burn mechanism, its storage endowment has a similar effect. As the tokens in the endowment will be gradually unlocked over a period of at least 200 years, we can safely assume that these tokens are currently out of the circulating supply.

From these numbers, we can clearly see two things. Firstly, Filecoin’s maximum supply is 30x Arweave’s maximum supply. Secondly, Arweave already has around 75% of its supply in circulation while Filecoin only has 22.5% of its supply in circulation. These two distinctions imply that Filecoin’s supply will cause a lot more downward pressure on its price in comparison to Arweave.

Furthermore, Filecoin has heavily subsidized the cost of storage on their network since their $200m ICO, making their current pricing model unsustainable in the long run. Increases in storage costs are inevitable for Filecoin, which will lower demand on the network, whereas Arweave’s prices should decrease over time as the actual cost of storage decreases each year.

Storage cost

Another difference between the two is the cost of using the protocol. On Filecoin, data can be stored by setting up deals with a storage provider or by using a 3rd party pinning service such as Infura or Pinata.

When using a storage provider, the cost of storage Filecoin is estimated to be $0.002 per GB per year. If you’d like to use a pinning service such as Pinata to store your data on IPFS, it will cost $4.8 per GB per year according to their current subscription plans.

On the other hand, when you store data on Arweave, you pay a one-time upfront fee. Currently it costs ~$7.4 per GB. This comes to around $0.037 per GB per year assuming the data is stored 200 years.

Once your data is successfully uploaded on Arweave, you can access it at any time without any additional fees. However, on IPFS and Filecoin, there are retrieval costs for when you would like to access your data.

Filecoin appears to be focused on competing with the big cloud storage providers – AWS, Azure, Google – on price, but based on hardware configuration and tokenomics described above they will not be able to outcompete the traditional storage providers over the long term.

Decentralization and data stored

Filecoin currently has 3,541 active storage providers. If you’d like to visualize it, you can view their map here. Arweave currently has 279 estimated nodes. Arweave’s blockweave is currently at 126.7 TiB, whereas 791 PiB worth of data is currently stored on the Filecoin Network. In order to understand how much data that is, 791 PiB is 809,984 TiB. That’s around 6,393 times more data than the blockweave.

Judging by these two metrics, it would seem that Filecoin is trusted to store way more data and its level of decentralization makes it more secure. However, the picture is not so clear as it is important to understand where and how the data is stored, as well as the type of data stored.

If you were to store your data on Filecoin, it would be stored on the node(s) of your storage provider for the duration of the deal. The average deal duration is 506 days which means that your data would be stored on that node for around a year and a half.

Now if you were to upload your data on Arweave, it would permanently be a part of the blockweave. Additionally, your data will be replicated across the nodes storing a full replica of the blockweave. It’s possible to determine the number of complete replicas of Arweave, as we know the hashes per second that a complete replica of Arweave data will produce. It’s estimated that there are currently 200+ replicas of the 126.7 TiB blockweave.

With Arweave, there is no single point of failure through which your data can be lost. However, when you store on Filecoin, your data can still be lost in the event of storage provider failures.

On Arweave, there is a built in incentive layer and each node is encouraged to store your data, so it can be considered to be more resilient. Pricing is determined so there will always be incentives for miners to store at least 15 replicas of the full blockweave, a number that is determined to balance cost efficiency along with extreme statistical improbability that all replicas are destroyed at once due to disk failure.

Finally, it’s important to note that as Arweave is a permanent data storage solution, the data stored is for the most part very valuable. So while Arweave stores less data that doesn’t imply it is less trusted, but rather that Filecoin makes more sense for bulk storage of low value data that is not likely to be needed in the long term.

Ecosystems

Moving onto ecosystems, Filecoin has more projects developing on it in comparison to Arweave. Currently, there are 400+ projects in the Filecoin Ecosystem. This includes projects such as Chainlink, Decrypt, ENS, Internet Archive, Lighthouse Storage, Livepeer, Ocean Protocol and The Graph.

Presently, there are 100+ projects utilizing the Arweave Ecosystem. This includes projects such as Akord, Alex, Ardrive, Bundlr, Community Labs, everVision, Forever Stories, Mirror, Permaswap, Stamp Protocol, Verto, Weve and Warp Contracts.

Interestingly, both Filecoin and Arweave released major upgrades last month giving a boost to their respective ecosystems. Filecoin released its Filecoin Virtual Machine and Arweave transitioned to Arweave 2.6.

Filecoin’s Virtual Machine has many implications for the protocol. It allows for ‘perpetual storage’ (as discussed in the ‘Contract-based vs. permanent storage’ section, above), replication of data and ‘Data DAOs’.

While the move to Arweave 2.6 does not really affect what can be built on the protocol, it dramatically improves the network in many ways.

This upgrade decreases storage costs for miners as they are incentivized to use more drives for storage rather than prioritizing drive speed. It’s also estimated to have a ~23.16% efficiency improvement, and founder Sam Williams stated the upgrade probably makes Arweave the most replicated data set in the world.

Conclusion: the next frontier

Arweave, IFPS and Filecoin represent powerful decentralized alternatives to traditional centralized services. They are the next frontier of data storage the decentralized web. But which is better? That may depend on what kind of data you are storing and your use case. Still, there are some clear conclusions that we can draw.

All three projects are concerned with the long-term preservation of data. But in the case of IPFS, even from their own documentation, the advice is to use Filecoin if you are serious about resilient and durable long-term storage.   

On the surface, it appears Filecoin is attempting to provide a cheaper decentralized alternative that competes with the likes of AWS, Google and Dropbox. Its architecture eliminates single-point-of-failure risks and provides users with censorship resistance, yet data is not truly permanent. Filecoin is ultimately a contract-based form of storage that requires ongoing effort and resources to continually store data.

A serious limitation, often not discussed, is while IPFS and Filecoin can not claim true permanence, they also can't claim access to a handy delete option . You can unpin data and end a storage deal, but you'll never be able to control if other providers and nodes on the network are still storing and serving your data, especially if the file becomes popular. 

With this in mind, we see IPFS and Filecoin fall into somewhat of a grey area – you can't fully trust it to be permanent, and you can't fully trust it to be temporal.

IPFS was ultimately conceived to decentralize the web and address issues of centralization around security, efficiency and link rot. It's purpose has stretched out along with Filecoin to address long-term storage of data. Arweave on the other hand, was designed from the outset to address permanent data storage, as well as the same issues of web decentralization, ushering in what it refers to as the Permaweb. The need for pinning services and the additional Filecoin protocol, provides an indication that IPFS and Filecoin have been plugging gaps to tackle persistent data storage. 

With that said, IPFS and Filecoin can present an interesting choice when data storage has a known lifecycle. For example, when data must be stored for 10 years for compliance reasons. It's heavily subsidised prices make it an attractive choice against centralised cloud services for this use case. However, if temporary storage of data is required then, equally, compliance reasons may rule out IPFS and Filecoin – data deletion cannot be guaranteed, complicating matters in the face of regulations like GDPR and CCPA. 

Ultimately, if there are no regulatory concerns and true permanence with the utmost resilience and durability is required, Arweave is the logical choice.