Arweave 2.6 Major Upgrade – More data, Less energy
Arweave in Asia 2022 is the first offline Asia-based Arweave community event, being hosted by Arweave and everFinance, and set to bring together 200+ Arweavers, including project founders, investors, developers, and the Arweave core team, to discuss the future of Arweave and Web3. And it is taking place as we speak.
Many important updates are sure to come from Arweave utilsing projects as well as the Arweave core team. Like yesterday, when Arweave founder, Sam Williams, took to the stage to present the new iteration of Arweave. Arweave 2.6!
I’m really excited to speak to you today because we’re this is essentially a culmination of about 12 to 18 months of work across three or four different work streams in the Arweave core team. And we saw that this event was coming together in Asia, and we were all gathering, and we thought, well, what better time to launch all of this than when we’re all in one place in physical space. So thank you again to the ever finance team for putting this all together-
Sam Williams, Founder, Arweave
Let’s take a closer look!
Arweave 2.6
The Arweave 2.6 upgrade took the core Arweave team 18 months to research and develop. It is a major upgrade to the protocol, so major, in fact, that initially, the Arweave team contemplated calling it Arweave 3.0.
As mentioned in the presentation, Arweave 2.6 is all about “more data, less energy”. Essentially, Arweave will now be much more efficient in how it enforces miners’ data storing while wasting less energy.
Overall, Arweave 2.6 achieves the following objectives:
- Lowers the storage acquisition cost for the network by encouraging miners to use cheaper drives to store data, rather than optimizing for drive speed.
- Lessens energy wastage in the network by directing a greater proportion of mining expenditure to the useful storage of data.
- Incentivizes miners to self-organize into full replicas of the Arweave dataset, allowing for faster routing to data and an even more ‘flat’ distribution of data replication.
- Allows for better dynamic price Per GB/Hour storage cost estimation in the network.
So what does all of this mean in practice?
With Arweave 2.6, the number of replicas per piece of data will increase from 430 replicas (as on version 2.5) to 9900, making it one of the most replicated data sets in the world, all while remaining democratised, and allowing anyone to put their data onto the network for a low cost per megabyte.
That makes the archive much more efficient and effective at achieving its long-term mission while procuring storage at a cheaper price for miners, all while merging Proof-of-Work (PoW) style security with Proof-of-Stake (PoS) style efficiency.
Secure
In his presentation, Sam mentions how the recent Ethereum Merge was great for reducing the amount of energy the network was burning but came with trade-offs. One such trade-off is long-range attacks, where a certain majority of voting power in the network can retrospectively write a new chain of sufficient length that claims an entirely different history.
And this is like not great. There are a lot of problems with this in Proof-of-Stake.
But the Arweave team has put their security budget towards making 1000s of replicas of each data set while making the process much more efficient.
So how does it work?
A hash chain enforces a limiter, which allows miners to perform challenges on the network at a maximum speed of 200 megabytes per second per replica of the data set.
Without being able to break the limit, miners are incentivised to make more replicas of the data, effectively filling more hard drives worldwide with copies of a user’s data set, further increasing its longevity.
In the process of doing this, Arweave 2.6 solves another fundamental problem all distributed data storage systems (like IPFS) face, in which building a system that gets from an address (like an identifier) to a piece of data, the dataset needs to first be organised. But dataset organising cannot be done with decentralised networks because data comes and goes as the nodes come and go.
As mentioned in the presentation, one of the best solutions till today has been Kademlia, an algorithm and early file sharing system created in 2002. Kademlia is widely used by distributed storage systems – like IPFS.
While building their new mining algorithm, Arweave created a miner incentive whereby it makes more sense for the miners to work together in groups to form full replicas of any given data set, which makes data retrieval more efficient. From a miner’s perspective, this makes no difference, as long as they know where the queried data is.
How is this achieved?
Every miner on the network gets one challenge at every time step, and to unlock a second challenge, they have to know how to route to the queried data. A challenge is when retrieving a piece of data means having to do some mathematical puzzle on it. So, if miners don’t know where the data is, they can’t mine as efficiently. This incentivises every miner to know where every other piece of data is at any given time. This ultimately means that when querying data, you are at most one hop away from a node that contains it.
Speed
Kademlia-based systems can take much longer to retrieve data (over seven minutes sometimes, as mentioned in the presentation). Arweave 2.6 can do this in roughly 400 milliseconds, while effectively retrieving data from anywhere in the world in a decentralised fashion.
Dynamic fee markets
Arweave 2.6 also enables dynamic fee markets. Arweave 2.5 has a necessary consensus step once every six months where miners need to agree on what to set the base storage price at.
But, with 2.6, the speed of data access allows calculating the number of replicas of a data set more accurately, in turn allowing the Arweave team to see how many tokens are being emitted per returned replica. This is an Oracle-less price discovery mechanism that incentivises miners to give the lowest possible price. That means miners are fighting each other to minimise the price users have to pay for storage.
A necessary component of this system is a Verifiable Delay Function (VDF), where everybody is limited to the same speed with no regard to (higher) computing power.
Scaling
Arweave scales arbitrarily, thanks to the fundamental way it works, due to permanent information storage needing less metadata to be stored about a new piece of information that has been added to the network. Subsequently, the network’s state is not bloated.
Energy efficient
Arweave uses about 800 SHA-256 hashes per second per node. For context, a MacBook uses about eight to ten million per second.
A strong foundation for innovation
A very crucial take from Arweave 2.6, is that it makes the Arweave network more complete than ever. With version 2.6, Arweave becomes a system that will require few future changes, as its features and scalability will allow it to persist through time.
Being complete is key to a protocol’s growth, increased use, and in turn value increase over time. The less it will require to change over time, the more it will be trusted and used to build on, as the risk of unwelcome upgrades is highly unlikely.
As mentioned by Sam during his presentation:
The core promise of a protocol rather than a company is we are not going to change from underneath your feet. You can trust to build on us. But last week, when the Ethereum Merge went through, $8 billion of GPUs were rendered essentially like useless. I mean, you can still use the GPUs for stuff but they were purchased for mining. And so that promise that the protocol made to those miners that said, look, you can build a business around this, is essentially broken. And I don’t say this to you know, trash talk Ethereum. I love Ethereum. But I’m just being realistic.
Going on to say:
So the Arweave protocol itself is essentially becoming stable. At this point. […] The value of protocols tends to get unlocked after they stop changing.
Conclusion
Arweave 2.6 brings new under-the-hood features that benefit everyone utilising the network (users, devs, token holders, etc.). And all these features combined make for a more “complete” – aka stable – protocol.
And at the end of the day, stable protocols come with a sense of trustworthiness, which leads to an increase in users and developers.
The success journey for Arweave, it seems, has only just begun!
And if you are a developer, you can read more technical details and specs here.
Join our
