Hyperledger Fabric: A Distributed Operating System for Permissioned Blockchains | Hacker News Reader

Hyperledger Fabric: A Distributed Operating System for Permissioned Blockchains

arxiv.org

113 pointszyngaro34 comments

I have done significant development work using hyperledger platform and aside from the fact that private blockchains are essentially pointless, this is (or at least recently was) pretty unusable, with very bad tooling and support.

I took a course on this whole thing on edx and I walked away with the same feeling.

Out of all the Ethereum contracts people have come up with so far, I still haven't found anything truly useful. Closest thing is just the ability to create "stock" like entities.

Fabric isn't Ethereum-based. It's the next most active smart contract platform, but way behind in developer interest.

Same. Admittedly, my work with it was with a fairly early version, but in spite of initial enthusiasm due to the fact that it is written in golang, I ended up very disappointed with a terrible developer experience. IBM's role in this mess didn't give me a whole lot of hope either.

Another comment from a technical perspective: Fabric doesn't even solve double spends. Say a logistics company uses Fabric to send diamonds to track (the prototypical blockchain business use case). Logistics Worker Alice goes to reserve Cargo Container C to ship her diamonds in. At the time, Logistics Worker Bob attempts to reserve Cargo Container C -- at the same time -- to ship his diamonds. This is a double spend (Company Cargo Container C -> Alice, Cargo Container C -> Bob).

All Fabric does is takes both transactions, issues the txns to the "Orderer" and then both transactions get put into a block and pushed out to the nodes. The nodes are then left to deal with the bloat of having two transaction.

If you prefer to have a immutable database that is anchored to a blockchain, rather then a permissioned blockchain, I suggest looking at the Politeia toolkit. Time stamping the sum content of any data repository makes it censorship resistant.

https://blog.decred.org/2017/10/25/Politeia/ https://github.com/decred/politeia

There is no point of having a private blockchain or distributed ledger. Full stop.

Nothing there was no possible before Bitcoin and what they are trying to achieve is some kind of API 2.0 where different companies can integrate within the same technology. This problem is political rather than technical and the blockchain is not a solution by itself.

I wouldn't go quite that far, I think in terms of simplifying your architecture and for providing sandboxing of third party code in a mission critical system there could be some modest benefits in a private blockchain.

However, these are not the reasons why people are usually interested in private blockchains, which as you say revolve mainly around political problems where blockchains can't help much.

I'd be horrified to discover an architecture that could be simplified by adding a blockchain to it.

I think some day this comment will be equivalent to "I'd be horrified to discover an architecture that could br simplified by adding a linked list to it".

I agree that for MOST use cases adding blockchain tech just makes things more complicated

This seems to be a popular opinion amongst a lot of ethereum developers I have met. I definitely don’t have as much experience in the field so honest question here. If a permissioned blockchain has nodes running across multiple organizations and third party auditing institutions, can that not solve real world problems?

Good question. There are more than one argument but I will use the security one: in a public blockchain, the blockchain is the truth and if a transaction happens it is immutable under certain assumptions. The connection between a public blockchain and the real world is through oracles which are the weakest chain since no matter the blockchain security it could trigger transactions beyond it. Now, what happens in a private blockchain? In this context there are a lot of outside the blockhain organizational systems that if they are being attacked all the blockchain security does not matter.

The text read like total mumbo jumbo. I have heard about hyperledger multiple times, but I still don't get what's the value prop. Maybe it is just a trap for banks so they waste their time and resources on that, while the real blockchain stuff moves forward elsewhere?

http://hyperledger-fabric.readthedocs.io/en/release-1.2/bloc... You can easily build a dapp! Not entirely without solidtity but you can use different projects because they are managed by a specification!!! https://www.hyperledger.org/projects/sawtooth Is my favorite! It is written in python! It is sponsored by intel and you can easly get started with aws images or docker images or you could even setup everything with Ubuntu!!!!!!

I can't tell whether the ludicrous ratio of exclamation marks to sentences (11:4 or 11:5 depending on how strictly you define sentences) is sarcasm or enthusiasm.

He has two other comments in this thread, albeit with slightly more judicious use of exclamation marks, yet I’m still just as confused as you.

Yeah, I studied blockchains before for a job (some business person saying: "hey we should really look into blockchain, and AI, and Big Data, and IoT, because errr...") and I always felt stupid and like I didn't understand it. Simple things like, why is this such a big deal? What problem does it actually solve? Why is it so amazing compared to a database? It's so, so difficult to actually read what these projects actually "do" sometimes. They often don't define what they think a blockchain actually is, or their threat model, or what their solution concretely does compared to regular databases. Once they got me to look at some project similar to this, and I was so confused as to why anyone would care I literally posted it on Hacker News because maybe they would care? But it got shot down. Which was actually quite a good moment for me because I realized that yes, I am right, this stuff is actually nonsense.

I eventually realized that, perhaps, they're just deliberately obtuse. Banks or other business people are obviously falling for it. Wait until these bankers learn what Git is.

Latest version https://arxiv.org/abs/1801.10228v2

For me the takeaway message is: "Fabric is the first blockchain system to support the execution of distributed applications written in standard programming languages"

Now you have to deal with all the issues of the traditional IT world (devops, cloudflare, vpns, containerization, cloud hosting configuration, load balancers, dns configuration, etc) AND all the issues of the new blockchain world (contract security, gas fee metering, block limits, replay attacks, etc)

...the saving grace of a "real" blockchain architecture is that at least most of the traditional IT issues go away.

> ...the saving grace of a "real" blockchain architecture is that at least most of the traditional IT issues go away.

All the work done related to blockchains I know of, involves lot of IT issues, both traditional ones and new ones.

Quite often work tends to become more complex as time passes on... In the past we had to deal only with fiat currency, now we have also to deal with cryptos, but fiat isn't going away as well.

Hyperledger Fabric: Not a Blockchain

Yes it is uses blockchain technology! http://hyperledger-fabric.readthedocs.io/en/release-1.2/bloc...

Rolls Royce uses "car technology" to build airplane engines, but that doesn't make an airplane a car.

So it's a just immutable, distributed linked list of signed nodes with a consensus protocol like two stage commit, paxos, isis and maybe overlay network and peer forwarding to bridge some gaps?

Sounds traditional distributed systems but now with signatures involved.

i can't find a reference to isis, google is pretty swamped with the term.

are you referring to the work by Ken Birman?

Yes, Birman and Joseph.

I see it as a way for technologists to satisfy non-tech people that "yes, we use blockchain now" when in reality this is just an immutable BFTDL with configurable consensus (perhaps similar to what aircraft have been using for many years).

Full, PoW/PoS blockchain like Bitcoin or Ethereum is hard to justify in an enterprise. Decentralization is rarely part of a viable business model.

BFTDL?

Byzantine Fault Tolerant Distributed Ledger

My advice is to say "yes, all our developers use blockchain-related technologies" and never let on that just means you use git.

Why is it called a distributed operating system? Not sure if it can help deal or orchestrate with distributed resources?

It uses blockchain technology like the EVM and adds a layer of abstraction so you can easily make applications untop of it. It is also funded by companies and is lead by the linux foundation! http://hyperledger-fabric.readthedocs.io/en/release-1.2/bloc...

Reply on news.ycombinator.com