VAKT is a blockchain startup that processes post-trade transaction for the oil market, initially Brent crude. It’s owned by a combination of banks, oil majors and oil traders including Shell, BP, Chevron, Total, Reliance Industries, Equinor, Gunvor, Mercuria and Koch Industries. Perhaps with that combination of shareholders, it is not surprising that it processes more than 90% of post-trade transactions for Brent crude.
Not only did they successfully execute, but they also did it quickly. So when the company lays out how it got there, it’s worth paying attention. In the interests of sharing, last week, along with technology partner ThoughtWorks, VAKT published a document “From PoC to Production: Implementing an enterprise blockchain solution.”
“The term ‘blockchain’ is becoming a mantra for change, even where the problem space may not need all of the characteristics,” the report notes. The features include persistence or redundancy; tamper resistance; provenance about where the data came from; shared data; the ability to include programming logic; and decentralisation, so a central authority does not control it.
There’s a couple of significant differences between public and private or enterprise blockchains. One is a far greater requirement for confidentiality for enterprises. Another is public blockchains are trustless. In contrast, enterprise blockchains have controls over who can host and access data.
It’s important to note that VAKT chose to use Quorum, the enterprise blockchain created by JP Morgan, which is based on Ethereum. This has very different characteristics to something like R3’s Corda, which is not strictly a blockchain designed more for workflow. We’ll return to that.
The report highlights the step change between a Proof of Concept (PoC) and production. Whereas the PoC is about innovation and experimentation, getting to production is different. Instead, it’s about usability, operability, confidentiality, integrity and availability.
The report refers to the “3 am test”. When considering any implementation issues, “is there a possibility that because of this decision, at some point in the operation of the system the CTO will be woken up at 3 am to deal with a production issue?”. If the answer is yes, then another solution is required.
Enterprise blockchain isn’t there yet
VAKT and ThoughtWorks contend that no enterprise blockchain solution meets the 3 am test, yet.
They contrast different approaches as top-down (starting from enterprise) versus bottoms up (starting from a public blockchain type of design). In terms of well-known implementations, it classifies Corda as top-down. With Corda, the data is completely segregated so that only parties to transactions store the data. It’s also not strictly a blockchain. The latter has a bit less redundancy, but also may encourage greater storage of information on the ledger.
All the other well-known solutions – Quorum, Pantheon (ConesenSys’s coming Quorum competitor), Hyperledger Fabric and Hyperledger Sawtooth – adopt more typical blockchain architectures.
And by adopting typical blockchain features, they also inherit the disadvantages such as the speed of throughput, limited scalability, and lack of privacy. While many of the issues are being addressed, they aren’t there yet.
Plus, all blockchain and DLT solutions are immature. The paper gives examples of missing features in some cases, such as identity and document storage.
We’re reminded of the Betamax, VHS example that the best technology won’t necessarily win. And the big three will prove attractive purely because of market penetration. VAKT and ThoughtWorks say that as a result of the weaknesses in current offerings, there’s a need to bolt together blockchains with other existing technologies.
VAKT and ThoughtWorks take a conservative approach: minimize use of the ledger and exposure to innovation. For the latter, they are saying combine blockchain with tried and tested technology wherever possible. By doing so, the high risk of the blockchain aspect is partially mitigated by other areas having low technology risk.
They emphasize the need to treat parts of the implementation as components, so that way, there’s isn’t technology lock-in. So if an innovation usurps some conventional software, it’s easier to replace.
“80 per cent of your effort will be ‘business as usual’ enterprise development, with a decentralised twist,” the report states.
The two companies emphasize that one size does not fit all, particularly in an enterprise blockchain context. Off-chain processing can be used to help the blockchain scale. They recommend minimizing the use of smart contracts. That’s because of versioning issues (all nodes have to upgrade at the same time), security concerns and relative complexity in implementing simple logic.
The report concludes with a list of potential features to use to assess the different blockchain solutions. These are:
- Transaction privacy
- Ledger robustness
- Document sharing
- Reference data
- Ledger index (for querying and searching)
- Robustness (availability, operability, deployment, tooling)
- Which logic in smart contracts?
- What to decentralise?
- Consensus model
- Immutability and tamper evidence