Posted by in MultiChain, Private blockchains.

Permissioned networks built by MultiChain partners in live production

This is a write-up of a talk given at the Consensus 2019 conference. A video of the talk is also available.

In the four years since the first alpha version of MultiChain, hundreds (if not thousands) of proof-of-concept and pilot projects have been built by our partners on the platform. While many of the early ones were pointless blockchains, over time we have seen a consistent rise in the proportion of projects using the technology appropriately. Now we rarely hear about a blockchain-based application which lacks a good answer to the question: “Why not just use a regular database?” What a relief!

Proof-of-concepts and pilots are all well and good, but to my mind, the most important signal comes from solid enterprise blockchain projects that make it to live production. To be clear, this means networks containing multiple blockchain nodes belonging to multiple parties, where more than one of these parties is involved in generating real transactions and participating in the blockchain’s consensus algorithm. Without these characteristics, the blockchain is providing little or no value compared to a centralized database.

This article is a survey of ten of the most interesting permissioned blockchain applications built on MultiChain that are in production today. Each application will be described briefly, along with an explanation of why it made sense to use a blockchain and some numbers to give a sense of scale. Note that confidentiality agreements prevent us from revealing some of these projects’ details, but we’re telling you as much as we can. After reviewing the ten projects, I’ll finish with a list of five important lessons that I believe we can learn.

Ready? Then let’s begin…

Blockchain #1: SAP for Pharmaceuticals

Some medicines bought by large customers such as hospitals don’t end up being used, and are returned to wholesalers unopened for resale elsewhere. However this process brings a significant risk of counterfeiting, where the so-called “returns” have been faked somewhere along the way. To help combat this problem, every box of drugs can be shipped with a barcode label that identifies its contents and origin, with the barcode being recorded in a database for future verification. But who should be responsible for managing this critical database of drug shipment barcodes? In Europe, a centralized EU-level body was set up for this purpose, but there is no corresponding governmental entity in the USA.

To solve this dilemma, SAP built a blockchain-based solution on top of MultiChain, where multiple drug manufacturers and wholesalers have their own node, granting them direct access for reading and writing the chain. Each barcode is recorded as an item in a MultiChain data stream, allowing it to be looked up directly by scanning a printed label. The system is already running live and has been successfully tested to scale to 1.5 billion recorded barcodes and 30 million verifications per year.

Blockchain #2: TruBudget

When donor countries finance public projects in developing countries, it’s vital to keep track of important events in each project’s lifecycle, including tenders, contracts and disbursements. Both the donors and recipients want to maintain these records in a database for easy searching, but who should be in charge of that database? Neither side in the relationship is politically comfortable with ceding full control to the other, so this has often led to both parties maintaining their own records, and trying to keep them in sync. The picture is complicated further when there are multiple donor countries partnering together.

TruBudget is an open source application which uses a MultiChain blockchain to solve this dilemma. Each of the important stakeholders maintains its own node, writing important events to streams while sharing an identical picture of the project’s progress through their own front end. The system was commissioned by Germany’s Federal Ministry for Economic Cooperation and Development and developed by Accenture and KfW, Germany’s third largest bank. Two blockchains are now running in production for projects in Brazil and Burkina Faso respectively, with each expected to record up to 300 projects and 5,000 events per project.

Blockchain #3: Connected Health

In order to improve patient care and reduce bureaucracy, an Indian state government is implementing an electronic medical record system to enable information sharing between hospitals and other healthcare facilities in the state. When designing the system, two particular concerns arose. First, how can the records be secured against loss or tampering? Second, how do we ensure that the information is available locally in each city, in the event of a temporary loss of Internet connectivity?

These requirements were solved together by building the system on a blockchain rather than a centralized database. MultiChain streams are being used to store the medical records – currently with text only but with richer data such as images to be integrated later on. Participating cities will have their own nodes running locally, which take part in the consensus process. The system was built by RapidQube and is already in early production, with around 2 million records stored for over 50,000 people.

Blockchain #4: Collateralizing livestock

In many developing countries, farmers find it difficult to access affordable loans, even if they own valuable assets such as cattle that could serve as collateral. In order for a farmer’s cow to be used in this way, it must be identified and tagged, immunized against diseases and insured against potential mishaps. In addition, each cow can only be collateralized once. All this requires extensive data coordination between a country’s animal healthcare system, insurance companies and financial institutions, each of which has different incentives and governance structures.

FarmTrek is a blockchain-based solution developed by InfoCorp which enables this coordination to take place without being controlled by a central party. Each major stakeholder runs one or mode MultiChain nodes which work together to store and secure the data written to streams. Each cow is physically tagged with a tamper-proof NFC (near field communications) device, which connects to an Android mobile application used by the farmer to sign transactions and publish them to the blockchain. The project is now in live production in Myanmar and expected to scale to 100,000 farmers within two years, with an additional pilot in the works in Rwanda.

Blockchain #5: Tagcash KYC

As in many countries, when somebody opens a new bank account in the Philippines, the bank must perform rigorous KYC (know your customer) checks to verify the customer’s identity and residence. This costs time and money, meaning that banks and other financial service providers would benefit by sharing KYC information through a single database. Once built, this database can also form the basis of a credit scoring system, by adding information about customer loans and repayments (or failures thereof). Unfortunately, the Philippines has no centralized KYC and credit scoring mechanism, so this integration has been difficult to achieve.

In order to address this problem, Tagcash has created a blockchain-based KYC and credit scoring solution, using a network of nodes belonging to banks and smaller fintech companies. Some nodes have write privileges while others are permitted to read only. The information is stored within MultiChain streams, using a hash of each person’s name and birth date as a unique key for identifying their data. With the initial roll-out, around 100 records are being written per day, and this is expected to grow to 10,000/day over time.

Blockchain #6: Bureau Veritas Origin

With increasing awareness of food supply chain scandals, interest has grown in giving consumers greater transparency into how their food is sourced, processed, transported and stored. The goal is to create a comprehensive record of the steps involved in preparing an item for sale, and to enable consumers to access this information directly. To increase transparency and prevent tampering or corruption, it is preferable not to centralize control of this database at any individual company or location.

Bureau Veritas, a global company focused on testing and certification, has partnered with Atos Worldline to develop Origin, a blockchain-based food traceability solution. Nodes are run by multiple companies within the food supply chain, with data written to streams in a proprietary binary format. The finished products are labelled with QR codes, which consumers can scan in order to browse a web-based summary. With the initial roll-out, up to 100 records are being written per day.

(To avoid a common fallacy, it should be emphasized that the sources of data still need to be trusted when using a blockchain. The chain only improves the security of that data once it is stored.)

Blockchain #7: ILSBlockchain

An insurance linked security (ILS) is a bond which enables an insurance policy to be covered collectively by a group of investors. For example, the owners of a ship could pay a premium to the holders of an ILS, but if catastrophe strikes and the ship sinks, those holders lose some or all of their original investment. As with any financial asset, digitizing ILS ownership allows sales and transfers to take place efficiently. This is traditionally achieved using a custodian such as Euroclear, but the cost can be prohibitive for smaller insurance policies in the $10-20 million value range.

This problem was solved by Solidum Partners who issue and track ILS bonds on a MultiChain blockchain, removing the need for a highly regulated centralized custodian. Each bond is issued as a MultiChain asset, with participants transferring and exchanging these assets on a peer-to-peer basis. Nodes are run by the bond trustee, investors and reinsurers, with the consensus generated by a small group of senior participants. So far, four bonds have been issued on the blockchain, for over $50 million in total value.

Blockchain #8: Air Quality Chain

When it comes to collecting environmental data, three particular challenges need to be addressed. First, each type of data is generated in a different location, due to the need for specialized equipment. Second, the data must be stored safely and reliably for the very long term, to enable trends and changes to be analyzed. And third, different types of data may need to be cross-referenced in real time, to create a full picture of anomalies at the moment they occur.

These requirements can be addressed together by using blockchain. The Air Quality Chain project, implemented by Baumann, aggregates data on levels of ozone, radiation and air quality in Austria, using a network of nodes which collect data from multiple sources. Raw data is written directly to MultiChain streams, and so automatically replicated to all of the nodes in the network, which collectively ensure that it cannot be lost or modified. The system is running in production and collecting 2.7 million records annually, containing around 4 GB of raw data.

Blockchain #9: Deepshore Archive

Metro Group, the world’s fourth largest retailer, is required to archive all point-of-sale data for internal and external auditing purposes. Whereas Metro used to rely on a single vendor for this purpose, they recently migrated to a more flexible model, where the data can be redundantly stored on a number of different cloud providers. This gives them much greater freedom and the ongoing ability to negotiate over pricing.

However, this fragmentation presents a challenge in ensuring that all of the data is stored correctly and cannot be changed. To solve this problem, Metro have deployed a blockchain-based system, built by Deepshore, where a hash and some other metadata for each data set is stored in MultiChain streams for verification purposes. Multiple nodes are running in different subsidiaries and locations within the Metro Group, so even though this is an “internal blockchain”, control is effectively decentralized within a vast organization. The system is already running live and notarizing approximately 9 million datasets per day.

Blockchain #10: Fantastec SWAP

Growing up in the 1980s in the UK, collecting football stickers was hugely popular. We spent our pocket money on packets of random stickers, containing the faces of players, team photos and badges, and swapped obsessively with each other in at attempt to complete each year’s album. Fantastec has now developed a digital equivalent, where users download the SWAP app and purchase limited edition “cards”, complete with player videos and interactive statistics. Naturally, this application needs some database to keep track of card ownership, but it wasn’t clear where this database should be hosted. On the one hand, each participating football club should maintain its own database, to guarantee the authenticity and rarity of its issued cards. On the other hand, much of the product’s value derives from the ability to swap cards that were issued by different clubs.

This dilemma was solved by building the system on a blockchain, where each club has its own node that issues its digital collectibles as MultiChain assets, all of which are tracked together on a chain that is managed by consensus. The system, which makes extensive use of MultiChain’s built-in atomic exchange functionality, was built by Fantastec with assistance from partners such as PricewaterhouseCoopers. SWAP was recently launched with three big-name partners: Real Madrid, Arsenal and Borussia Dortmund. After less than 3 months it has grown to 15,000 users with over 250,000 collectibles issued.

Lessons learned

Now that we’ve reviewed ten of the most interesting MultiChain-based networks in production, what can we learn from this group as a whole? What differentiates these projects from the hundreds and thousands of proofs-of-concept and pilots that never made it to the next stage?

Lesson #1: Focus on new applications

While there has been much talk about blockchains as an upgrade for existing systems, for now at least, we’re primarily seeing them deployed in new applications. I can think of two related reasons why this might be.

First, blockchains are still a new technology, and are perceived as more risky than centralized databases. This uncertainty can be tolerated when building new applications, which inevitably comes with some risk of failure. However, it makes blockchains less attractive for replacing something that is already known to work.

Second, any running centralized application must already have a trusted intermediary, who has presumably proven their reliability over time. While moving to a decentralized architecture might save money in bypassing this intermediary, this has to be weighed against the cost and risk of rebuilding the system from the ground up.

Lesson #2: Find a strong motive

Every application implemented on a blockchain must answer a crucial question: Why use a blockchain instead of a centralized database or file server? Blockchains will always be slower, less scalable and more complex than centralized systems, as a result of their fundamental design.

So if you have a suitable trusted intermediary who can host an application centrally, you should use it! The only reason to use a blockchain is if there is a strong motive to avoid this kind of centralization. In practice we see four main types of motive appear:

  1. Commercial concerns. The participants in a network do not want to grant too much power to a competitor or some other central body, who could charge a lot for the service.
  2. Regulatory requirements. Some regulation prevents the deployment of a centralized system, or would render it too expensive in terms of compliance.
  3. Political risks. There is no place where the database could be hosted that would be politically acceptable to all of its users.
  4. Secure replication. Multiple copies of the data need to be stored for redundancy, so using a blockchain provides the additional benefit of proven synchronization and tamper resistance.

Lesson #3: Think about data in general

Early discussions about enterprise blockchains were triggered by the rise of cryptocurrencies, in which the blockchain allows users to directly hold and transfer a virtual asset while preventing double spends. While some of the production networks we described (#7, #10) are using MultiChain in this way, the majority are doing something fundamentally different – building a decentralized architecture for storing and securing data.

Any database or file system, whether it is holding structured or unstructured data, could be implemented on a blockchain. Each piece of data can be stored in full on the chain, or notarized as a short on-chain hash (fingerprint) which serves to verify the data which is delivered off-chain. Unlike in asset use cases, there is no notion of ownership changing over time. The blockchain’s sole purpose is to enable some information to be stored and secured by a group, without relying on a central party.

In data-driven applications, “smart contracts” are the wrong transactional model, since they require every piece of data to be represented as a message sent to a contract, rather than being validated and then directly embedded (or hashed) in the chain. The central issue is the scale and speed with which information can be stored, indexed and retrieved.

Lesson #4: Look beyond “transformation”

For too long, the enterprise blockchain narrative has focused on buzzwords like “revolution” and “transformation”. But in reality, if we look at those blockchain projects that actually make it to production, only a few are doing things that would be impossible to achieve using more traditional technologies such as centralized databases, replication and point-to-point messaging. So what exactly is being transformed?

In most cases, a blockchain is being used simply because it’s the most appropriate and convenient tool for the job. It enables a new application to be easily built on top of a unified data store, while avoiding some concern about that store being centrally controlled. The blockchain provides additional robustness and tamper resistance, whose value outweighs the complexity and cost of running multiple nodes. While this all might seem rather unromantic, since when has enterprise IT been anything else?

But there is an additional, more subtle, part of the story. In rare cases we see projects being built on a blockchain, where there is no immediate justification for that choice. It turns out that the application’s users are happy for it to start out centralized, but want to keep their options open for the future. Using a blockchain (even with one node!) rather than a database allows the intermediary to be swapped or removed just by adding or removing nodes and changing some permissions. All this can happen with zero downtime and without touching the application’s code.

Lesson #5: Be very patient

With all of the noise surrounding blockchains, it’s easy to forget just how new this industry is. MultiChain, along with most other enterprise blockchain platforms, only reached a version 1.0 release in mid-to-late 2017 (it’s now at version 2.0.2). Since it’s quite common for enterprise IT projects, whether based on blockchains or not, to take two years from initiation to go live, it’s no surprise that the number of real blockchain networks in production is still rather small.

Indeed, two particular phenomena demonstrate just how early things are. First, we often find our partners performing the most basic tests on MultiChain just to convince themselves that it actually works! Second, we see some participants in production blockchain networks lacking the confidence to take responsibility for their own node, instead relying on some third party to host it on their behalf.

So as with any other new enterprise technology, people working in the blockchain space should hunker down for the very long term. I expect it will take another ten years before blockchains are commonly considered as an alternative for information system architectures, and another ten after that before they reach their full potential. By then, bandwidth, storage and cryptography will be so cheap and fast that it may seem quaint (if not ridiculous) for shared applications to store their data in only one place.

 

Please post any comments on LinkedIn.