With so much speculation about blockchain technology and its potential impacts on existing business models, it is a time to be realistic.  In Gartner's most recent Hype Cycle for Emerging Technologies report, blockchain is said to have entered the trough of disillusionment phase, the third phase of the company's hype cycle metric, as shown in the following screenshot:

Between Q4 2017 and Q1 2018, the price of 1 BTC reached above 19K and then quickly dropped down 10k in a matter of days. It is 3.4K on December 11, 2018. This eye-popping roller-coaster in BTC price has led to the cooling down of frantic speculations on cryptocurrencies and consequently a reduction in investment in cryptocurrency projects. In addition, the limitations of the technology are also a factor, making the technology unsuitable for certain use cases. A well known issue is the low rate of transactions per second (TPS) inherited by most blockchain platforms from Bitcoin, as discussed in Chapter 2, Ethereum Fundamentals. Successful blockchain applications are still scarce and they are mainly concentrated in the financial industry. Ripple is a success story, which focuses on cash payments. 

Since blockchain technology is not suitable for all use cases, it is important to choose a proper use case before jumping into action. The following comments  can be useful to help you in determining which use case to work on and how to choose a blockchain platform for its implementation along with other architectural considerations:

• Not every use case is suitable for blockchain. For instance, many use cases can be implemented with traditional technology. It is true that blockchain is a data repository. If the sole purpose is to host data, choosing a regular database may be sufficient. Currently writing to a blockchain is still much slower than writing to a database. Insertion into a blockchain takes seconds or minutes. Insertion into a database takes only milliseconds. This makes databases a better choice in many use cases requiring high throughput, such as capturing credit/debit card transactions or equity trading market data. In the future, as performance and scalability improve, blockchain technology can be used for these use cases.
• In IoT use cases, an issue to consider is how to integrate an IoT device with a blockchain network. An IoT device is not a computer. As a result, an IoT device cannot be a node of the blockchain network. One possible solution is to link the device with a node on the network via APIs. The node interacts with the blockchain ledger and triggers the corresponding smart contracts upon receiving a signal from the IoT device. Performance is also an issue. Some IoT devices, such as airplane sensors, generate high-frequency measurements. A low TPS blockchain network cannot respond quickly to requests from these devices.

• The block size of a blockchain platform is limited. For example, Bitcoin has a block size restricted to around 1 MB. The following graph (from blockchain.com) shows its average block size history up to October 2018. For use cases such as an IP market for selling a novel or a movie, detailed information on a digital asset requires a lot of storage space. One can consider an architectural design of combining on-and off-chain storage to resolve the limited block size issue. Details about a digital asset can be saved off-chain at a centralized location. Ethereum has already adopted the on-and off-chain data storage approach:

• If a use case involves a digitized asset, a few issues need to be addressed for managing the underlying physical assets:
  • The physical asset needs to be notarized in order to prove its authenticity.
  • A solution is needed to ensure that the underlying asset remains the same and is not changed between the time it is notarized andwhen it is transferred in its title.
  • Similar to the digital coin double spending issue, a solution is needed to ensure that a physical asset is mapped to one and only one digital asset.
  • Valuable physical assets require secured places for storage. When ownership is transferred digitally, the corresponding ownership on the underlying asset needs to be recorded and transferred. A possible solution could be borrowed from bullion (gold) trading. The physical gold can be stored at a safe place, like the NY Fed gold vault. When a pile of gold bars changes ownership, the physical gold bars do not leave the vault.
• Although Ethereum is a generic platform supporting DApp development, its implementation involves a digital coin.  Whenever a cryptocurrency is part of a solution, one may need to deal with legal complications associated with it. For example, in the US, BTC is defined as an asset, not a currency. In other words, there is a tax implication (for instance, sales tax) when a BTC is sold to a buyer. Certain countries such as China prohibit cryptocurrency trading. As a result, an enterprise blockchain solution such as Hyperledger Fabric (HF) may be preferred since its implementation does not involve a cryptocurrency.  
• Many use cases such as healthcare data sharing or credit data digitization are not suitable for public use. A permission-based enterprise (or private) blockchain is needed instead of a public blockchain such as Ethereum.
• Another advantage of a private chain such as HF or R3's Corda over Ethereum is that both HF and Corda support development in Java, while Ethereum requires a programmer to learn a new language such as Solidity.  Given the scarcity of talent in Solidity, it is difficult and expensive to find qualified developers. On the other hand, turning a Java programmer into a HF or Corda developer could be an easier solution.
• Blockchain technology implies the guaranteed execution of a scripted legal document, a smart contract, which makes untrusting parties feel comfortable doing transactions. If a use case does not require a guaranteed transaction, then it is not a suitable use case. For instance, blockchain is not needed to replace the traditional internet dating site. Dating is very personal and it does not lead to a guaranteed transaction—a marriage.
• If blockchain is only used for the purpose of being a distributed ledger, it is not justifiable due to the cost associated with a blockchain solution. If fault tolerance and providing transparency are the primary goals, a distributed ledger can be implemented in a traditional way by making identical copies of the ledger at multiple nodes without the need to involve additional components, such as Bitcoin's mining and consensus mechanism. The consensus component was introduced to resolve the double spending issue. A distributed ledger does not involve double spending. In other words, blockchain is overkill if one only needs a distributed ledger.

• Blockchain currently is still not a suitable solution for many use cases requiring high throughput, such as stock trading or credit card transactions. Existing blockchain platforms are many magnitudes slower than other platforms (for instance, traditional databases) for hosting transaction data. For example, a specialized database, KDB, is needed to save market data in  terms of billions of records a day.

• Since smart contracts are scripted legal documents, there are legal challenges to be dealt with:
  • Are local laws applicable when a smart contract runs? If the answer is yes, how do you deal with conflicts with local laws, a scenario when the contract is legal at some locations of nodes and not at other locations of nodes?
  • Regulations and laws are not fully developed on blockchain and cryptocurrency. A US lawmaker recently pushed the IRS to clarify regulations on blockchain.
  • Since the execution of a smart contract is automatic and unstoppable, a blockchain application could be ruled to be unlawful when it cannot sufficiently address illegal activities such as money laundering.
  • Cryptocurrency receives different statuses in different countries. For example, in the US it is defined as an asset and in Singapore, it is considered to be a currency.