What is blockchain technology?

Conceptual illustration of blockchain records and a network of nodes

Blockchain is a way of keeping a shared digital record in blocks linked by cryptography. Participants follow agreed rules for accepting new transactions and maintaining the record. The point is not simply to keep copies on several computers. It is to make the accepted history difficult to rewrite without detection or acceptance by the network. The strength of that protection depends on how the network is designed and operated. [1]

Consider a hypothetical maintenance job. A motor manufacturer has the original test report, a repair workshop has the service history, and the factory has its own equipment file. When the motor fails again, the records disagree. Was the bearing replaced? Which report belongs to this serial number? Did the factory approve the work? Sharing another spreadsheet would not settle who was entitled to change the record, or whose approval was required. This is the kind of record-sharing problem a blockchain application could be designed to address.

It also helps to separate Bitcoin from blockchain. Bitcoin uses a blockchain to record digital currency transactions and prevent the same funds from being spent twice. Blockchain is the broader concept. Not every blockchain needs a cryptocurrency or mining. Hyperledger Fabric, for example, supports permissioned networks without requiring a native cryptocurrency. [2] [4]

How do the blocks form a chain?

A block groups transaction records with information used to check them. Except for the first block, known as the genesis block, each block links back to its predecessor through a cryptographic hash. In Bitcoin, that link uses the previous block's header. Altering an earlier transaction affects the cryptographic links that follow it. [2]

A hash is a fixed-length value calculated from data. Think of it as a digital fingerprint, not a shortened version of the document. A secure hash function is designed to make it impractical to find different contents that produce the same fingerprint. Recalculating the hash allows a later copy to be checked against the recorded value. [3]

There is a catch: someone who controls a local copy could change it and calculate new hashes. Hash links alone do not decide which history everyone should accept. That also requires the network's validation and consensus rules. Understanding how blockchain works means looking at both the linked records and the rules used to accept them. [2]

Who controls a blockchain network?

Distributed and decentralised describe different things. Records can be spread across many computers while one organisation still controls membership, software updates and operating policies. A permissioned network can run within a single enterprise or across several organisations. Counting servers therefore tells us much less about control than asking who runs them and who can change the rules. [4]

In a permissionless network, participation does not depend on an administrator approving membership, although protocol requirements still apply. A permissioned network restricts participation to approved members. A consortium can share governance among several organisations. These arrangements have different trust assumptions; none should be described simply as a network with no one in charge. [4]

The participants in a blockchain network may also have different jobs. Fabric distinguishes clients, peers, ordering nodes and administrators, with identities and policies determining their permissions. Permission to submit a record is not necessarily permission to approve it, read every detail or administer the network. [5]

What does “immutable” actually mean?

It does not mean that history is protected under every imaginable condition. NIST describes blockchains in terms of resistance to tampering and evidence of tampering, with accepted records remaining unchanged during normal network operation. The guarantee belongs to a particular security model, not to the word “blockchain” itself. [1]

Timing matters too. In Bitcoin, recent blocks can be replaced during a chain reorganisation. A transaction being broadcast, appearing in a block and gaining further confirmations are different stages. This is one reason questions about reversing blockchain transactions need a platform-specific answer. [2]

There is another distinction that matters in everyday use: preserving history does not mean freezing the current situation. Fabric separates the current state of an asset from the transaction history that produced it. A motor can change owners or move from “awaiting inspection” to “approved” through new transactions while the earlier history remains available. [6]

In our maintenance example, an incorrect report could be followed by a correction that references the original entry. The application would need rules for who can submit and approve that correction. Correcting the record openly is not the same as quietly rewriting what happened.

A shared record does not have to reveal every detail

The workshop may need to prove that a report matches an accepted record without disclosing its price agreement to every supplier. Fabric's private data collections provide one concrete approach: authorised organisations hold the private data, while its hash is recorded on the shared ledger for the relevant channel. A party later given the data can compare its hash with that record. Shared verification and unrestricted visibility are not the same requirement. [7]

It is also worth keeping three terms separate. Hashing supports content comparison. A digital signature lets a recipient check that a message was signed using the private key corresponding to a particular public key, and that the signed content has not changed. Encryption protects readability by turning the content into ciphertext. These serve different purposes; signing a document does not hide its contents. [3] [5]

When evaluating how blockchain protects data, ask two separate questions: who can verify this record, and who can read it? For an industrial project, decisions about data privacy should be made before sensitive information is distributed, not after. Private data features work within defined access policies; they are not a reason to assume that every blockchain entry is confidential. [7]

How does the network agree on a record?

Consensus is the process by which network participants maintain an agreed ledger under common rules. Those rules address which transactions are valid and which history to follow when competing versions appear. It is not necessarily a unanimous vote from every computer each time a record arrives. [8]

Bitcoin uses a proof-of-work system involving computational effort. Ethereum uses proof of stake, with validators committing assets and facing rewards or penalties under the protocol. These mechanisms are parts of larger consensus designs, which also include transaction checks and rules for selecting the accepted chain. Their security assumptions are different; the names are not interchangeable. [8]

For a maintenance network, the useful design question is more concrete: whose approval should be necessary before a repair is accepted? The answer might involve both the workshop and the factory. That requirement should guide the application design, rather than be added as an afterthought once a platform has been chosen.

What does a smart contract actually do?

A smart contract is program logic executed through a blockchain platform. It checks inputs against its rules and can update the recorded state. On Ethereum, a state-changing interaction starts with a transaction; the program does not independently notice that a delivery arrived or a motor stopped. [9]

In our hypothetical application, a maintenance contract could require a service report and a factory approval before marking the job as complete. This illustrates the appeal of industrial smart contracts: the parties can define a workflow rather than rely on each organisation interpreting a status field differently.

The code does not know whether the bearing was actually replaced. An external application, person or data service must supply that information. Systems that bring outside information to on-chain programs are known as oracles. The program's result still depends on the reliability of those inputs. Running the code correctly does not establish that the physical event happened as reported. [10]

Changing the rules is another separate task. Fabric, for example, uses a contract lifecycle in which organisations approve definitions and upgrades according to the network's policies. Software can support changing business rules, but the right to change them must also be defined. [9]

An industrial example: following a motor through its service life

Take the same example a little further. The manufacturer could register the motor's serial number and factory test reference. The workshop could submit its signed service report, and the factory could record acceptance. Later, an authorised buyer could examine the relevant history before taking over the equipment. This is an illustrative design, not a claim about a deployed installation. It connects maintenance records with supply-chain traceability around one asset identity.

There is no need to place every photograph, report and temperature sample directly on the chain. Large files can remain in an archive, while process measurements stay in a historian or data logging system. The ledger can hold document identifiers, approvals and hashes. Fabric's performance guidance specifically recommends considering off-chain storage for large data, with the hash stored on-chain. [11]

Later, the hash of a retrieved report can be compared with the accepted value. But a hash is not a backup. It cannot reconstruct a lost report, so somebody must still retain the file and make it available to authorised users. Nor does a matching hash establish that the report was truthful in the first place. [3] [10]

A temperature sensor with poor calibration can produce an inaccurate measurement that the system preserves perfectly. A correct measurement can also be assigned to the wrong motor. The distinction is fundamental to industrial data integrity: proving that information has not changed is different from proving that it was correct when collected. Device identification and measurement checks remain necessary parts of the proposed design. [10]

What are the practical limitations?

Shared verification takes resources. Transaction size, approval requirements, hardware and network conditions all affect performance. More participants do not automatically mean more processing capacity. In Fabric, for instance, adding ordering nodes can increase communication overhead even where other tasks can be spread across peers. Blockchain scalability therefore involves choosing the right workload and architecture, not just buying more servers. [11]

Energy use also needs context. Proof-of-work mining, proof-of-stake validation and permissioned networks do not have the same operating requirements. Ethereum's move away from mining illustrates how much the consensus design can affect electricity demand. Any discussion of blockchain's environmental impact should identify the network, infrastructure and measurement boundaries rather than apply one figure to every implementation. [12]

Integration brings its own questions. Before connecting a second factory, check whether both sites use the same asset identifiers, measurement units and approval meanings. Review how existing industrial communication and record-keeping systems will supply the required information. For our motor example, receiving “inspection complete” is not enough unless both parties agree on what that status confirms.

Finally, protect the keys that authorise transactions. A valid signature links a message to a key; it does not establish that the key has never been stolen. Identity systems such as Fabric's support certificate revocation, but the organisation still needs a response procedure. This is part of assessing blockchain security, not an unrelated IT detail. [5]

Where does the technology go from here?

Scaling approaches such as layer-2 systems move some activity away from the base chain while retaining a defined relationship with it. Other approaches divide data or processing responsibilities. These are different engineering choices with different trade-offs, not a universal upgrade that removes every limitation. The important questions remain what is verified, where it is verified and which security assumptions the application relies on. [13]

For a project decision, start with the record-sharing problem rather than the technology. Do several independent organisations need a common history that none should be able to rewrite on its own? Compare that requirement with a conventional database offering access control, audit logging and backups. Include integration, administration and long-term record access in the comparison. These are practical evaluation questions, not a claim that one architecture always wins.

Blockchain does not remove the need for trust; it changes which rules, participants and systems must be trusted. Its value is in making a shared record verifiable under those rules. The motor example works only when the device identity, approvals, source information and retained documents are all handled properly. A stronger ledger helps protect the history; it does not do the inspection for you. [1] [10]