The Supply Chain Management Blog covers the latest
in SCM strategy, technology, and innovation.
-
Blockchain for Supply Chain
August 15, 2020
Advancing Enterprise Blockchain with the Latest Generation of Oracle Blockchain Platform
As more blockchain applications go live and existing production deployments scale to larger business networks and bigger transaction volumes, Oracle has recently released the second generation of Oracle Blockchain Platform Cloud Service. This post discusses in detail the advantages it brings to customers.
According to Deloitte’s “2020 Global Blockchain Survey: From promise to reality,”[1] which surveyed 1,488 senior executives and practitioners in 14 countries, 39% are using blockchain in production in 2020 compared to 23% in the 2019 survey, with the figure rising to 46% for organizations with revenue over US$1 billion. In addition, IDC’s “Worldwide Blockchain 2020 Predictions”[2] report highlights that over one-third of the nearly US$11B forecasted spend on Blockchain Services by 2023 will be on managed services, with more than 80% of organizations in 2020 using Blockchain-as-a-Service (BaaS) as the first priority for blockchain infrastructure when implementing blockchain initiatives. As Blockchain implementations move from experimentation to production, and the technology turns the corner from early adopters to wider acceptance, the products that support customer implementations must also evolve to meet customer needs.
Driven by customer feedback, the new generation of Oracle’s managed Blockchain-as-a-Service platform on Oracle Cloud Infrastructure (OCI) provides significant new advantages for production blockchain deployments and decentralized business networks.
Numerous industry studies have noted that many blockchain projects do not reach production stage, with some hitting limitations due to immature technology stacks, which do not provide enterprise-grade resilience required for business-critical applications or are difficult to scale with sufficient agility and granularity to meet the needs of the production workloads in an affordable manner. When enterprise blockchains involve multiple competing organizations (e.g., competing suppliers to a major manufacturer or competing carriers in a shipping network) the transparency of a distributed ledger may run contrary to confidentiality or privacy needs in a business relationships or regulatory requirements, and the limited options for transaction or data confidentiality in the current frameworks may not be sufficiently powerful or granular.
“With this release, Oracle advances the Oracle Blockchain Platform to a new level of dynamic scalability, high availability, and confidentiality for enterprise blockchain applications running on Oracle Cloud,” said Frank Xiong, Group VP, Blockchain Product Development in Oracle. “We’ve enhanced our architecture to take advantage of the higher resilience and improved price/performance enabled by Oracle Cloud Infrastructure. This provides options for customers to dynamically scale-out and scale-up their configurations to better match their evolving workloads. The new version of Oracle Blockchain Cloud Service is designed to help customers comply with regulations, confidentiality, and information governance needs. We also continue to leverage the advancements of community-verified open source Hyperledger Fabric releases, including support for a decentralized RAFT consensus protocol.”
This new generation of Oracle Blockchain Platform has been architected to meet three key objectives:
- Strengthen its ability to meet production-grade customer needs – higher resilience, dynamic scalability, and enhanced security – by re-envisioning the architecture leveraging Oracle Cloud Infrastructure capabilities
- Extend blockchain capabilities based on the Hyperledger Fabric community advancements
- Enable new pricing metrics to better match customer expectations and avoid surprises as transaction volumes grow
“We jumped at the chance to upgrade to the new Oracle Blockchain Platform Cloud Service because we prefer the more predictable pricing model as well as dynamic scaling,” said Douglas Johnson-Poensgen, CEO of Circulor. “The migration was fast and painless.”
Strengthening Production-Grade Customer Needs
Since the launch of Oracle Blockchain Platform two years ago, many of our customers have gone live with a broad range of blockchain applications and workloads. These use cases include tracking provenance of food and fashion products, electric vehicle battery components, and luxury goods; tracking millions of maritime shipments; recording thousands of e-KYC verifications for instant bank account opening, and many other production applications. As these customers expand their networks and scale transaction volumes, the new release delivers:
- Enhanced resilience, high availability, and recoverability with automatic distribution and replication of all platform components over three Availability Domains (ADs) or three Fault Domains (FDs) in single AD data centers. This means the instance can survive any outage limited to a single AD or FD by automatically deploying all the components in an “anti-affinity” pattern to help ensure full replication of all service components. Customers with business-critical blockchain applications benefit from this new release’s ability to provide stronger resilience and recoverability, with the SLA for the Enterprise SKUs of at least 99.95% (see OCI PaaS and IaaS Services Pillar document for details.)
- Decentralized ordering using RAFT-based replicated state machine approach to replace the Kafka/ZK-based ordering in previous releases. Avoiding reliance on additional service components (Kafka/ZK) simplifies the infrastructure used to support blockchain operations and helps to increase the overall resilience and efficiency of the infrastructure. At a blockchain network level, using RAFT-based ordering enables multiple organizations to run ordering nodes as part of a “consenter set”, and different channels can be supported by separate ordering clusters. This results in greater overall scalability and more granular scaling of the specific ordering nodes based on the transaction volumes of the channels they support. And it further helps with privacy and confidentiality by enabling specific channels to use orderers from specific member organizations, rather than relying on a global ordering cluster in previous releases.
- Multi-region ordering cluster provides an even higher degree of resilience. The ordering cluster that sequences the transactions and creates the ledger blocks is the heart of the blockchain network. By setting up multiple instances across the Oracle Cloud Infrastructure regions and linking their RAFT-based ordering service nodes (OSNs) into a disaster-proof ordering cluster, the blockchain network can survive even a region’s failure or a major network outage as long as the remaining OSNs can meet the quorum majority requirements (e.g., 2 out of 3, 3 out of 5, etc.)
- Dynamic scale-up and scale-out of blockchain nodes and storage to enable admins to match resources to workload needs. Unlike the previous release of Oracle Blockchain Platform, where Oracle Cloud operations would scale your instance based on a service request, this version provides customers the controls for a granular scale-out and scale-up as a self-service. For example, you can dynamically:
- Add peer and ordering nodes, specifying which AD or FD to deploy them in and how much OCPU resource to add
- Scale-up or down existing peer nodes within the limits of their VM
- Increase or decrease the number of replica copies for the CA or REST Proxy
- Expand the storage in 1 TB increments
- Expand the total number of OCPUs available to the platform
- Enhanced confidentiality and privacy with support of the new fine-grained on-chain access control mechanism – providing a new chaincode API and library to create, update, manage, and check access control lists (ACLs) that can be defined on any data or functions within a smart contract. This enables developers to use normal blockchain transactions to create Groups, Resources, and ACLs on the ledger and use the provided APIs to verify access privileges in their business logic. For customers who need to balance distributed ledger transparency with business confidentiality or regulatory privacy requirements, this mechanism is not only more granular than anything available in Hyperledger Fabric today, but it is implemented on-chain and can be deployed in a shared ledger, providing built-in transaction history for auditing of any ACL changes.
- Audit of Control Plane operations through an Oracle Cloud Infrastructure audit log. Audit records are available through an authenticated, filterable query API or can be retrieved as batched files from Oracle Cloud Infrastructure Object Storage. A similar audit log for blockchain Console operations is planned in an upcoming release.
- Support for Oracle Database Blockchain Tables via the rich history database feature. This feature, when enabled, asynchronously shadows blockchain transaction updates to an Oracle Database or Autonomous Data Warehouse, which is often used for analytics and reporting, as well as Autonomous Transaction Processing. However, once in regular tables, the data is not under the same tamper-proof controls as it was on-chain, so some auditors may question the integrity of the analytics reports built from these tables, particularly if they are used for financial applications, SOX-404 compliant reporting, etc. Using the new Oracle 20c Blockchain Tables as targets for the rich history – which this release enables – provides highly tamper-resistant, insert-only database tables with hash-linked rows, PL/SQL validation, and other immutability features to ensure the same strong auditability and tamper-resistance spans from on-chain data to its off-chain shadow in the database.
"The dynamic scaling changes everything. We can allow retailers, who use our platform for tracking product provenance information, to run at full throttle with massive engines when they need them, but cut costs when they are not required,” said Peter Merkert, Co-founder of retraced. “Further, there is a central audit log mechanism out of the box in OCI for all our cloud services, which is now also used by the new generation of the Oracle Blockchain Platform. Unlike before, we can see configuration changes for all services in one place.”
Extending Blockchain Capabilities for Decentralized Business Networks
Permissioned blockchain networks typically start with a few member organizations and grow to include many more members over time. Increasingly, the governance of the network becomes an important consideration in providing equal rights and treatment among the members. This includes the ordering responsibilities – the process of sequencing transactions and grouping them into ledger blocks. In prior releases this ordering was done by ordering service nodes using Kafka messages to achieve consensus on the sequence of transactions in the block. Because Kafka’s best performing configuration is when its brokers all run in a single site, the ordering nodes were also provided by a single organization – the network Founder.
With the new release, the ordering consensus mechanism has been upgraded to RAFT – a dynamic leader-based protocol based on a replicated state machine approach, which removes the need for Kafka and enables orderers from multiple organizations to cooperate in an ordering cluster. With this change, any blockchain instances in the network can run OSNs and join them to selected channels in order to participate in a decentralized ordering process. The result is more uniform and distributed governance with all members being able to participate equally. In addition, removing dependence on Kafka and Zookeeper to keep the platform running reduces operational complexity and saves resources previously consumed by these components.
As mentioned above, with RAFT, this release can use groups of orderers provided by blockchain platform instances deployed across multiple Oracle Cloud Infrastructure regions, and even non-Oracle Fabric nodes in third-party vendor clouds, to allow a widely distributed geo-redundant ordering cluster that can continue to operate even if one of the regions or its communications are down. With RAFT, it is important to remember that its operation requires a majority of the initial quorum of OSNs to be available. Thus, if you start with a quorum of N OSNs, you can lose some of them as long as the remaining number still constitutes a majority (N/2 + 1) of the initial N OSNs. The speed of the network links between orderers is also important to keep in mind when benchmarking for high throughput requirements. Oracle Cloud Infrastructure regions use high-bandwidth, fault-tolerant network links with at least 100 Gbps between them and consistent latency (as low as 75ms within US and as low as 75ms US to EU).
Adjusting Pricing Metrics to Better Match Customer Expectations
When customers are experimenting and running pilots, transaction-based pricing of the current instances is elastic and affordable. But as workloads move to production and transaction rates become much higher, customers may want a less elastic pricing model and one that’s more predictable and not dependent on external transactions.
In this release we introduce OCPU-hour based pricing metrics to respond to these needs. The configuration of the instances is measured in OCPUs and storage resources allocated to them. In fact, there are two SKUs: Standard – providing a low cost fixed configuration for development and testing, and Enterprise – providing a highly resilient and dynamically scalable configuration, where OCPUs and storage can be expanded. The table below shows the two newly available SKUs and their configuration details.
The initial number of OCPUs and storage in an Enterprise instance can be adjusted up or down (but not below the minimum) dynamically using the Control Panel, and the new numbers will be reported into the billing system once the scaling operation has been completed. For the pricing of OCPU-hr and storage, please refer here.
In addition to the new pricing metric, two new customer options are introduced in this release:
- Ability to Stop and Restart instances on demand using Control Panel. During the time the instance is not running, the charges for OCPUs are reduced to 25% of the allocated capacity (e.g., an Enterprise Large instance with 16 OCPUs will only be charged for 4 OCPUs during the hours it is not running).
- Oracle’s Bring-Your-Own-License (BYOL) program enables customers to leverage existing software licenses for Oracle PaaS at a lower cost. Specifically, licenses of Oracle Blockchain Platform Enterprise Edition (an on-premises version of the blockchain platform) can be registered in your Oracle Cloud Infrastructure account and applied to the Enterprise SKU OCPU licenses at a rate of 1 OBP EE Processor license = 2 OCPUs.
For customers using current versions of Oracle Blockchain Platform Cloud Service, Oracle will provide a migration capability to the new release with no loss of ledger data or configuration information.
This overview blog post kicks off a series over the coming weeks, which will further explore the technical advantages of this release in depth, covering topics such as:
- High Availability & Resilience
- Dynamic Scalability
- Decentralized Ordering Based on RAFT
- Fine-grained Access Control
- Rich History Using Blockchain Tables
- Using Oracle Cloud Infrastructure SDKs and CLI for Blockchain Platform life cycle management operations
For more information:
- For further details on the new version read What's New.
- For additional resources on the managed service offering, including customer case studies, pricing, documentation, partner solutions, and more, please visit the Oracle blockchain platform cloud service page.
- For more information on Oracle Blockchain solutions and related news please visit Oracle.com/Blockchain.
[1] Deloitte, “Deloitte’s 2020 Global Blockchain Survey,” Linda Pawczuk, Jonathan Holdowsky, Rob Massey, Brian Hansen, June 2020, https://www2.deloitte.com/us/en/insights/topics/understanding-blockchain-potential/global-blockchain-survey.html?id=us:2el:3pr:4di6608:5awa:6di:061620:&pkid=1007151
[2] IDC, “IDC FutureScape: Worldwide Blockchain 2020 Predictions,” James Wester, Alison Close, Steven D’Alfonso, Rivka Gewirtz Little, Yu Xue, Scott Lundstrom, Simon Piff, Jeff Rivkin, Jeff Yong Xun Xie, Shawn P. McCarthy, October 2019, https://www.idc.com/getdoc.jsp?containerId=US44538119