As Doug Fisher, one of Intel’s vice presidents, once said, “Data is to the 21st century what oil was to the 20th.”[i] It is therefore no surprise that, after autonomous cars and combat drones,[ii] autonomous databases have now arrived. In March 2018, the world’s first database of this kind, Oracle Autonomous Data Warehouse Cloud, fully managed by AI (Artificial Intelligence) algorithms, was launched. Everything therefore suggests that the profession of database administrator may soon become a thing of the past. To answer the question posed in the title—revolution or evolution—it is definitely evolution. The trend towards automation has been visible in IT for many years, while autonomy is only now gaining momentum.
What are autonomous databases?
Based on the available information, an autonomous database can be defined as a cloud solution that uses ML (Machine Learning) algorithms to automate routine tasks within a database management system. In particular, this includes installing updates, optimizing databases, ensuring security, and creating backups. Some estimates suggest that demand for database administration services could fall by more than 90%
[iii]. Dozens, if not hundreds, of initialization settings, database configuration parameters, operating-system parameters, and business-application parameters are already managed by algorithms capable of learning independently. Moreover, in a cloud environment, the end user does not have to worry about many other parameters and settings, because the service provider ensures that the entire environment remains up to date and secure.Autonomous databases provide a high degree of automation for operational tasks, including indexing and changing configurations and settings. The working environment must be secure and well protected against external attacks—the autonomous database must itself ensure that the security policy is implemented and kept up to date. All data is encrypted by default. In terms of security, autonomous databases offer a very high level of protection, while backups are fully automated. AI algorithms monitor the condition of the database. As a result, autonomous databases can guarantee data availability of 99.995%.
How do autonomous databases work?
If a database is autonomous, the need for human work is reduced to an absolute minimum. The roles and functions previously performed by people have been taken over by AI and ML algorithms. Together, they carry out all the processes involved in managing the database and the necessary infrastructure.Autonomous databases create storage indexes on their own, intelligently manage the flash cache, and optimize queries. These advanced background algorithms eliminate the need to manually create indexes, partitions, materialized views, and statistics in order to optimize database workloads. ML algorithms observe database activity, identify anomalies, and take appropriate corrective action by tuning key resources such as processor computing power, storage capacity, and network bandwidth. Skillful balancing of resources and workloads prevents business processes from slowing down and stops a single batch job or application from monopolizing resources.For example, an Oracle autonomous database provides the resources that are actually needed at a given moment and, most importantly, switches them off when they are no longer required. Under this model, charges are based solely on actual database use, which helps optimize and reduce the cost of maintaining database infrastructure within an organization.Additional savings in autonomous databases are generated by mechanisms that automatically download and install patches and updates. Eliminating administrators’ work in this area is not only a financial matter; it also improves data security by removing human errors and omissions, which account for most database downtime. Full native data encryption and an authorization system prevent access by people without the appropriate permissions. Automatic, rapid security updates protect against cyberattacks and threats from within the organization.Perhaps for the first time, as Oracle’s CEO Larry Ellison put it, bots are fighting bots
[iv]: the bots protecting autonomous databases can repel large-scale attacks by malicious bots much faster and more effectively than a human could. It is estimated that autonomous databases may require less than 2.5 minutes of downtime per month
[v], mainly to install patches. In addition, if business applications operate continuously, backups are also performed smoothly and without downtime, minimizing any risk of data loss.Running such databases in a cloud environment also provides virtually unlimited scalability. In response to changing user demand, the database server can simultaneously adjust its requirements for computing power and storage resources. In this way, the system can smoothly change the number of processor cores, entire compute nodes, or storage servers it uses. For example, when quarterly reports are being generated, an organization can order eight cores instead of the four it normally uses. It can also switch off all computing power for the weekend and restart the necessary infrastructure on Monday. This approach to resource optimization produces real and by no means insignificant savings.
How does an autonomous database differ from a traditional one?
As the name suggests, an autonomous database operates independently, without human intervention. Administrative tasks are supervised by AI and ML algorithms. Removing people from the process means removing the weakest link in the database security chain, because people are the ones most likely to make mistakes, neglect their duties, and forget basic administrative tasks. Each of these factors creates a risk of data loss or data becoming outdated.To use a simple analogy, the difference between an autonomous database and a traditional one is like the difference between
an Excel spreadsheet and an ERP system. The functions may appear similar, but the qualitative difference is enormous.Autonomous databases are used in both data warehouses and transactional systems. This means the range of possible applications is very broad, but due to their technological sophistication and cost, they are primarily an enterprise-class solution. In the near future, it is difficult to expect autonomous databases to replace popular free solutions such as MySQL. In large enterprises with very large databases, however, the time for autonomous databases has already arrived. Their adoption will also improve the effectiveness of enterprise-management ERP systems, allowing organizations to focus more closely on optimizing business processes.
What are the benefits of using autonomous databases?
With traditional duties removed from database administrators’ task lists, they can focus on much more important work. Their knowledge and experience should be directed towards planning data-aggregation strategies, as well as data modeling, processing, and management. Administrators can serve as an effective link between application developers and databases, helping to use the available functionality efficiently while minimizing changes to application code.From the perspective of the organization as a whole, the broad flexibility of autonomous databases is particularly important. It improves business agility by significantly reducing the time required to launch IT services and infrastructure for specific tasks and projects.Because an autonomous database engine is optimized for maximum performance and availability, it requires fewer resources and less computing power than traditional databases. When workloads are variable and uneven, easy, automated scalability makes it possible to pay only for the resources actually consumed, without stopping or slowing business processes—additional resources are allocated or released almost immediately. Consequently, users of autonomous databases can expect a significant reduction in the solution’s total cost of ownership.
Are autonomous databases the future of big-data analysis?
As 5G and the Internet of Things (IoT) become more widespread, the amount of data generated and used by companies will be extraordinarily large.
Datasets containing hundreds of petabytes will be nothing unusual. The challenge for organizations will be not only to store this data securely, but also to put it to meaningful use.A fast, smooth stream of data needed for business decisions at every management level will require a new, complex database-management architecture. Without support from autonomous databases, maintenance tasks would place an excessive burden on administrators, who already have to search, transform, mask, and cleanse data so that analysts can then perform their work.Autonomous databases and administrators who strengthen an organization’s analytical capabilities are the right combination of people and algorithms for business-data analysis. With large datasets, even a minimal improvement can deliver substantial savings at scale.Data modeling in autonomous databases will become the most sought-after skill. The ability to create and modify logical models, multidimensional data cubes, and data-type models will be highly valuable when planning and implementing the best ways to use data in business applications.
[i] https://businessinsider.com.pl/technologie/nowe-technologie/oracle-open-world-premiera-autonomicznej-bazy-danych/gp5my62
[ii] https://businessinsider.com.pl/technologie/nowe-technologie/oracle-open-world-premiera-autonomicznej-bazy-danych/gp5my62
[iii] https://www.computerworld.pl/news/Autonomiczna-Baza-Danych-Oracle,412160.html
[iv] https://diginomica.com/oracle-openworld-2018-ellison-makes-convincing-pitch-on-automation-and-security-for-oracle-cloud-2-0-but-cant-resist-trashing-aws
[v] https://www.oracle.com/pl/autonomous-database/what-is-autonomous-database/
