Customers choose Oracle Autonomous Data Warehouse vs Amazon Redshift, a cloud data warehouse service that eliminates all the complexities of operating a data warehouse, securing data, and developing data-driven applications. It automates provisioning, configuring, securing, tuning, scaling, patching, backing up, and repairing of the data warehouse. It includes tools for self-service data loading, data transformations, business models, automatic insights, and built-in converged database capabilities that enable simpler queries across multiple data types and machine learning analysis. It’s available in both the Oracle Cloud Infrastructure (OCI) and customers’ data centers with Oracle Cloud@Customer.
Customers choose Oracle over Amazon Redshift for several reasons:
Autonomous Data Warehouse provides built-in analytics and machine learning (ML) capabilities that are not available with Amazon Redshift, reducing costs incurred for add-on services. When customers compare total cost of ownership (TCO) for cloud data warehouses not only should the core data warehouse subscription cost be included, but also additional charges for analytics modules, storage services, as well as the integration and secure administration of any services required to support data scientists and business users.
Capability and evidence |
Oracle Autonomous Data Warehouse |
Amazon Redshift |
---|---|---|
Can customers build a self-service data mart for reporting without having to integrate multiple additional services?Autonomous Data Warehouse includes comprehensive capabilities to address a broad set of analytical use cases. Customers can easily create self-service data marts using built-in capabilities for data ingest, transformation, ML, business analytics, graph analytics, geospatial analytics, and application development.Amazon Redshift is missing an extensive set of data and analytical capabilities that are included in Autonomous Data Warehouse. For instance, organizations must pay for and integrate Lambda, Kinesis, and S3 just to move data from Amazon Aurora to Redshift—or master Amazon Glue, which requires a DBA to define the ETL job whereupon Glue generates PySpark code, which often needs to be customized based on validation and transformation requirements. Customers who want to use graph analytics need a separate specialty database, data scientists who want to use ML must export data to and reimport it from another AWS service, and data ingestion and transformation require separate AWS or third-party services. These additional services add complexity to customers’ data marts, fragment data, introduce security risks, and require additional management, significantly increasing Redshift’s TCO. |
yes |
no |
Does the data warehouse include built-in machine learning capabilities for the training and deployment of ML models?Autonomous Data Warehouse includes in-database ML algorithms and automated machine learning (AutoML) capabilities. Data scientists and data science-savvy non-experts use these capabilities to accelerate modeling tasks, automate the discovery of new insights, and quickly generate predictions. Building ML models inside Autonomous Data Warehouse eliminates the time-consuming requirement to extract, reformat, and move data to separate additional cost services, enabling users to create models without having to learn new tools. Running inferences inside the data warehouse also eliminates the need to extract data, reduces time-to-insight, and data proliferation risks while lowering costs.Redshift lacks in-database advanced analytics and ML algorithms. Customers must use additional Amazon services, such as SageMaker Data Wrangler, SageMaker Studio, SageMaker Pipelines and other services to move and reformat data, develop/test algorithms, and generate predictive insights. Moving data out of Redshift and using separate services delays insights while increasing complexity, security risks, and costs. |
yes |
no |
According to an industry analyst, the cost of system downtime can range from US$140,000 to US$540,000 per hour, depending on organization size. Autonomous Data Warehouse protects users from service interruptions due to the failure of multiple components by running on Oracle Exadata infrastructure with server, network, and storage redundancy. As many as 24 drives and 8 flash cards within 2 storage cells can fail in an Exadata and operations will still run nonstop. Autonomous Data Guard and other built-in capabilities enable fully automatic, zero-data loss failover during unplanned outages at the system or availability zone level. Online upgrades and maintenance of Autonomous Data Warehouse eliminates costly downtime incurred during scaling and maintenance operations.
Capability and evidence |
Oracle Autonomous Data Warehouse |
Amazon Redshift |
---|---|---|
Does the data warehouse SLA include both planned and unplanned downtime?Autonomous Data Warehouse enables organizations to run data warehouses with approximately 4.4 hours of total downtime per year--including both planned and unplanned downtime.Redshift availability SLAs have a goal of 8.8 hours per year of unplanned downtime. Redshift’s SLA does not cover planned downtime for scaling and regular maintenance operations. Redshift requires a scheduled maintenance window of at least 30 minutes per week, or over 26 hours per year, and even planned cluster resizing operations can require additional hours to days of downtime. If planned downtime were included, Redshift’s SLA would be 34.8 hours, 7.9x worse than Autonomous Data Warehouse. |
yes |
no |
Can the data warehouse remain online during upgrades?Autonomous Data Warehouse can be upgraded without disruption as the service continues to run.Redshift must shut down while undergoing regular maintenance. When maintenance is performed on Redshift, the cluster is not available for normal operations and it terminates any queries or other operations that are in progress. |
yes |
no |
Can the data warehouse scale without downtime?Autonomous Data Warehouse can scale up to increase query performance and concurrent user throughput or scale down to minimize costs—both without downtime. Resizing occurs instantly, fully online, and without DBA intervention.Scaling Redshift clusters up or down to match query performance requirements is a manual process that requires a snapshot of the data warehouse and system execution paused for 4 to 8 minutes. Once the Redshift cluster restarts, it’s in a read-only state until elastic resize data movement completes, about 10 to 15 minutes after the resize operation begins. However, elastic resize is not available in all situations, forcing customers to use “classic resize” which can take up to 2 days to complete. During resizing, Redshift data becomes stale and analyses may no longer be accurate. Concurrency scaling to support more users happens automatically in a matter of seconds by creating read-only copies of the entire Redshift data warehouse. Concurrency scaling does not improve query performance and can substantially increase per-hour costs. Concurrency scaling is not available for single-node warehouses. |
yes |
no |
Agea, a leader in digital media in Argentina, uses Autonomous Data Warehouse and Oracle Analytics to create a 360-degree view of customers by analyzing more than 20 billion clicks and searches a day. With Autonomous Data Warehouse, Agea reduced marketing campaign lead times from three days to less than 24 hours, cut costs by 50%, and achieved significantly higher availability.
Executives and Line of Business leaders need reliable access to the latest data to make decisions, even when query complexity and underlying data change. Autonomous Data Warehouse’s self-tuning and self-scaling capabilities continuously track performance, optimize execution, and scale compute resources so individual queries and multi-user workloads run at peak speed. These automated capabilities eliminate over-provisioning and the need for DBAs to monitor and adjust cloud data warehouse resources as workloads change. Without this level of automation, DBAs and systems admins continue to have to manually manage and tune Amazon Redshift.
Capability and evidence |
Oracle Autonomous Data Warehouse |
Amazon Redshift |
---|---|---|
Does the data warehouse provide automatic SQL tuning and query optimization?Autonomous Data Warehouse automatically tunes and optimizes SQL query performance, regardless of data structures or data volumes. No manual intervention by DBAs is needed.Amazon Redshift provides a limited set of automated features for tuning data warehouse performance. While Redshift does help optimize data table design, DBAs must manually implement recommendations on query tuning, sort key selection, and other factors to address performance issues. The result is higher operational costs and higher TCO. Not surprisingly, some customer benchmarks show that Autonomous Data Warehouse is 4X faster than Redshift. |
yes |
no |
Does the data warehouse automatically scale up compute resources to increase query performance and scale down to minimize costs?Autonomous Data Warehouse automatically scales compute resources up and down to increase the performance of complex data warehouse queries and minimize costs. There is no need to experiment with node types and cluster sizes to determine the best configuration for each workload. With Autonomous Data Warehouse, scaling takes place instantly on a granular, processor core at-a-time basis without manual intervention by DBAs.Amazon Redshift does not automatically reconfigure itself to increase query performance. DBAs must experiment with workloads to identify the best type of node and cluster size for each one. Changing the number of cluster nodes using “elastic scaling” requires DBAs to manually initiate a 3-step process that can take up to 15 minutes to complete, including 4 to 8 minutes when the cluster is unavailable. Redshift warehouses running on 2 vCPU dense compute (DC2) nodes and 4 vCPU dense storage (DS2) nodes may only be scaled up or down by a factor of 2 from their original size. In addition, even when Redshift clusters are scaled up they “might not have enough storage because of the way data is redistributed,” creating uncertainty for IT teams. |
yes |
no |
Does the data warehouse automatically scale compute in a granular manner to minimize costs while supporting more users?Autonomous Data Warehouse automatically scales compute resources to support more simultaneous queries against the data warehouse. Additional resources are instantly added one processor core at a time, and they are removed when concurrent workloads no longer need them.Amazon Redshift provides automatic concurrency scaling to support additional users, but with significant restrictions and incremental costs. Concurrency scaling is not available for small data warehouses with single-node clusters. When available, Redshift concurrency creates additional read-only copies of the entire data warehouse, so all users that need to update the data warehouse must run on the primary cluster, creating potential bottlenecks. Redshift concurrency scaling can also be expensive since it requires paying for additional full clusters instead of just the minimum amount of additional compute resources needed. For instance, concurrency scaling of a moderately-sized Redshift data warehouse with eight ra3.4xlarge nodes requires adding 96 vCPUs of processing power at a time—even if the customer’s workload only needs a small amount of extra resources. |
yes |
no |
11880.com uses Oracle Autonomous Data Warehouse to provide over 80 million online recommendations to customers of more than one million small- and medium-sized businesses in Germany. 11880.com achieved a 4X performance improvement compared to Amazon Redshift and was able to consolidate seven DBAs roles into one architect, realizing a return on investment in under twelve months.
Data warehouses potentially contain personal information on millions of consumers and organizations, making them prime targets for cybercriminals. According to a study by the Ponemon Institute, the average cost of a data breach in 2020 was US$3.86 million while the loss of 50 million records resulted in costs of more than US$390 million. Autonomous Data Warehouse includes extensive built-in security from always-on encryption to data discovery and security monitoring tools that make it easy for IT organizations to ensure data and privacy aren’t compromised by external or internal bad actors.
Capability and evidence |
Oracle Autonomous Data Warehouse |
Amazon Redshift |
---|---|---|
Does the data warehouse include data masking, data discovery, and user scoring to secure customer data?Autonomous Data Warehouse includes built-in security with Oracle Data Safe, helping organizations understand the sensitivity of their data, mask sensitive data, and evaluate risks to their data. Customers use Data Safe to implement and monitor security controls, assess user security risks, monitor user activity, and address data security compliance requirements.Amazon Redshift lacks this level of built-in functionality. Customers must implement the equivalent of Oracle’s built-in features using multiple add-on tools and services, increasing complexity, potential security risks, and operational costs. |
yes |
no |
Does the data warehouse provide security controls to prevent privileged users from accessing sensitive data?Oracle Database Vault provides powerful security controls to help customers protect application data from unauthorized access while helping address privacy and regulatory requirements. IT teams use Database Vault to block privileged account access to application data and to control sensitive operations inside the data warehouse. Database Vault transparently secures data warehouses, eliminating costly and time-consuming changes at the application level.Amazon Redshift does not have built-in preventive controls to block privileged users and DBAs from accessing sensitive data in the data warehouse. As shown here, Amazon employees have accessed and disseminated customer data. This would not be possible with the data protections engineered into Oracle Autonomous Data Warehouse. |
yes |
no |
Does the data warehouse enable encryption by default?With Autonomous Data Warehouse, encryption of customer data is always turned on—it’s not an option. Data is encrypted at rest, in motion, and as part of backups. Always-on encryption helps organizations protect sensitive data, address compliance requirements, and minimize human errors that could expose data.Amazon Redshift requires customers to manually turn on encryption—it’s not enabled by default. DBAs must choose a hierarchy of encryption keys to encrypt the database. Encryption in Redshift is more complex than in Autonomous Data Warehouse and may introduce human errors during the encryption process. Data security is limited by the weakest link in the chain and, as reported here, AWS S3 security misconfigurations may leave data unprotected and easily accessible to hackers. With AWS, the burden for security is on the customer. This would not be possible with the data protections engineered into Oracle Autonomous Data Warehouse. |
yes |
no |
DX Marketing relies on Autonomous Data Warehouse and its integrated machine learning capabilities to analyze terabytes of consumer data and help its clients reduce the cost of finding new customers by 52%. Autonomous Data Warehouse increased the performance of DX Marketing’s data warehouse by up to 70% and protects the privacy of more than 260 million licensed records of US consumers.
Data residency and regulatory compliance requirements sometimes prevent customers from running data warehouses in the public cloud. Oracle Autonomous Data Warehouse can be deployed either serverless or on dedicated infrastructure in Oracle Cloud Infrastructure regions as well as on Cloud@Customer solutions located in customers’ data centers. As a result, organizations achieve cloud benefits while also meeting data residency and security requirements—and providing low-latency connectivity to existing IT infrastructure.
Capability and evidence |
Oracle Autonomous Data Warehouse |
Amazon Redshift |
---|---|---|
Does this solution provide options for public cloud and on-premises deployments?Autonomous Data Warehouse provides a choice of deployment in OCI regions around the world or in customers’ data centers with Exadata Cloud@Customer or Dedicated Region Cloud@Customer solutions. By providing identical architectures and capabilities on-premises and in the cloud, organizations can easily move workloads to meet changing business needs and regulatory requirements.Amazon Redshift is available only in the AWS public cloud. Redshift is not available for deployment on AWS Outposts, so customers do not have the option of deploying it in their data centers. |
yes |
no |
Can the data warehouse meet data residency requirements?For certain applications, industries, and geographies, data must always reside locally in customers’ data centers and behind customers’ firewalls. Autonomous Data Warehouse can be deployed as a fully managed cloud service in customers’ data centers using Oracle Cloud@Customer solutions to meet data residency and security requirements or to provide low-latency connectivity to existing on-premises resources.Amazon Redshift is available only in the AWS public cloud. Redshift cannot be deployed on AWS Outposts in customers’ data centers, failing to meet data residency or security requirements. |
yes |
no |
Does this solution have options for multi-cloud interoperability?Autonomous Data Warehouse provides direct, transparent, in-place access to data stored in Oracle Object Storage, Azure Blob storage, and Amazon S3. Users can seamlessly join and analyze datasets from all three object stores with a single SQL query.Amazon Redshift users cannot directly perform joins and analyze datasets from non-AWS object stores with a single SQL query. As described here, data must first be migrated into Amazon S3 using a series of added-cost data ingestion tools and/or connectors before it can be loaded into Redshift for analysis. |
yes |
no |
Entel, one of Latin America’s largest telecom providers, is accelerating its digital transformation by moving more than 30 databases to Oracle Autonomous Database on Exadata Cloud@Customer, which is now fully managed by Oracle and running in Entel’s data center. Entel realized a 3X performance increase and reduced the security exposure for its mission-critical databases with Autonomous Database’s self-security capabilities.
Try Oracle Autonomous Database for free