Databricks Lakehouse Platform vs Oracle Data Warehouse

• Process raw data in One Lake (S3) env to relational tables and views
• Share notebooks with our business analysts so that they can use the queries and generate value out of the data
• Try out PySpark and Spark SQL queries on raw data before using them in our Spark jobs
• Modern day ETL operations made easy using Databricks. Provide access mechanism for different set of customers

Read full review

• Able to handle very large data sizes efficiently from a performance, high availability and manageability perspective. This is accomplished through the Oracle Partitioning functionality. Partitioning allows large segments (tables, IOT index-organized tables, indexes) to be broken into smaller segments at the physical layer but treated as a whole at the logical layer.
• Provides support for dual-format architecture through Oracle In-Memory functionality. Without any change to application code one can obtain in-memory performance. This functionality enables us to have the tables represented in both the row format and the column format using in-memory format. This is a huge boost for BI/analytic queries since the Oracle optimizer is able to intelligently choose the appropriate format.
• Provision to materialize a subset of table data or table joins. This is through materialized views and the optimizer will rewrite the query against the base tables to make use of this materialized view. This provides a huge performance boost and is critical in VLDBs as in a data warehouse. The query rewrite is fully transparent to users.
• Provides multiple compression capabilities. This is very useful not only for deducing the storage foot print but as well as increase performance at different layers of the infrastructure including query performance. The compression functionality can be applied against both structured and unstructured data.
• With the advent of Engineered Systems (Exadata, Database Machine, SuperCluster) there are specific features and functionalities that can further boost the Oracle data warehouse. These are related to consolidation, Smart Scan, Storage Indexes, EHCC (Exadata hybrid columnar compression) and much more.
• RAC - Real Application Clusters (with 2 or more nodes) provides functionality for high availability, performance and scaling as the work load increases. The parallelism is provided both within a node and as well as across nodes. If for any reason a node goes down the data warehouse is still available through other nodes and the running queries are transparently failed over to the surviving nodes.

Read full review

• Better Localized Testing
• When they were primarily OSS Spark; it was easier to test/manage releases versus the newer DB Runtime. Wish there was more configuration in Runtime less pick a version.
• Graphing Support went non-existent; when it was one of their compelling general engine.

Read full review

• Customer support isn't the best out there. We usually have to wait about an hour to get some form of assistance.
• Pricing is a bit higher than many of its competitors such as AWS Redshift.
• Tweaking features requires dedicated staff. Software is fairly advanced. Would be difficult to use for newcomers.

Read full review

• Machine learning is a very new concept and not many universities offer to teach it. My school and a few others have been utilizing Databricks as one of the tools to teach and learn machine learning. By doing this, my university is creating a strong future workforce for the job market.

Read full review

• Very cost effective for large databases.
• Very fast results on simple queries.
• Analytical functionalities are of wide range, which makes it very cost effective.
• For smaller businesses, it might be a great asset.
• Some features are supposed to be bought separately, so one needs to consider this before licensing with Oracle DW.

Read full review