Written by: Roger Bitar, Offering Manager for Db2 & Big Data at IBM‘s Data and AI division.
View the original Medium article here.
View the Db2 11.5.4 Nebula Webinar Series here.
On June 30, IBM is releasing Db2 11.5.4, the latest version of the Db2 database. This new release includes lots of enhancements in the areas of Enterprise Readiness, Cost Savings, Multi-Cloud support, Graph Database, Spatial Analytics, Machine Learning, Security, and enhanced Developer Support. The following sections provide more details on these enhancements, and present the case on why you should consider upgrading to the Db2 11.5.4 release.
The enhancements in Enterprise Readiness include Faster Database Activation, with as much as a 30x reduction in activation time for configurations with large buffer pools based on internal benchmarks. Further enhancements apply to Db2 DPF (Database Partitioning Feature) in the area of “Federation Parallelism”, and to Db2 BLU in “Columnar Query Speed-up”. More detailed information about these enhancements can be found in this blog.
Db2 11.5 release introduced a simplification in the number of Db2 supported editions. The number of supported editions was reduced from 13 to 3, namely “Advanced Edition”, “Standard Edition”, and “Community Edition”. This release introduces a new “Base Edition” for small to medium workloads. All editions support the same Db2 features such as HADR, pureScale, BLU Acceleration, and other features at no extra costs. IBM has created a migration path to the new editions without any impact to your compliance status. Existing customers with legacy editions will not be impacted during migration to the new editions. You can start with the free “Community Edition”, and you can try all the other editions for free for the first 90 days, to determine the edition that is the most suitable for your business.
This release introduces “Adaptive Workload Manager for Columnar”. With different types of queries from short queries, to complex nested long running queries all touching the database, they will be competing for system resources such as memory and CPUs. This AI based enhancement will improve query performance many times over, especially in very large columnar databases. It fully automates the tuning of resources, and further classifies and prioritizes the queries based on an estimated need of resources to maximize performance, thereby saving time and efforts for the DBAs.
For customers who spent a lot of time tuning their complex queries to make the optimizer more efficient, only to see that the optimization is lost when they upgrade to a new Db2 release, we heard you. With this Db2 11.5.4 release, you have the option to lockdown the optimizer execution plan you used previously even after an upgrade.
The other important cost saving enhancement improves on the compression technology used in Db2. This applies to VARCHAR, VARGRAPHIC, and VARBINARY which did not compress with the dictionary-based compression. By using a new page-based String/VARCHAR compression algorithm, it will save almost 50% of the disk space used compared to now.
With Db2 11.5.4, we made it easier than ever to migrate to the cloud with IBM Cloud Pack for Data. The same Db2 functionality comes in many form factors to meet customer needs. You can deploy it on-premise on bare metal or virtualized environment, in a private cloud, hybrid cloud, public cloud, hosted or managed cloud, or in a containerized deployment on RedHat OpenShift platform.
Db2 is a fully integrated service in the IBM Cloud Pack for Data. IBM Cloud Pack for Data is our premier hybrid cloud and multi cloud platform powered by RedHat OpenShift. Cloud Pack for Data includes the infrastructure layer which could be a physical server, a virtualized environment, or any cloud platform of your choice. This layer is managed by Kubernetes using RedHat OpenShift platform. The next layer optionally integrates with IBM data services, data virtualization, and AI platforms seamlessly. IBM Cloud Pack for Data eases your way to containerize or cloudify your existing Db2 on-premise deployment if you decide it is in your future. This blog discusses Db2 as part of the IBM Cloud Pack for Data in more details.
Db2 11.5.4 introduces a graph engine for the first time in this release. A graph database stores information as relationships between entities, and represents this data using nodes and edges. Graph databases are highly beneficial in fraud detection, getting 360-degree customer views, and in recommendation engines. Rather than licensing a separate Graph database, and constantly duplicating data between databases, Db2 Graph will enable Graph analytics on top of Db2, allowing you to perform Graph analytics and SQL on the same copy of data, further saving you money and administrative efforts.
Db2 Graph will create a virtual graph view of the underlying data using the existing relationships already defined in Db2. Alternatively, you can create your own graph model, by defining how the tables and views already defined in Db2, map into nodes and edges in your graph. Db2 Graph uses the Apache TinkerPop framework to expose the graph model so you can execute Apache Gremlin queries. Db2 Graph fetches only the necessary data from Db2 at the time of query execution, so any updates made to data in Db2 will be reflected. More detailed information about Db2 Graph can be found in this blog.
Spatial analysis is a type of geographical analysis that seeks to explain patterns of human behavior and its spatial expression in terms of locational analysis. Examples include nearest neighbor analysis and Thiessen polygons. Spatial analysis lets us ask, and find answers to, a wide range of questions. Businesses can use spatial analysis to increase profits by defining more efficient sales territories, minimizing transportation or manufacturing costs, or gaining a better understanding of their customers. Scientific researchers can use spatial analysis to gain a better understanding of COVID-19 infections, and develop better models to predict the effects of different influences on the spread of the disease.
While Db2 Spatial Extender provided spatial analysis on row organized tables, with Spatial Analytics introduced in this Db2 11.5.4 release we can now perform spatial analysis on both row and column organized tables. This enables you to perform spatial analysis on analytical workloads that are typically deployed in a columnar database.
Db2 11.5.4 release introduces in-database machine learning. AI and ML enables us to learn from data, identify patterns, and make smarter decisions that augment human capabilities. When you gather and analyze data by looking at patterns, you can better identify customer needs, uncover trends, and innovate faster than your competitors.
ML models are built using neural networks, and rely on large datasets for model training. The accuracy of the models depends on the depth of the neural networks and the amount and quality of the training data. The more data you can collect and analyze, the more accurate are your models. By leveraging all your existing data in Db2, you save cost, time, and efforts versus moving your data to a data lake to do processing, cleansing, normalizing, and indexing in order to build and train the models.
Db2 11.5.4 supports data preparation activities, pre-processing activities, and model training activities, to enable the data scientists to prepare, train, and score the data all in Db2. Currently we support supervised learning algorithms in the form of Classification and Regression. While Classification can help with fraud detection, Regression helps with prediction and estimation. More detailed information about AI/ML in Db2 can be found in this blog.
This Db2 11.5.4 release introduces JSON Web Token (JWT) for Single Sign On (SSO). JWT defines a compact and self-contained way for securely transmitting information between applications as a JSON object. With this enhancement when a user authenticates on the front-end application that can create JWT tokens, the application will be able to generate a user-specific token that can then be passed to Db2 to authenticate in the backend using the Single Sign On.
Another security enhancement in this release is Authentication Caching. With this enhancement, the first-time login credentials will be cached and allowed for subsequent attempts. Customers will thus be able to eliminate overloading the authentication system for high-frequency applications by using the cached credentials. This helps improve the performance for all applications, and eliminate the chatter effect.
Enhanced Developer Support
As described in the Machine Learning section, Db2 11.5.4 supports data preparation activities, pre-processing activities, and model training activities. This is done via Python UDF support. With python UDF customers can repackage their python function as a UDF and store it in Db2. When they want, they can call their Python code directly, as database operations or functions similar to a stored procedure.
New in this release is Db2 REST Service support. This helps developers build their applications without worrying about deployment specifics or language dependencies, since they leverage HTTP methods to invoke the service. Each developer-defined service is associated with a single SQL statement. Now you can develop web, mobile, and cloud applications to interact with Db2 through a set of scalable RESTful APIs.
With all these enhancements in the Db2 11.5.4 release, customers are showing increased interest in upgrading. Not only they can save money and efforts by leveraging Db2 Graph, and ML model development inside Db2, they also gain performance and increased functionality in terms of security, spatial analytics, multi-cloud and developer support. All these enhancements will improve your ROI, help grow your business, and keep your customers and developers happy.