IBM Db2 Analytics Accelerator can now be deployed on IBM Z

IBM Db2 Analytics Accelerator can now be deployed on IBM Z

 Sowmya Kameswaran, IBM, Technical Lead - Modernization, DevOps and Db2 zTools/Accelerator Management with DSM

Yes! The IBM Db2 Analytics Accelerator for z/OS V7.1 (IDAA) now deploys directly on IBM Z. The general availability announced in Nov 2017 opens up interesting new possibilities for IDAA - it now has two deployment alternatives - one on the IBM Integrated Analytics System appliance and another one running in a dedicated SSC LPAR on IBM Z Mainframe.

 

The latest Analytics Accelerator is powered by Db2 with BLU Acceleration engine, which is an internal change without any external change to the product. This keeps it simple for the customers as external interfaces continues to be same as previous releases. The new engine is IBM’s strategic in-memory, columnar database with a much bigger SQL coverage and compatibility between Db2 for z/OS and the accelerator. It also promises faster data ingest for incremental updates leading to less HTAP query delay once HTAP is available on the new platform. We will not be diving deep into the specifics of the BLU engine in this blog, to learn more about it please refer to the following link - IBM Db2 with BLU Acceleration.

 

Deployment on IBM Z - Why is this exciting? Many customers are already heavily invested in IBM Z hardware. Having the ability to deploy the Analytics Accelerator on Z allows them to reuse the same infrastructure to run Analytics workloads. Also, the operational processes that have been built over many years around the Mainframe can be reused, which is a major advantage. Further, customers who did not find the need to invest in dedicated hardware for Db2 Analytics Accelerator, now have the option to accelerate analytical workloads using existing hardware and infrastructure without having to maintain a separate hardware stack. In other words, improvement in ROI with reduction in CAPEX can be achieved.

 

What’s in the package? IBM Db2 Analytics Accelerator for z/OS V7.1 uses Docker to containerize the accelerator server, the database engine and other components used for high availability, monitoring and security for easier deployment and setup. The IBM Secure Services Container contains the OS (to launch the Docker container) in addition to infrastructure management capabilities. This model makes it a self-contained software application that can be launched as an isolated entity on IBM Z14. Outside the container, the compute resources - CPU, Memory and Storage need to be provided by the customer based on usage requirements.

 pic3-1.JPG

 

What are the benefits? With this new model it is much easier to get started with deployment on Z and scale on demand. Scale-up is performed by adding memory, CPU and storage as needed to grow the install without having to reinstall. The deployment is faster and instances are currently limited to CPU resources in 1 physical drawer. This model is perfectly suited for organizations who prefer to start small using any available compute resources as well as for organizations who want to provide dedicated compute resources for analytical processing.

 

The set-up is a single container that runs multiple Db2 Warehouse partitions(MLNs) per LPAR. If multiple Accelerators are needed on a system, it can be installed on different LPARs. Based on usage requirements, each LPAR needs to be assigned CPU, storage and memory. Higher instance granularity is possible considering the fact that characteristics of Analytics Accelerator deployment on Z lead to more dedicated and smaller instances which is beneficial for many small test/dev instances that could be re-provisioned as needed.

 

To enable the deployment, IBM Z hardware has a new SSC (Secure services Container) partition. (The SSC provides is an infrastructure to build, deploy and manage IBM Z Systems Appliance.)

For IDAA, the SSC partition or LPAR encapsulates the IBM Secure Services Container that contains the Accelerator Server and other additional components. It is managed using the Hardware Management Console (HMC) for IBM Z.

 

Download, Install and Management

 

The latest version of the IBM Db2 Analytics Accelerator for z/OS can be downloaded from Fix Central using the following link - Download IBM Db2 Analytics Accelerator for z/OS-Deployment on IBM Z

 

  • The image can be downloaded to any workstation which has access to a web browser.
  • Define(provision) the SSC LPAR and any additional resources through the HMC Console.
    • Select the physical machine which will serve as host
    • Customize the activation profile
    • Define the number of processors to use in LPAR
    • Define Storage
    • Define network for Appliance Installer and IDAA Admin UI
    • Activate profile (clean memory and set up)
  • Then, from the Appliance Installer (a browser based interface) install and set up the IBM Secure Services Container (containing the Accelerator server and resources) from the downloaded image.

 

The SSC LPAR comes with an installer or loader which when pointed to the downloaded Accelerator Image puts it to the disk and reboots Analytics Accelerator from there.

 pic3-2.JPG

 

 

The HMC Console Dashboard

pic3-3.jpg

 

Appliance Installer UI

pic3-4.png

The above Interface allows the System Engineer to manage the Appliance.

 

Start small and scale on demand:

As part of some of the simple, getting started use cases, customers can use any existing IFLs and memory for smaller production, test or development environments. The minimum recommended configuration is 4 IFLs with 256 GB memory for test/dev and 8 IFLs with 512 GB memory for production. For scale up use cases, a dedicated z14 drawer with 35 IFLs and 2.56 TB of memory is recommended.

 

An example configuration when a dedicated z14 drawer is to be shared between production, test and dev can resemble the below image. It is better for non-production environments to share a certain number of IFL’s as opposed to splitting them. IFL’s can be shared but memory has to be individually assigned. The usage configuration can even be set up in such a way where for example, test environment can be given more capacity and priority as needed.

 pic3-5.png

 

The more memory, the better performance because the Db2 BLU Analytics Engine is an in-memory columnar data store and performs at its peak if everything that is needed to run data analytics can be kept in memory.

After initial deployment, configuration changes like adding/removing IFL’s or memory can be performed, both of which require shutdown and reactivation of the Accelerator. Changes to storage is also allowed which does not require any shutdown or reactivation.

 

In conclusion, the new deployment option opens up new markets and possibilities. This presents a great opportunity for both customers wanting to run small scale analytical workloads with data in Db2 for z/OS as well as customers who run massive Analytical workloads without needing any additional hardware for IDAA. Why not start small and see the power?

 

 

 

1 Like
Recent Stories
Db2 12 SQL Pagination

SQL Percentiles

Spring Batch using Db2