Virtualized Oracle RAC databases on premises with a cloud experience

[StorageExperts]

Learn about Oracle RAC database deployment using HPE GreenLake for private cloud using HPE Alletra dHCI

Your Oracle deployments demand efficient on-premises infrastructure with cloud management. Whether in your data center or edge site locations, deploy your Oracle database solution with the data availability, security, and flexibility needed for your business-critical Oracle workloads. 

– By Richard Wuehler, Solutions Engineer, HPE Storage

Solution overview

At the heart of this solution is HPE GreenLake for Private Cloud Business Edition using the power of HPE Alletra dHCI built on HPE Alletra Storage 6000 to create business-critical virtualized Oracle RAC database deployments with six-9s data availability.

Our recent technical paper details how to create and manage the Oracle infrastructure. I’m covering some of the highlights presented in the technical paper here, but I recommend you check out the complete paper for more details.

 Benefits of this solution

• Full-stack lifecycle management with catalogue-based updates
• Fully integrated server, storage, network, and virtualization solution
• Simplified storage performance-tuning using vSphere storage policies
• Cloud-based management with automated storage insight and maintenance features
• Simple and flexible to scale either compute or storage or both together

Now let’s look under the hood

Here’s the Alletra dHCI solution stack including HPE ProLiant servers, HPE switches, and HPE Alletra Storage 6000.

The Oracle RAC database builds off the highly scalable six-9’s data available solution. The Oracle database is clustered across two VM nodes and utilizes virtual datastore volumes for storage.  

Configuration guidance

Both the VMware vSphere Client and the HPE GreenLake Data Services Cloud Console (DSCC) were used to configure the solution. HPE GreenLake cloud provides context aware guidance at the Cloud Console Articles icon.

This solution entails provisioning a virtual machine with storage, installing the Oracle database, and creating a template. Major tasks include:

1. Deploy the Alletra Storage dHCI system and add it to the HPE GreenLake for Private Cloud Business Edition service area
2. Create storage provisioning policies for intent-based provisioning
3. Create datastores and a content library
4. Create and configure a virtual machine to create a template from
5. Create and store an OVF template in the Content Library
6. Create at least two virtual machines from the OVF template
7. Create anti-affinity rules to help keep the VMs on separate ESXi hosts

Critical settings

The virtualized RAC database storage requires specific VMware disk settings. These settings allow the shared storage to be accessed from more than one VM at the same time. Important settings and information include:

• Use Multi-writer for Oracle RAC cluster disks.
• VMware vSphere® vMotion® (live migration) has limited support depending on the ESX/ESXi version.
• VMware vSphere® Storage vMotion® is not supported (shut down all cluster VMs first).
• Snapshots have limited support (Set Oracle disks to Independent-Persistent disk mode to take snapshots of the OS disk only; use Oracle RMAN to back up the Oracle data).
• Virtual disks from vVols datastores can be thin or thick.

Here is an example of a clustered database disk with the correct settings applied.

For those not as familiar with a virtualized solution, be aware that multipathing is all handled by the hypervisor, which was configured during the Alletra dHCI installation process. While the multipath.conf file is not used, the udev permissions file can still be used to help set permissions on volumes not yet integrated into Oracle Grid. Anti-affinity rules are available to help keep the RAC VMs nodes on separate hosts. Keeping VMs on separate hosts is an important step in maintaining the HA capability of the RAC database.

Tuning

SwingBench was used to help confirm the environment as shown here. One benefit of a virtualized solution is that after running some tests, CPU and memory can be adjusted as needed. With Alletra 6000 enabled policies, quality of service can also be tuned with both IOPs and MB/s tuning is available.

The benefits add up

HPE GreenLake for Private Cloud Business Edition using HPE Alletra dHCI with intelligent features like intent-based provisioning enables faster on-premises virtual machine provisioning for Oracle RAC databases.

By separating storage from the server in the HPE Alletra dHCI, implementation benefits include:

• No need to overprovision the new solution for unknown future storage or compute needs
• Lifecycle management tools are a key part of the solution
• Mixed workloads can be combined on the flexible dHCI solution
• On-premises speed, security, and great ROI can be achieved.

Whether you’re an IT decision maker, infrastructure administrator, or database specialist, I recommend the complete technical paper, Oracle RAC using HPE GreenLake for Private Cloud Business Edition, to learn more about how to provision a virtualized Oracle RAC database solution using HPE GreenLake for Private Cloud Business Edition and HPE Alletra 6000.

---

Meet HPE Storage Experts blogger Richard Wuehler

Richard has been working with storage for over 25 years. Currently he is part of the HPE Storage Solutions Engineering team.

Storage Experts
Hewlett Packard Enterprise

twitter.com/HPE_Storage
linkedin.com/showcase/hpestorage/
hpe.com/storage