Built to Scale with VCF on VxRail and Oracle 19C RAC
Fri, 17 Apr 2020 05:21:03 -0000|
Read Time: 0 minutes
Built to Scale with VCF on VxRail and Oracle 19C RAC
The newly released Oracle RAC on Dell EMC VxRail with VMware Cloud Foundations (VCF) Reference Architecture (RA) guides customers to building an efficient and high performing hyperconverged infrastructure to run their OLTP workloads. Scalability was the primary goal of this RA, and performance was highlighted as the numbers were generated. As Oracle RAC scaled, TPM increased to over 1 million TPM, while read IOPs showed sub-milli-second (0.64-0.70 ms) performance. The performance achieved with VxRail is a great added benefit to the core design points for Oracle RAC environments of which the primary focus is the availability and resiliency of the solution. Links to a reference architecture (“Oracle RAC on VMware Cloud Foundation on Dell EMC VxRail”) and a solution brief (“Deploying Oracle RAC on Dell EMC VxRail “) are available here and at the end of this post.
The RAC solution with VxRail scaled-out easily — you simply add a new node to join an existing VxRail cluster. The VxRail Manager provides a simple path that automatically discovers and non-disruptively adds each new node. VMware vSphere and vSAN can then rebalance resources and workloads across the cluster, creating a single resource pool for compute and storage.
The VxRail clusters were built with eight P570F nodes; four for the VCF Management Domain and four for the Oracle RAC Workload Domain.
Specifics on the build, including the hardware and software used, are detailed within the reference architecture. It also provides information on the testing, tools used, and results.
This graph shows the performance of TPM and Response Time when increasing the RAC node count from one to four. Notice that the average TPM increased with near-linear trendline (shown by the dotted line) as additional RAC nodes were added, while total application response time was maintained at 20 milliseconds or less.
Note: TPM near-linear trendline is shown in the above graph (blue dotted line), As additional RAC nodes are added, an increase in performance is seen as well as an increase in RAC high availability. TPM linear performance (scale equal performance per each note) growth is not achieved due to RAC nodes’ dependency on concurrency of access, instance, network, or other factors. See the RA for additional performance related information.
Summary of performance
Different-sized databases kept the TPM at the same level (about one million transactions) while keeping the application response time at 20ms or below. When increasing the database size, the physical read and write IOPS increased near-linearly, as reported from the Oracle AWR. This indicated that more read and write I/O requests were served by the backend storage, under the same configuration. Overall, when the peak client IOPS was up to 100,000, vSAN still provided excellent storage performance at sub-milliseconds at read and single-digit milliseconds latency at write.
Sidebar about Oracle licensing: While not mentioned in the RA; the VxRail offers several facilities to both control Oracle licenses and in some cases eliminates the need for costly licensed options. These include a broad choice of CPU core configurations, some with fewer cores and higher processing power per core, to maximize the customer’s Oracle workload performance while minimizing the license requirements. Costly add on options such as encryption and compression can be provided via vSAN and are handled by VxRail. Further, and the vSphere hypervisor features, like DRS, allow Oracle VMs to be contained to only licensed nodes.
You can speak to a Dell Technologies’ Oracle specialist for more details on how to control Oracle licensing costs for VMware environments.
Oracle Database 19c on VxRail offers customers performance, scalability, reliability, and security for all their operational and analytical workloads. The Oracle RAC on VxRail test environment was first created to highlight the architecture. It also had the added benefit of showcasing the great performance VxRail delivers. If you need more performance, it is simple to adjust the configuration by adding more VxRail nodes to the cluster. If you need more storage, add more drives to meet the scale required of the database. Dell Technologies has Oracle specialists to ensure the VxRail cluster will meet the scale and performance outcomes desired for Oracle environments.
Reference Architecture - Oracle RAC on VMware Cloud Foundation on Dell EMC VxRail
Solution Brief - Deploying Oracle RAC on Dell EMC VxRail
Author: Vic Dery, Senior Principal Engineer, VxRail Technical Marketing
Special thank you to David Glynn for assisting with the reviews
Related Blog Posts
VMware Cloud Foundation on Dell EMC VxRail Integration Features Series: Part 1 -- Full Stack Automated LCM
Fri, 03 Apr 2020 21:39:05 -0000|
Read Time: 0 minutes
VMware Cloud Foundation on Dell EMC VxRail Integration Features Series: Part 1
Full Stack Automated Lifecycle Management
It’s no surprise that VMware Cloud Foundation on VxRail features numerous unique integrations with many VCF components, such as SDDC Manager and even VMware Cloud Builder. These integrations are the result of the co-engineering efforts by Dell Technologies and VMware with every release of VCF on VxRail. The following figure highlights some of the components that are part of this integration effort.
These integrations of VCF on VxRail offer customers a unique set of features in various categories, from security to infrastructure deployment and expansion, to deep monitoring and visibility that have all been developed to drive infrastructure operations.
Where do these integrations exist? The following figure outlines how they impact a customer’s Day 0 to Day 2 operations experience with VCF on VxRail.
In this series I will showcase some of these unique integration features, including some of the more nuanced ones. But for this initial post, I want to highlight one of the most popular and differentiated customer benefits that emerged from this integration work: full stack automated lifecycle management (LCM).
VxRail already delivers a differentiated LCM customer experience through its Continuously Validated States capabilities for the entire VxRail hardware and software stack. (As you may know, the VxRail stack includes the hardware and firmware of compute, network, and storage components, along with VMware ESXi, VMware vSAN, and the Dell EMC VxRail HCI System software itself, which includes VxRail Manager.)
With VCF on VxRail, VxRail Manager is integrated natively into the SDDC Manager LCM management framework through the SDDC Manager UI, and through VxRail Manager APIs for LCM by SDDC Manager when executing LCM workflows. This integration allows SDDC Manager to leverage all of the LCM capabilities that natively exist in VxRail right out of the box. SDDC Manager can then execute SDDC software LCM AND drive native VxRail HCI system LCM. It does this by leveraging native VxRail Manager APIs and the continuously validated state update packages for both the VxRail software and hardware components.
All of this happens seamlessly behind the scenes when administrators use the SDDC Manager UI to kick off native SDDC Manager workflows. This means that customers don’t have to leave the SDDC Manager UI management experience at all for full stack SDDC software and VxRail HCI infrastructure LCM operations. How cool is that?! The following figure illustrates the concepts behind this effective relationship.
For more details about how this LCM experience works, check out my lightboard talk about it!
Also, if you want to get some hands on experience in walking through performing LCM operations for the full VCF on VxRail stack, check out the VCF on VxRail Interactive Demo to see this and some of the other unique integrations!
I am already hard at work writing up the next blog post in the series. Check back soon to learn more.
Twitter - @vwhippersnapper
The Dell Technologies Cloud Platform – Smaller in Size, Big on Features
Wed, 20 May 2020 13:07:08 -0000|
Read Time: 0 minutes
The latest VMware Cloud Foundation 4.0 on VxRail 7.0 release introduces a more accessible entry cloud option with support for new four node configurations. It also delivers a simple and direct path to vSphere with Kubernetes at cloud scale.
The Dell Technologies team is very excited to announce that May 12, 2020 marked the general availability of our latest Dell Technologies Cloud Platform release, VMware Cloud Foundation 4.0 on VxRail 7.0. There is so much to unpack in this release across all layers of the platform, from the latest features of VCF 4.0 to newly supported deployment configurations new to VCF on VxRail. To help you navigate through all of the goodness, I have broken out this post into two sections: VCF 4.0 updates and new features introduced specifically to VCF on VxRail deployments. Let’s jump right to it!
VMware Cloud Foundation 4.0 Updates
A lot great information on VCF 4.0 features was already published by VMware as a part of their Modern Apps Launch earlier this year. If you haven’t caught yourself up, check out links to some VMware blogs at the end of this post. Some of my favorite new features include new support for vSphere for Kubernetes (GAMECHANGER!), support for NSX-T in the Management Domain, and the NSX-T compatible Virtual Distributed Switch.
Now let’s dive into the items that are new to VCF on VxRail deployments, specifically ones that customers can take advantage of on top of the latest VCF 4.0 goodness.
New to VCF 4.0 on VxRail 7.0 Deployments
VCF Consolidated Architecture Four Node Deployment Support for Entry Level Cloud (available beginning May 26, 2020)
New to VCF on VxRail is support for the VCF Consolidated Architecture deployment option. Until now, VCF on VxRail required that all deployments use the VCF Standard Architecture. This was due to several factors: a major one was that NSX-T was not supported in the VCF Management Domain until this latest release. Having this capability was a prerequisite before we could support the consolidated architecture with VCF on VxRail.
Before we jump into the details of a VCF Consolidated Architecture deployment, let's review what the current VCF Standard deployment is all about.
VCF Standard Architecture Details
This deployment would consist of:
- A minimum of seven VxRail nodes (however eight is recommended)
- A four node Management Domain dedicated to run the VCF management software and at least one dedicated workload domain that consists of a three node cluster (however four is recommended) to run user workloads
- The Management Domain runs its own dedicated vCenter and NSX-T instance
- The workload domains are deployed with their own dedicated vCenter instances and choice of dedicated or shared NSX-T instances that are separate from the Management Domain NSX-T instance.
A summary of features includes:
- Requires a minimum of 7 nodes (8 recommended)
- A Management Domain dedicated to run management software components
- Dedicated VxRail VI domain(s) for user workloads
- Each workload domain can consist of multiple clusters
- Up to 15 domains are supported per VCF instance including the Management Domain
- vCenter instances run in linked-mode
- Supports vSAN storage only as principal storage
- Supports using external storage as supplemental storage
This deployment architecture design is preferred because it provides the most flexibility, scalability, and workload isolation for customers scaling their clouds in production. However, this does require a larger initial infrastructure footprint, and thus cost, to get started.
For something that allows customers to start smaller, VMware developed a validated VCF Consolidated Architecture option. This allows for the Management domain cluster to run both the VCF management components and a customer’s general purpose server VM workloads. Since you are just using the Management Domain infrastructure to run both your management components and user workloads, your minimum infrastructure starting point consists of the four nodes required to create your Management Domain. In this model, vSphere Resource Pools are used to logically isolate cluster resources to the respective workloads running on the cluster. A single vCenter and NSX-T instance is used for all workloads running on the Management Domain cluster.
VCF Consolidated Architecture Details
A summary of features of a Consolidated Architecture deployment:
- Minimum of 4 VxRail nodes
- Infrastructure and compute VMs run together on shared management domain
- Resource Pools used to segregate and isolate workload types
- Supports multi-cluster and scale to documented vSphere maximums
- Does not support running Horizon Virtual Desktop or vSphere with Kubernetes workloads
- Supports vSAN storage only as principal storage
- Supports using external storage as supplemental storage for workload clusters
For customers to get started with an entry level cloud for general purpose VM server workloads, this option provides a smaller entry point, both in terms of required infrastructure footprint as well as cost.
With the Dell Technologies Cloud Platform, we now have you covered across your scalability spectrum, from entry level to cloud scale!
Automated and Validated Lifecycle Management Support for vSphere with Kubernetes Enabled Workload Domain Clusters
How is it that we can support this? How does this work? What benefits does this provide you, as a VCF on VxRail administrator, as a part of this latest release? You may be asking yourself these questions. Well, the answer is through the unique integration that Dell Technologies and VMware have co-engineered between SDDC Manager and VxRail Manager. With these integrations, we have developed a unique set of LCM capabilities that can benefit our customers tremendously. You can read more about the details in one of my previous blog posts here.
VCF 4.0 on VxRail 7.0 customers who benefit from the automated full stack LCM integration that is built into the platform can now include in this integration vSphere with Kubernetes components that are a part of the ESXi hypervisor! Customers are future proofed to be able to automatically LCM vSphere with Kubernetes enabled clusters when the need arises with fully automated and validated VxRail LCM workflows natively integrated into the SDDC Manager management experience. Cool right?! This means that you can now bring the same streamlined operations capabilities to your modern apps infrastructure just like you already do for your traditional apps! The figure below illustrates the LCM process for VCF on VxRail.
VCF on VxRail LCM Integrated Workflow
Introduction of initial support of VCF (SDDC Manager) Public APIs
VMware Cloud Foundation first introduced the concept of SDDC Manager Public APIs back in version 3.8. These APIs have expanded in subsequent releases and have been geared toward VCF deployments on Ready Nodes.
Well, we are happy to say that in this latest release, the VCF on VxRail team is offering initial support for VCF Public APIs. These will include a subset of the various APIs that are applicable to a VCF on VxRail deployment. For a full listing of the available APIs, please refer to the VMware Cloud Foundation on Dell EMC VxRail API Reference Guide.
Another new API related feature in this release is the availability of the VMware Cloud Foundation Developer Center. This provides some very handy API references and code samples built right into the SDDC Manager UI. These references are readily accessible and help our customers to better integrate their own systems and other third party systems directly into VMware Cloud Foundation on VxRail. The figure below provides a summary and a sneak peek at what this looks like.
VMware Cloud Foundation Developer Center SDDC Manager UI View
Reduced VxRail Networking Hardware Configuration Requirements
Finally, we end out journey of new features on the hardware front. In this release, we have officially reduced the minimum VxRail node networking hardware configurations required for VCF use cases. With the introduction of vSphere 7.0 in VCF 4.0, admins can now use the vSphere Distributed Switch (VDS) for NSX-T. The need for a separate N-VDS switch has been deprecated. So why is this important and how does this lead to VxRail node network hardware configuration improvements?
Well, up until now, VxRail and SDDC management networks have been configured to use the VDS. And this VDS would be configured to use at least two physical NIC ports as uplinks for high availability. When introducing the use of NSX-T on VxRail, an administrator would need to create a separate N-VDS switch for the NSX-T traffic to use. This switch would require its own pair of dedicated uplinks for high availability. Thus, in VCF on VxRail environments in which NSX-T would be used, each VxRail node would require a minimum of four physical NIC ports to support the two different pairs of uplinks for each of the switches. This resulted in a higher infrastructure footprint for both the VxRail nodes and for a customer’s Top of Rack Switch infrastructure because they would need to turn on more ports on the switch to support all of these host connections. This, in turn, would come with a higher cost.
Fast forward to this release -- now we can run NSX-T traffic on the same VDS as the VxRail and SDDC Manager management traffic. And when you can share the same VDS, you can get away with reducing the number of physical uplink ports to provide high availability down to two and reduce the upfront hardware footprint and cost across the board! Win win! The following figure highlights this new feature.
NSX-T Dual pNIC Features
Well, that about sums it all up. Thanks for coming on this journey and learning about the boat load of new features in VCF 4.0 on VxRail 7.0. As always, feel free to check out the additional resources for more information. Until next time, stay safe and stay healthy out there!