VMware Validated Design (VVD) VxRail for the Software Defined DataCenter (SDDC) is a prescriptive architecture whose value customers easily recognize and appreciate, but in some cases may also need to deviate-from based on their own requirements. One such use case is where existing VxBlock customers want to leverage VVD on VxRail but also want to retain their existing infrastructure.
The Technical Problem: VVD has a Management Domain which contains all of the components required to manage the SDDC, including the vCenter Server and NSX Manager instances from which the Compute Workload Domains are managed.
VxBlock systems use an AMP (Advanced Management Platform) within which their own vCenter Server must reside.
Right away we have an alignment problem, with that problem being that, today, we cannot fully align VxBlock to the tested, validated and documented VVD architecture!
So let’s be clear in this post, that VVD and VxBlock is NOT VVD-aligned, therefore is not a VVD-certified or VVD-supported architecture.
But that doesn’t mean that customers cannot leverage VVD with VxBlock. They absolutely can, but that brings us to another problem for some customers. The overall solution comes under the label of ‘Custom’. Yea, the C-word. And while some customers really do not like having the word Custom associated with their solutions, the overall solution while technically Custom, actually combines to two very comprehensive validated offerings.
So let’s look at what is technically possible for a solution which will meet customer requirements while providing customers with the best of both worlds (well, mostly!).
Being able to align to a fully tested, validated and documented Support Matrix or Bill of Materials (BOM) is crucial to many customers. It’s not simply a collection of off the shelf components. Vast amount of hours go into the comprehensive test and validation. And many more go into documenting the whole thing!!! VVD and VxBlock provide exactly that with their respective VVD and RCM hardware and software matrices.
We are coming at this from the top of the vRealize stack down. And by the Top, we mean vRealize Automation. We want to be able to have a VVD-aligned, single cloud management platform from which we can consume our VxBlock infrastructure (and Public cloud if and as required!).
At a high level, the suggested solution is to use VVD on VxRail for the CMP, from which we can point vRA, vROps, Log Insight, vRealize Business for Cloud at the VxBlock as an endpoint.
(This is also applicable for other CI/HCI platforms where the vCenter and NSX Manager instances must be local to the Infrastructure)
What is required to manage and consume the VxBlock infrastructure as an endpoint?
Each endpoint must consist of the following:
- Dedicated vCenter Server / PSC
- Individual SSO Domain for per endpoint
- vRA Agents
- vROps Collectors
- Log Insight Forwarders
- vRealize Business for Cloud Remote Data Collectors
- NSX Manager
- NSX Controllers
- NSX Enterprise and Tenant Edges
- NSX Distributed Logical Routers
A standard VVD on VxRail deployment consists of a minimum of 2 x 4-node VxRails, with the first VxRail hosting the CMP and the second VxRail hosting the Shared Edge components as well as the Tenant workload domain.
In this model using a single 4-node VxRail for CMP only, each endpoint must provide its own vCenter/PSC, NSX infrastructure and operate as its own workload domain.
A standard VxBlock instance will have the required vCenter and NSX instances in place on the AMP, so what remains would be to deploy the required agents, collectors and forwarders in the AMP. The requirements are very similar (though not the same!) to what is required for the VVD ROBO.
Benefits of doing this?
- Customer avails of the solution they require
- Single tested and validated CMP to manage private and public cloud resources
- Retain all benefits of VVD on VxRail for CMP
- Test, validation, build guides for CMP
- CMP Life-cycle with vRealize Suite Lifecycle Manager (vRSLCM)
- Retain all benefits of VxBlock
- Fully tested and validated Release Certification Matrix (RCM)
- Retain full support of both VMware and DellEMC for all components
- Decoupling of Management Platform from the workload infrastructure
- Can life-cycle the CMP independently of the Infrastructure so long as vRA supports the vSphere and vCenter versions of the endpoint e.g. VxBlock RCM
- Can easily expand endpoints with additional infrastructure
- If Multi-Site DR is required, then Multi-Region VVD on VxRail can be applied for the CMP using vSphere Replication and Site Recovery Manager (SRM), while the VxBlocks can use SRM with the relevant Storage Replication Adapters (SRAs), or whichever replication and recovery technology is preferred, as using the VxBlock as a Compute Workload Domain is not bound by VVD constraints.
- Similarly for Continuous Availability, the VVD on VxRail CMP can be stretched as per the Multi-AZ VVD, while the VxBlock workloads can be stretched and protected by VPLEX.
While this combined solution is being adopted by a number of customers globally in order to meet their requirements, please keep in mind that this is not VVD-certified, and does not come with a ready-made end-to-end build guide as with VVD-certified solutions. The associated VVD and VxBlock teams can stand behind their fully tested and validated BOMs for CMP and CI respectively, but what is described above is 2 independent solutions being combined for a (cough) Custom (cough) solution.
Hope that helps, and thanks for reading.
I’m off to get something for that nasty cough I seem to have developed 🙂