Summary
VMware Tanzu Architecture for Dell EMC VxRail is a carefuly considered, fully inclusive and regularly validated approach to using appropriate hardware and software elements to produce a highly scalable PaaS installation for today and into the future as you grow your system. The maturity path to follow starts with a minimally viable, low cost platform that features all the capability needed for small and remote installations and concludes with an Enterprise-grade, full scale and expandable system that can be grown and upgraded over time without re-architecture. These models can be used and reused over and over around the business as needs arise. You can use these with confidence, as each edition has been fully tested on real world, full scale hardware exactly as described. You can be confident that it will operate as described. Each validation cycle tests the target platform for functionality and performance.
TAS for VMs and TKGI Installation
A fully meshed TKGI and TAS for VMs installation based on best practices and reference design considerations will look as follows:
View a larger version of this diagram.
The common elements of these two platform technologies are the NSX T0 router and the associated T1 routers. This approach allows for any possible cross traffic between TKGI and TAS for VMs applications to stay within the bounds of the T0 router. Cross traffic does not need to exit the NSX overlay. This also provides a convenient, singular access point to the whole installation, making deployments of multiple, identical installations easier to automate.
Further Considerations
With a single vSphere Cluster, the Host Groups feature introduced in Ops Manager v2.7 can be used in conjunction with Resource Pools in vSphere Clusters as AZ constructs to stack different installations of PaaS. Host Groups provide a way to pin hardware resources to an AZ. Resource Pools provides a way to aggregate host resources.
As server capacity continues to increase, the efficiency of deploying independent clusters of a server just for one installation of one product is low. As customers are commonly deploying servers approaching 1 terabyte of RAM, stacking many installations in these clusters improves overall resource utilization.
You may be tempted to split the TAS for VMs and TKGI installations into separate network parent/child configurations, behind separate T0 routers. Before doing this, review VMware’s best practices for T0 to T0 routing efficiencies and weaknesses to ensure that the approach meets your needs.
