Over the past few days, quite a few articles have been appearing about VMware vCloud Director, including some excellent networking articles from Duncan Epping yellow-bricks.com and a great architecture piece from Frank Dennemanfrankdenneman.nl but I havent seen anything about how to deploy a VMware vCloud Director environment with resilience or high availability, so thought I would write an overview of the basic design.
So within the VMware Cloud Team we have taken a building block approach to deploying vCloud Infrastructure. By using this approach a resilient infrastructure can be guaranteed. A building block approach will enable the environments to easily scale horizontally. Building blocks within a vCD environment are referred to as “Resource Groups”.
A vCD environment consists of all components needed to run a VMware vCloud excluding the underlying vSphere infrastructure:
• VMware vCloud Director (vCD)
• VMware vShield Manager
• VMware Chargeback
• VMware vCenter (supporting the Resource Groups)
The diagram below shows how all the components fit together. I have posted this previously, but thought it would be beneficial to show again.
A key tenet of a cloud architecture is resource pooling and abstraction. vCD further abstracts the virtualised resources presented by vSphere. vCD provides two logical constructs that map to vSphere logical resources:
• Organization – Organizational unit to which resources(vDC’s) are allocated
• Provider Virtual Datacenter – vSphere resource groupings that power vDCs, further segmented out into organization vDCs
• Virtual Datacenter (vDC) – Deployment environments, scoped to an organization, in which virtual machines run and is a subset of a Provider vDC
• External Network – A customers connection to an external network
• Network Pool – A grouping of network segments used by Org and vApp network
• Org Network – Intra cloud network which uses resources (segment) from the Network Pool.
• vApp Network – Intra cloud network which uses resources (segment) from the Network Pool.
The following diagram depicts how a physical environment would look like corresponding to the logical design that has been shown in the previous diagrams.
As per the diagram, you have 3 Clusters:-
Cluster01 is for the Management VM’s and Database VM’s, including vCD, vCenter02 and vCenter03.
Cluster02 and Cluster03 are the resource clusters which are used for the Provider pVDC Clusters.
The physical infrastructure consists of the underlying vSphere components VMware vSphere ESXi 4.1 and related components.
This article is to help provide a basic understanding of how you would design and deploy your own cloud environments for use within your own Development and Proof of Concept environments.
Nice work. On your physical diagram you might want to extend another line from the storage to the second aggregate switch. I realise that it says “switches” plural, but the way you split the ESX hosts across both sets of aggregate switches make it look you are missing redundancy.
Anyway, nice overview that I’m sure lots of people will benefit from.
Thanks for the tip, will look to update the diagram
VCD with vshiled have no redundancy , if VSE instance is lost (the VM down, host down) all the underlying networks and VMs are down, this takes 3-4 minutes or more until new VSE is lunched somewhere else, there is no stateful protocol to keep that gateway IP ‘alive’ like in normal networking. This was verified by vmware , i would not use VCD in production with this limitation, also because you have tons of VSE instances running (one per port-group, many for the same ORG)
basically application stops working when 4 minutes downtime, many times you need to restart them, this product can not go production in DC at this point.
I recognise most of that diagram that I created whilst at VMware 😉 the one with the three clusters. Kev James…… Some nice content though
Haha yeah, the last diagram ended up in vCAT 😉 I claim the top diagram though 🙂 this is an old post now.