How best to showcase the power of a Private Cloud? Build one onsite at TechEd North America and run Hands on Labs on it! Ronald Beekelaar in the video below talk about how they leveraged Hyper-V and Systems Center in a Private Cloud to deploy, monitor, and manage a system that delivers thousands and thousands of Hyper-V virtual machines for TechEd attendees to complete labs. The talk was briefly about the solution and how its evolved from year to year then Ronald took us over and showed us the HP servers that are running the entire infrastructure.
They built us a Private Cloud to host all of the demo vms for the SIM TLC Pods. Using System Center Virtual Machine Manager 2012 Beta, they created a “Cloud” for each POD, and we were able to log into the system and only see the VMs we had access to, but could manage them ourselves as needed. This showed the true tenants of Private Cloud. We had a shared environment, that was elastic as resource needs changed, it allowed for Self Service, and we were metered by use.
Here is a description of the solution:
TechEd offers our demo stations a workstation class machine with 8GB RAM. Product teams in the System Center and Identify Management tracks are showing environments consisting of dozens of VM’s at times with high disk and memory requirements. Examples.
1. Service Manager: 9 VM’s, 40GB RAM, 450GB Storage
2. Orchestrator: 24 VM’s, 40GB RAM, 250 GB Storage
In addition, we wanted to showcase some of Microsoft key private cloud solutions and how we are actually using those solutions to solve our own challenges.
Sponsor HP provided two C7000 bladecenter systems containing 4 blades each. Blades were 72GB RAM per blade, with SSD local storage. The enclosures were connected together with a 4GB fiber backplane for storage and 10GB Ethernet.
Storage was provided by an HP EVA storage array containing 24 400GB SAS drives configured as RAID 10.
Storage and Hyper-V hosts were configured and deployed using VMM. Entire build time for the solution was 8 hours, including operating system installation, storage provisioning, cluster creation and management.
Each bladecenter was configured as a failover cluster (named StormCloud and Cyclone). Storage was 4GB Fiber with multiple paths of redundancy.
Individual demo VM’s were deployed as highly available resources and assigned a primary node. Live migration was used to move running demo environments between nodes as needed for maintenance. In addition this allowed the intelligent placement and dynamic optimization features in SCVMM 2012 to ensure that all demo environments were equally balanced across all nodes.
SCVMM 2012 includes the ability to present physical resources as “clouds”. A cloud is an abstract view of resources that you own, which hides all aspects of the underlying physical implementation. In addition, clouds can be shaped to ensure that no one cloud or cloud owner can consume more resources than they have been allocated.
1. SCVMM was used to bare metal deploy the Hyper-V hosts.
2. SCVMM was used to allocate storage to the servers.
3. SCVMM was used to deploy and configure the cluster.
4. SCVMM was used to create clouds representing slices of the overall fabric (hardware) to be allocated to users.
5. SCVMM PowerShell was used to create the clouds and self-service roles as a batch.
6. SCVMM PowerShell was used to manage cloud and VM properties.
Sample: Get-VM –Name *ERDC* | Set-VM –Cloud (Get-SCCloud –Name *Orchestrator*) was used to assign all VM’s in a group to a cloud.
For each demo station, three steps are performed.
1. A cloud is created.
2. A self-service user is created.
3. The VM’s are dynamically provisioned to the correct cloud.
Each demo station owner is provided a standard Windows desktop and a copy of VMM 2012 administration console. When a demo station owner launches the VMM 2012 administration console, they are prompted to choose a self-service role. Each self-service role is mapped to one or more clouds with specific rights. The following series of screenshots depicts aspects of the solution.
This was a fantastic solution for the SIM Pods, and allowed us to showcase our solutions on our solution. Some key Benefits Realized
1. Users do not need to know anything about the physical implementation to access resources. All details of physical implementation are fully hidden from the cloud manager, while still giving the cloud manager administrative access to all resources.
2. Resources are automatically highly available and fault tolerant. VM’s are migrated between cluster nodes as needed for maintenance and capacity management.
3. Deployment and setup was very quick and easy. Platform was deployed from bare metal to fully operational in under 8 hours.
4. Automation via PowerShell has resolved most troubleshooting and support issues. Powershell scripts were used for repeated tasks such as cloud creation, troubleshooting, and VM management.
5. SCVMM offers extensive control over user resource consumption allowing fair managed allocation or resources.
If you would like for information on the solution, please contact HynesITe at the details below, or comment here. We would love to hear your feedback.