Kevin Remde's IT Pro Weblog
In part 10 of our series today, I wanted to kick of a sort-of-sub-series of my own. During our IT Camps this month (our afternoon portion of our TechNet Events), I’ve been taking some time to show off my personal test environment, and how I set it up. I’m finding that there is a lot of interest in learning and seeing just how easy it is to build a private test environment using cheap hardware and evaluation software. And since I have had some pretty good success in doing just that, I thought I would document the process for you all.
So today I’m going to outline some of the goals in building what I’m calling my “Private Private Cloud”. Here is what I hoped to achieve as my desired outcome:
In my role at Microsoft I have been fortunate in a couple of key ways:
For various presentation / event / launch needs, hardware has come our way to support it. And though I would likely have to return if I left Microsoft for any reason, I am able to keep it for now. The need for a good test environment is putting this older hardware to good use.
To support my test lab, I am using 3 Lenovo laptop computers. (You could do it with 2, and I’ll outline that in the post that talks specifically about the hardware setup.) These are all connected to my test network using a 1Gbps Linksys Router.
Here is a diagram showing my configuration (click to enlarge):
The two T61Ps at the bottom of my simple diagram are the machines that make up my 2-node failover cluster. Also, to support a large quantity of shared storage, I have a 2TB drive (7200RPM) connected via USB3 to my W520 “storage server”.
Short of having a TechNet Subscription (which I highly recommend – for many reasons – but more relevant to this topic because the evaluation software doesn’t time-out), you have everything you need at your disposal FOR FREE for building a private cloud based on Microsoft’s solutions. Check out http://aka.ms/evals for a list of and links to the evaluation software that you need for building out our test lab. Most importantly, you’ll be starting with Windows Server 2008 R2 SP1.
Additionally, in order to build the shared-storage for supporting Windows Failover Clustering, you’ll need to download the Microsoft iSCSI Software Target. This is what we’ll install on the “server” that we’re designating as our “storage server”.
On all three of my “servers” I have added the Hyper-V role. Originally my test server was just the W510 (top of the diagram), and I have two SSDs (Solid State Drives) for the sake of really fast OS and Virtual Machines; so it’s still a good place to build and support faster-to-build + faster-performing demo examples. And on the two T61Ps I installed the Hyper-V role also, as they are the two nodes of my Hyper-V cluster.
“Hey Kevin, what does Hyper-V cost?”
Ah.. Virtualization from Microsoft is simply included with Windows Server. So is Windows Failover Clustering. And so is Live Migration. (How much are you paying VMware for vMotion?)
Speaking of failover clustering.. that also was a goal of this test environment. Until recently the option to build relatively cheap (free?) shared storage for implementing failover clustering was non-existent. But now with the ability to install the free Microsoft iSCSI Software Target, it’s easy to create and become familiar with your own high-availability test bed. So I wanted to take advantage of that here.
“What is CSV?”
CSV stands for Cluster Shared Volumes. This is a technology introduced in Windows Server 2008 R2 for the sake of clustered Hyper-V nodes that supports shared access to files. Once you have Windows Failover Clustering configured, you can dedicate some storage to be used for CSV, which is then the recommended location for your virtual machine files that will represent HA (highly available) virtual machines.
Because the live migration (the moving of a running virtual machine from one cluster node to another with zero downtime) requires Failover Clustering, which in turn requires shared storage, it wasn’t easy until recently to implement this in a test-lab-on-a-budget. But now we can.
Key components of a true private cloud are the tools to manage and support it. Virtualization alone does not a Private Cloud make. So to fully implement my own private private cloud, I need to acquire, install, and work with the System Center products; both the current set as well as the ones coming in the System Center 2012 wave. And fortunately, like the Operating System, you can find links to free trials, betas, and release candidates at http://aka.ms/evals.
So there you have both the goals and some of the higher-level components of my own personal “mad house of cloud”. Watch for future posts (and perhaps some screencasts) where I’ll take you through step-by-step in the building of my private private cloud.
If you have missed any of the series posts, check out my summary post for links to all of the articles available so far at http://aka.ms/cloudseries.