Tuesday, March 6, 2012

Storage Replication Configuring HA for VM - Part 6

Achieve high availability by investing on technology

Introduction
Who would like to be called during the midnight when you are having rest because of the system interruption due to hardware availability and etc. With high availability, user can continue to access the system without knowing there is a downtime happen. Look at the following configuration whereby you will see how to configure a high availability VM with the storage replication aka SteelEye. Once you have configure this, the VM can perform live migration or quick migration on a volume without any shared storage. The best practice is to have 1 VM per volume, which mean if your hyper-V host is loaded with the drive A(floppy Drive), C(system drive) and D(CD ROm), you just can host another 23 VM.

 What would happen if you host 2 VM in a single volume? 
The detail would be this, storage resources only can be assign to one VM and not to 2. What happen if you are performing live migration on vmA from HyperV-A to HyperV-B, you will be expecting a single ping drop for vmA. The volume in HyperV-A will be lock with the vmB still running. Once you are trying to perform quick migration or live migration on vmB to HyperV-B the system will be restarted and corruption might could happen. This is just a test to know what would happen, and in the real production server of course you don't want to do that:).
 
Configuration
In my previous article I have cover the following, configuring the storage replication and also configuring the cluster. On this article I will cover what need to be configure in order the high availability VM to work with the steeleye.

As on the screen below, we need to create an empty service to host the steeleye information. Since the live migration just work on a specific volume. You might want to skip this configuration and direct configure the steeleye datakeeper volume it self.

Integrate Steeleye into windows cluster

The screen above is nothing special, as this would be the standard way of creating a high availability VM on the hyperV host.
Configure HA on Windows cluster

You need to choose the type of application, in this case we will choose virtual machines.
Configure VM as the high availability resources

select the VM that you want it to be part of the high availability.
Configure VM as the high availability resources

Click next.
Configure VM as the high availability confirmation

Below will be the summary of the configuration, you can ignore the warning.
Configure VM as the high availability

This is how it look like once we have configure the HA option. The next step will be adding the volume resources to be part of the VM.
Configure dependency for VM

Adding datakeeper volume to the VM.
Configure steeleye as part of the cluster volume

By default the service is offline, you will need to choose the drive first and bring it online manually.
Configure the mirror volume as part of the datakeeper volume

In my case, I have use drive Z as my mirror for replication. In some cases, you might not able to see any volume there. What you can do is to delete the datakeeper in the storage area under the failover cluster manager.
Overview of datakeeper volume

The next is to configure dependency on the VM it self.

Select the datakeeper volume and press OK.

Now we are good to go, we can test the live migration


Reference
Part 1 : Installing SteelEye Storage Replication software
Part 2 : Consideration Of Storage Replication 
Part 3 : Scenario of Storage Replication
Part 4 : Configure Storage Replication mirror
Part 5 : Configure cluster
Part 6 : Configure high availability VM(Live migration without shared storage)
Summary 
Let us know what you though , learns and hope for next articles!. Connect with us on GOOGLE+ , TWITTER and FACEBOOK.

No comments:

Post a Comment