Managing and protecting all enterprise data


Storage Bin 2.0: Virtualization challenges

Some of the biggest challenges in the data center revolve around backing up and recovering data in a virtual machine environment.

Backing up and recovering data from virtual machines is easier said than done.

Last month, I used this space to discuss the pervasiveness of virtual machines (VMs) in the data center and how they change practically everything. For many, VMs are a no-brainer value proposition. But let's not forget that the devil is in the details.

If you use VMs within a Fibre Channel (FC) network, you may have a bit of a problem. FC is a Layer 2 protocol and every VM in a physical server shares a common World Wide Name (WWN). That's like everyone in your firm sharing the same phone number or email address--everyone sees everything. FC host bus adapter (HBA) vendors fixed this problem through N_Port ID Virtualization (NPIV), which lets you assign a different WWN to each VM. However, you have to upgrade all of your HBA drivers. No big deal, right? Wrong. iSCSI and NFS don't have this problem because these protocols are IP based and you can assign different IP addresses to your VMs.

Some of the biggest challenges in data centers revolve around backing up and recovering data in a VM environment. There are so many issues to address I could dedicate a book to the topic. Let's look at a select few.

Backup agent in the guest OS:

  • May be burdensome on the host's shared resources, especially if multiple VMs share the same host resources and are scheduled at the same time.
  • Requires setting up backup scheduling and policies, as well as necessitating a client agent license, for each virtual machine.
  • Lacks bare-metal recovery options, so a VM can't be restored as a whole.
  • Could scramble the VMs and cause performance degradation when the backup process is initiated.
  • Running agents in the guest operating system can lead to backup agent sprawl (a management quagmire).

Backup agent in the hypervisor:
  • No file-level recovery.
  • No individual VM-level backup or recovery.
  • Doesn't support enterprise applications such as Exchange, Oracle and SQL Server.
Infrastructure management is another concern. Most of the management tools we use were built for physical systems. Each has its own limitations and challenges, but it turns out these tools are even less efficient when it comes to managing a virtual infrastructure. To a large degree, storage admins are blind to the availability, performance and optimization of the virtual infrastructure because the tools are lacking. For example, if multiple VMs share the same LUN and a storage admin discovers an IO problem, they need some tool that will allow them to determine which VM is causing IO contention.

While IT vendors tout their technologies, they need to focus more on teaching people how to effectively and seamlessly use their stuff with VMs. Vendors should create interactive communication with their users. It seems to me that the entire world--except for high-tech vendors--is into social networking and Web 2.0.

The journey toward a virtual data center is essential and (dare I say) may even be requisite, but it's fraught with obstacles and challenges. My intent isn't to scare anyone away from implementing VMs, but rather to encourage users to make well-reasoned decisions. Virtualization offers a ton of tactical value to the business and the data center. However, an impulsive, frenzied move toward virtualization can lead to confusion, misinformation and big mistakes. We must leverage knowledge to mitigate risk.

Article 13 of 13

Dig Deeper on Storage virtualization

Start the conversation

Send me notifications when other members comment.

Please create a username to comment.

Get More Storage

Access to all of our back issues View All