Home > Blogs > Virtual Reality


Debunking Myths About vSphere Data Protection

The release of VMware’s vSphere Data Protection 5.5 (VDP) seems to have caused a stir in the virtual backup industry. It appears we have hit a soft spot with some of the other vendors offering backup solutions for vSphere and have seen some confusing messaging coming from our partners/competitors in this market. While we’re certainly proud of the technology partner ecosystem built around VMware solutions I would like to take this opportunity to set the record straight on vSphere Data Protection.

The Myths

  1. Other vendors claim they never need agents to perform backups, where VDP Advanced does.
  2. If you cannot directly view your backup files you may lose your ability to recover from a disaster
  3. Replicating your backups off-site requires expensive “WAN acceleration” technology
  4. VDP lacks any kind of fast VM recovery technology
  5. VDP cannot automatically verify the recoverability of a VM

We’ll dive in to each of these a little bit to get to the truth about vSphere Data Protection.

Myth 1: Agents, or the lack thereof

Some vendors claim they require no agents to do vSphere backups, even for application aware backups of Exchange, MS SQL, and SharePoint, whereas VDP Advanced does require agents for these applications.

The fact of the matter is, the vast majority of VMs do not require agents because of the way our vSphere data protection APIs work. This is the case for VDP and every other vSphere certified backup solution. But, a proper application consistent backup of Exchange, MS SQL, SharePoint and other application does require an agent, even for vendors like Veeam. Need proof? Here’s a quote from page 235 of the Veeam Backup & Replication Version 7.0 User Guide:

To coordinate proper indexing and VSS activities, Veeam Backup & Replication injects a runtime process inside the VM...In the Guest OS credentials section, specify an account with local administrative privileges for injecting the process.

Call me crazy, but a runtime process injected on a VM via admin credentials to do indexing and other activities on behalf of another server is the very definition of an agent. The biggest difference between VDP and Veeam’s agent approach is that VDP’s agents are a one-time install via wizard, whereas Veeam’s agents are installed and uninstalled each and every time a backup job runs.

And don’t forget: our VDP Advanced agents also run on physical servers so you can backup your entire Exchange, SQL, or SharePoint environment with VDP Advanced.

Myth 2: If you can’t directly access your backup files you may suffer dire consequences

First things first, it really doesn’t matter which backup system you choose – your backup files are useless without the backup servers. Further, if you’ve lost your backup infrastructure I’d say the odds are good you’ve lost other critical parts of your infrastructure as well. In cases like this, perhaps backups aren’t the best option for getting up and running. You might want a disaster recovery solution like our Site Recovery Manager or vCloud Hybrid Service – Disaster Recovery for this situation.

But what about smaller, localized issues? What if your backup server gets wiped out? First and foremost I’d recommend you use a product that includes backup replication so you always have 2nd and 3rd copies of your backups, hopefully on-site and off-site. With VDP Advanced your backups could be replicated directly to another VDP Advanced virtual appliance so you could immediately restore from the 2nd appliance – no additional configuration or setup needed. (Even if vCenter is down!)

So what happens if you have your backup files but your backup server is gone? Nothing! At least not until you re-install the backup server and database and maybe some proxies and repositories so that you can actually use those files, stealing precious minutes or hours from your recovery time objective.

Even if you’re using our basic version of VDP, which is included with most versions of vSphere and which does not have built-in replication, keep in mind that everything you need to protect your backups – the backup files, database, everything! – is contained within a single VM. Simply copy the VM to secondary storage periodically to avoid a single point of failure.

Myth 3: You’re going to need to buy a WAN accelerator to handle replication

VDP Advanced includes highly efficient, secure backup data replication across any link at no additional cost. How do we do it and why don’t you see some special “WAN accelerator” configuration inside VDP Advanced? VDP Advanced is based on EMC Avamar and uses the same enterprise-class deduplication algorithm and replication engine as Avamar. What this means to you is VDP does all the required deduplication as soon as the backups are created, across all backups stored on the appliance. No additional steps are needed to further optimize the data for WAN transfers. Plus you get the added benefit of using less storage for the primary backups so you save money on your overall backup solution!

Myth 4: VDP lacks any kind of fast VM recovery mechanism

“Instant Recovery” is the hot marketing item in the backup world (it’s kind of a boring world). Strategies for restoring data quickly is a topic I’d like to explore further in a more detailed article so we can look at how we’d approach some common scenarios with VDP. For now I want to say this about “instant” recovery: the feature looks good in the brochure, but instant recovery techniques from nearly every vendor end up with VMs that are pinned to a single host, running from your backup storage, with IO shuttled through some sort of proxy VM. Add it all up and you’re left with a significant performance and usability hit to the recovered VMs. If you later decide to move that VM from backup storage to production, it often requires multiple steps to move and rehydrate the VMDKs and then rebuild them from the delta disks that were written while the “instant” VM ran.

In contrast, VDP Advanced can utilize Changed Block Tracking to restore a VM directly on full production storage. This means only the blocks that have changed since the selected restore point will be restored. As a result, restore times can be dramatically reduced – up to 6X versus traditional restore methods according to the VDP Advanced study performed by ESG Labs.

Myth 5: You cannot verify a backup with VDP

This myth is just plain wrong. VDP Advanced does include automated backup verification. And we’re not just talking about verifying a file checksum. A VDP backup verification job can be created to automatically restore and verify the full functionality of a VM on a scheduled basis, e.g., once per week. Results of the backup verification jobs are reported in the VDP Advanced user interface and email reports so that administrators have the utmost confidence that important VMs can definitely be restored when needed.

Where VDP Advanced Shines

We’ve designed VDP and VDP Advanced to offer a great value to our customers, who often struggle to setup a good backup system and cannot afford the high price of some of the enterprise backup solutions. We think VDP excels in many areas but especially with features like:

  • Reduced storage consumption to save you money (up to 75% less storage)
  • Fast and simple recovery for VMs, applications, and files (up to 6x faster)
  • Application protection for Exchange, MS SQL, and SharePoint (virtual and physical)
  • Integration directly with vCenter
  • Overall simplicity (go from zero to your first backup in 30 minutes)
  • Leverage the best technology available (powered by EMC Avamar)

As I said at the start, we’re very proud of the ecosystem of partners we’ve built around vSphere, even those we compete with at times. While we at VMware focus on building products that are “better together” we realize that no single product will fit every customers’ needs and at the end of the day it’s you – the customer – who has to navigate the maze of features and jargon and figure out the solution that’s best for you. I hope this article makes that task a little bit easier.

5 thoughts on “Debunking Myths About vSphere Data Protection

    1. Jim Armstrong Post author

      Thanks! Should be easy to setup but if you have any questions let me know.

  1. Frank

    Some nice FUD here against Veeam.

    1. Agents are old and outdated. They can crash and fail. Veeam will always succeed. They also focus on virtual infrastructures. No use pointing out the physical servers.

    2. Maybe you should be better informed. In the case of Veeam you don’t need to reinstall a server or proxy. They have an easy fast extract & recovery tool.

    3. If this is something VMware uses as a point to win that’s sad. None of the competitors use WAN Acceleration for replication. Veeam uses this to copy backups to multiple locations but that is something else then replication.

    4. Instant VM recovery by Veeam is outstanding and as a consultant I have ran this at many customers without the ‘problems’ you have listed. I do understand that you need some more FUD against Veeam.

    5. So you need to do a full restore to test everything? Seems like a lot of time lost. And from all my tests with VDP it’s true. Veeam has surebackup which isn’t optimal either but automatic testing by Veeam works faster in 90% of the times.

    Again nice FUD against Veeam and nice proof that VMware (and EMC?) fear Veeam. Who will be next to FUD on?

  2. Pingback: Newsletter: May 4, 2014 | Notes from MWhite

  3. Daphnis

    A big +1 to Frank’s comments. Further elaboration to dispel the rampant FUD this post incites:

    1.) No agent is persistent on the VM, and, unlike VDP’s and other solutions, doesn’t need to be maintained (patched, updated, babysat) as an updated copy is always used when it is processed. This means no reboots are necessary to use a new version of the agent.

    2.) Point already said. Fast extraction tool works very well. Furthermore, if you’re using Veeam’s default configuration backup and have access to the backup media itself, spin up a new VM, slap Veeam on it, import the config, and show it your data and you’re back to the races.

    3.) WAN acceleration is NOT required for any replication activities. It is an extra feature designed for off-site replication of the backup data itself. In addition, it also uses advanced caching and deduplication to reduce the amount of data sent over the wire.

    4.) Most of this is just pure FUD. Instant Recovery has worked every time I’ve called upon it and requires but a single svMotion to get it back to production storage.

    5.) The difference between VDP and Veeam’s verification is that with a Sure Backup, you can verify that not only the VM boots but your apps come up as expected by using predefined (or even custom) verification actions. After all, what good is a recovered SQL or Exchange server if the application doesn’t start??

Comments are closed.