Enterprise VM Guest Backup Solutions

[phowarth]

I'm looking into an Enterprise Backup Solution that will allow me to backup the encapsulated VM Guest restore the encapsulated Vm Guest and allow individual file restores. What are the current solutions and is Vmware or anybody else coming out with a good solution in the near future? The environment size has 250 hosts with about 15-25 guest running on each host. Current solution is using Netbackup with agents inside the guests.

I looked at vRanger Pro but have questions about it. Is it Enterprise Class to support the size environment above? And for those that use it for a large environment how does it offload it's work load? One little vRanger box cannot handle the backup of 5000 guests. After reading the documentation it sounded like it pushes the processing of the backup to the source ESX Host hosting the VM is this correct? I can then make a fiber connection or nfs connection to a Data Deduplication Appliance on other ESX Hosts in the Cluster and tell the job to use those connections on the ESX Hosts to stream the backup to the destination device to store the archive. Is this correct?

Thanks

Pete

[gary1012]

VCB will provide you the framework to perform both image and file-level backups. What is your current backup solution? That probably will impact the answers you get back from the community.

[phowarth]

Current solution is that Netbackup backs up all VM Guests using an agent installed inside of each guest. This gives me the ability to restore individual files, but if I loose an entire lun or a VM doesn't let me restore the entire encapsulated guest back to it's lun. I have the questions above about vRanger Pro and like the fact that it gives me the ability to restore the encapsulated VM Guest if a lun gets blown away and allows me to mount the vmdk file and search and restore individual files if needed. But I'm concerned in how it offloads processing power with regards to processing the backup and where it streams the backup to. I read the entire manual and it seems like it pushes the processing of the backup to the ESX Host sourcing the VM guest being backed up and allows you to choose either a shared nfs mount point or a fiber attached shared lun on another ESX Host in the cluster to send the archive. I'm not sure on this. That' why I have the questions about that solution. But am open to other Enterprise Class Backup Solutions that give me abilities I need mentioned above that can handle a 5000 VM Guest Environment.

[gary1012]

Valid concern... VCB was designed to offload backup processing from the ESX to the VCB Proxy. Vizioncore has published a whitepaper on how their product interacts with VCB. Most of the major backup players also have hooks into VCB to provide seamless image (encapsulated) and file-level backups. Some are more elegant than others, but most will provide the image backup you desire.

[cwenzel]

What version of NBU (NetBackup) are you running?

In version 6.5 they introduced integration with VCB that allows for the LUN/vmdk level backups you want to perform. The VCB framework is still required, but it is managed through by NBU so you still retain the media and device management that NBU provides.

[Texiwill · Accepted solution]

Hello,

Check out http://vmprofessional.com/index.php?content=esx3backups for a comparison on backup tools used with VMware VI3.x.

Best regards,

Edward L. Haletky

VMware Communities User Moderator

====

Author of the book 'VMWare ESX Server in the Enterprise: Planning and Securing Virtualization Servers', Copyright 2008 Pearson Education.

Blue Gears and SearchVMware Pro Blogs: http://www.astroarch.com/wiki/index.php/Blog_Roll

Top Virtualization Security Links: http://www.astroarch.com/wiki/index.php/Top_Virtualization_Security_Links

[PaulCooper]

Considering the size of your environment you probably want to look at this holisitically - ie not just at the software that will work with ESX, but backup data paths/ storage devices, RTOs & RPOs, data retention etc - and also make sure that the company networking and storage strategies are aligned.

There are some hard limits too on what many products can achieve - and although you give a number of VMs you don't mention their sizes, backup windows etc etc all of which need very careful consideration.

I went through an exercise with the guys I was working with about 7 months ago. We had one guy - very up-to-speed on current storage technologies (and NetApp certified), looking at bandwidth/data flow etc across some different options and looking at what was needed there - for about 60TB of data across two datacentres (iirc). We also looked at Netbackup with VCB, and also NetApp OSSV. We were constrained by the storage backend to not be able to use SAN snapshots/SRDF but again that is something that may be a suitable solution for you.

In the end we came up against too many technical constraints (the products just weren't mature enough, and VCB has far too many limits around image-level backups - especially if you fancy adding deduplication to the mix) and providing enough network/storage bandwidth to meet the backup windows would have cost several hundreds of thousands of pounds. Guess what we did? Went for NetBackup clients in the VMs, and a decision to review the marketplace in 2009. It's not the most elegant solution, but it works

[phowarth]

Thanks. I found after reading through the manual again for the Visioncore Product that it uses the Host's as a resource to offload the backup using storage attached to other hosts. You can offload that processing by making use of VCB as proxies which run the jobs faster and get it off your hosts. But the question is how many proxies will you need to be be able to backup all those guests.... I didn't know that Netbackup 6.5 worked with VCB to get the LUN backups. Nice. I'l have to look into that.

[phowarth]

I think that's the way we are going to go. I don't think there's a real robust solution yet for an environment of this size. Right now they backup all the guests with Netbackup and Agents. If a guest tanks it just gets rebuilt and restored. We did have a problem where an entire lun went and we had to rebuild all the guests that were on that lun from scratch and restore. Was not a fun experience with higher ups constantly looking for status etc. That's why I was looking at what other people are doing for environments of that size.