View Full Version : Questions regarding ZCB system compatibility and compression

October 26th, 2009, 04:56 PM

I'm looking to using ZCB but had a few questions that the tutorial videos did not cover.

- We're running our file servers as Server Core 2008 x64 instances and given the stripped down nature of the installation, I was wondering if Java x64 and ZCB would run correctly on it? I assume if it works I would either need to launch the management GUI via command line or somehow hook into a remote instance of the agent.

- We are also running our file server as a Hyper-V guest and could alternatively run ZCB at the host level which does have a full OS install. If we do it this way, how good is the compression related to relatively empty .VHD files? We uses static disk files so although the VHD files are 400GB in side, we are only using a small percentage of that at a given time. Is ZCB capable of handling large file compression and if so, how efficiently would it work in terms of CPU utilization and file compression before transmitting to S3? I would not want to have to pay for a huge amount of S3 storage for a Hyper-V guest file server that is lightly utilized but has been pre-sized for growth.

Thanks for any insight into these questions.

- zifnab

October 26th, 2009, 05:21 PM
Are you planning to use ZCB for disaster recovery? Are you interested in file level recovery?

We have not tested ZCB on Server Core installation. Java JRE 1.6 is not listed as supported on Server Core configuration (http://java.sun.com/javase/6/webnotes/install/system-configurations.html).

ZCB does not have command line tools for users.

So, the only option is to install ZCB on the host OS. ZCB uses deflate algorithm for compression (http://en.wikipedia.org/wiki/DEFLATE). Are the
Hyper-V files sparse files or are they zero-ed out?


October 26th, 2009, 06:02 PM
Initially it would be for file level recovery and later for bare-metal restore via VHD backup once the client could justify backing up VHDs. They're not using sparse files for the VHD files for performance reasons and I don't believe the default VHD format spec zeros out or defragments the disk files regularly.

I'm wondering if anyone has actually tried installing java on a server core implementation since I didn't want to attempt it on the production box. I would hope this would be a supported platform in the future given the prevalence of virtualization in the enterprise and the many benefits (less resource utilization, reduction in updates required and significantly reduce attack surface) for using a stripped down guest.

If it can't be done from the guest itself, is there any way to access the VSS snapshot remotely somehow to back that up?

October 27th, 2009, 12:04 PM
Yes. We are considering Server core for one of the supported platforms in the future. Java support on Windows 2008 Server core has problems. for example: http://forums.sun.com/thread.jspa?forumID=32&threadID=5379613

VSS snapshots cannot be performed remotely. You can perfom VSS snapshot from host OS that will contact guest OS VSS writers to sync data to disk. For file level recovery, the only option is to run backup program on the guest OS.