Windows Home Server “Code Name Vail” (WHS Version 2) has been in public beta for awhile, in fact it was refreshed over a month ago. I finally got around to installing it last weekend and I really, really like. Of course, I haven’t really dug into it so there may be some ugliness under the covers.
It’s beta software so bugs are to be expected. I ran into a couple minor or annoying things that already had a bug report opened for them. But for the most part I like what I see and I’m even more eager for the final release.
Some random comments and observations.
Data Protection – The amount of disk space allocated to “Data Protection” seems high. I did find information that there’s approximately 12% of overhead for data protection.
To protect against silent storage errors (bit flips, misdirected writes, torn writes), additional information is appended to each 512-byte sector stored on drive. In particular, each sector is protected by a CRC checksum, which enables Drive Extender to detect data read errors, perform realtime error correction and self-healing (up to 2 bit errors per sector if duplication is disabled, and any number of bit errors if duplication is enabled) and report the errors back to the user and application. The overhead for this additional data is roughly 12% of drive space.
To turn that into real numbers, I turned off all duplication and shadow copies and it still said 891.2GB was allocated to data protection which was 11.9% of the space in my storage pool. This was for a 7452.1GB storage pool with 341.8GB of data and 6GB worth of computer backups. When I turned on duplication for the data and backups the amount used by data protection skyrocketed to 1273.1GB, an increase of 382MB for 365MB worth of data.
Exclude the system drive from the storage pool – I like that the system drive can be explicitly excluded from the storage pool. When Vail installed it created a 60GB system partition for the OS, and the rest was a data partition that could be used by the storage pool. This excludes the data partition from being used for data storage. In theory, this should improve performance by not having the system related i/o mixed with data i/o. It does mean that the data partition is “wasted” space. Since I’ll be going this route (excluding it from the storage pool) I can go with a small system drive.
Hardware Limits – Vail does have some hardware limits. Despite being based on Windows Server 2008 R2 Standard Microsoft made a “business decisions” to limit CPUs and RAM. It will only recognize one physical CPU and up to 8GB of RAM. That one CPU can have up to 4 cores although my test server has only two cores.
While this has brought some complaints on the forums it tends to make sense from Microsoft’s pov considering the market they see for this home server. The new business server (Aurora) will have greater support (and probably cost more).
While I would like the flexibility of multiple CPUs this isn’t something that would impact my current plans.
Improved Storage Pool & Backups – While I haven’t tested each of these bullet points, I’m excited about the benefits they promise.
- Folder duplication happens in real time, not in regular hourly passes. This should provide better data protection. I also hope it improves performance as I sometimes see a performance hit in v1 when demigrator runs.
- File Level Encryption is supported by drive extender. This could be a benefit to security although I;’m not sure I’ll use it.
- Files in use are properly duplicated so the file conflicts they generate are gone. If I leave iTunes open for a long time (as I’m likely to do) I often get file conflict errors because the library file couldn’t be duplicated. Not a huge deal as shutting iTunes long enough to allow the duplication resolves the error.
- Storage operations occur in the background. The server doesn’t go offline during drive removals (this is the one item in this list I have successfully tested)
- Drives can be names in the console, making identification easy.
- Storage operations are a low priority i/o to avoid impacting media streaming.
- Data drives in storage pools can be migrated between servers. This provides some interesting possibilities for moving to a new server. But it also scares me. One thing that’s lost is the ability to read a data drive on another Windows PC. In version 1 the drives were formatted NTFS and could thus be read on another computer. In v2 the drives can only be read by Vail.
- Computer backups are prioritized if there are multiple computers queued to backup. The computer with the least current backup gets priority. Frequently disconnected computers, such as laptops, will backup outside the scheduled backup window if they can reach the server.
I’ve still got a lot of testing to do. I’m impressed with Vail so far but I’m resisting the urge to move my stuff over to Vail for my day-to-day usage. It’s still beta and probably has bugs. Plus I want to do some more testing to come up with an optimum configuration. I want to break free of iTunes for my Video library (just so damn easy to use) so I want to dive deeper into the video streaming options that Vail will open up.
Now that I’ve got the test box up and have had some hands on I’m more comfortable ordering hardware that will be my new Vail server, the first system build on my system roadmap.
The Vail beta and information is available from Microsoft Connect (Connect registration required).