As many people have seen, the Windows Home Server team stated that Microsoft is abandoning Drive Extender V2 in Vail. Many are crying foul and have even started to petition Microsoft to bring the new Drive Extender back which currently has approximately 1,500 signers. While most other sites are crying foul, as Drive Extender in Vail added some RAID 10 features with check summing, I have been of the position that Windows Home Server V1 and Vail are better off without using Drive Extender V2.Before going any further, let me say this, there are a lot of overreactions stating there is no compelling reason for Vail now. In some cases, where people want the lightest NAS possible with great Windows backups, this may be true. With that being said, 64-bit support, support for GPT disks, and from a stability standpoint, the community needs Vail. With Windows 7 and Windows Server 2008 (R2) adoption rates being what they are, and the cost of hardware continually plummeting, Server 2003 and Windows XP are slowly becoming less relevant. Personally, I use Windows Server 2008 R2 daily and cannot think of a compelling reason to put Windows Server 2003 on machines.
Next, despite what may amount to a few thousand petitions, Microsoft will likely not bow to pressure and bring DE V2 back, and they should not (at least due to the petitions). I would be willing to wager that a very high proportion of WHS licenses are sold through OEMs like HP, Acer, and others where the software is pre-installed. Frankly, people that buy new home server units from these vendors are going to buy whatever is offered, and Vail will be a compelling enough upgrade from WHS to make the vendors simply use Microsoft’s latest and greatest. My guess is that Microsoft sells under 50,000 licenses to end customers doing home builds a year, so we are talking <$50m in revenue at risk (resellers take a cut) for a company that brings in $60-70B a year in revenue. I know at every large technology firm I have consulted for, a $50m product line is not a top priority.
Where Microsoft makes more money is selling Windows Server Products to small businesses. Let’s face it, RAID 10 or RAID 6 with a hot spare and BBWC performs better than WHS ever did. Even DE V2 could not match the performance of hardware RAID. As someone that has run 20+ drive (Hyper-V/ bare metal) WHS v1 installations with hardware RAID and directly attached drives using DE, WHS (and Server 2008 R2 for that matter) run much better when they have little idea that a drive failed. For example, removing a drive properly from WHS requires that all shares are taken offline for what can be a long rebuild process. I had a 2TB drive (Western Digital Green for those wondering) die this morning on a machine that was using the device directly not through RAID 10 or RAID 6. It was 96% full at the time and it is looking like a 6-7 hour removal time. New 3TB drives are 50% larger than previous generation 2TB drives but only 20-30% faster, so removal times for full drives are only going up.
I digress, where is the opportunity in this? With Microsoft removing DE, Vail is now compatible with the old NTFS stack and all of the software built over the years for it. NTFS is clearly not the best file system around, and ZFS is not either, but at least NTFS is a known quantity because it has been used all over the place. Frankly, hardware RAID works much better than Windows Home Server’s Drive Extender technology ever did.
What I think we should all be hoping for is not DE v2, instead it is Microsoft building a version of “MS-ZFS” that is production ready across the board from desktop to server versions of Windows. Right now, EMC, NetApp, HP, Hitachi and a myriad of smaller vendors are all hot on solid state caching. With next-gen eMLC and new consumer 25nm SSDs hitting lower price points, Microsoft is probably better served looking at something that can address multiple storage tiers, SSDs, cheap SATA storage, Windows Home Server/ NAS storage, and cloud-based storage. Being able to integrate those four tiers while focusing on data integrity would do wonders for Microsoft. The big opportunity here is moving everything to a better file system, and having one Windows file system, not one for Windows 7 and Server 2008 and another for NAS appliances and very small business storage servers.
Aside from all of this, one reason I am excited is that this announcement opens up a lot of possibilities for something better. The question I have been asking myself over the past few months with test Vail setups is, could DE v2 + NTFS be the future of Windows file systems. My hope has been no, and it appears, at least for the time being, that it is not. I understood file corruption I saw because Microsoft clearly labeled this as beta software. Looking at WHS v1’s release, file corruption was not an infrequent “feature” of the software, but it has gotten better with time. Let us hope that Microsoft has taken the experience of WHS V1’s launch and has made this decision with the express intention of building something better, across the board, down the road. One of the first steps one sees in companies transitioning to a new technology platform is curtailing investment in new features for old platforms, hopefully making this transition away from NTFS + DE V2 a precursor to a new, improved, and universal MS-ZFS file system.
I am sure a lot of people were disheartened by Microsoft’s announcement so feel free to vent in comments if need be.