felipe-alfaro.org

These are the Top 10 new features in Windows Server 2008: Is this innovation? Maybe. Let me go through them all:

1. The self-healing NTFS file system:

   A self-healing NTFS file system sounds like a cool feature, but any modern journaling filesystem is pretty good at recovering from errors by replaying the journal (usually at mount time).

2. Parallel session creation:

   Parallel session creation sounds like a minor improvement to Terminal Services in Windows. Basically, the fact that a Terminal Server can’t concurrently create new user sessions seems like a big deficiency to me. NoMachine’s NX and the old and good X11 system allow for concurrent user session creation since basically there is no writable, shared state between users/sessions. X11 authorization cookies are created in the user’s home directory and the session itself is abstracted by processes and some IPC resources.

3. Clean service shutdown:

   To be honest here, a malfunctioning daemon could potentially slow down the shutdown of a UNIX system, so improving on this front sounds like a good idea. On the other side, how many times do you have to reboot a UNIX server? Hint: uptime.

4. Kernel Transaction Manager:

   I’m not entirely familiar with Windows kernel internals, but this sounds like a cool feature. I’m not sure what UNIX systems already implement this, if at all.

5. SMB2 network file system:

   It’s glad to know that the old, insecure and not very efficient SMB/CIFS network filesystem is being revamped. However, I am wondering what kind of changes is Microsoft introducing and how will they affect open source implementations like SAMBA. I don’t foresee any patent issues on this, though, since file system development (either local or networked) has stalled considerably during the last years, and even ZFS and Reiser4 only bring concepts but I wouldn’t call them revolutionary.

6. Address Space Load Randomization (ASLR):

   I don’t see anything fascinating here. The Linux kernel has been sporting this with the introduction of Exec Shield.

7. Windows Hardware Error Architecture (WHEA):

   This seems very Windows-centric, so I don’t see any real innovation here. I would call this another privative feature of an already privative operating system.

8. Windows Server Virtualization:

   Nothing to see here. Xen support is extremely well integrated in Linux (read as in Red Hat Enterprise Linux, CentOS or SUSE Novell Enterprise) and NetBSD, and it offers support for paravirtualized and HVM-based virtualization. Xen has proven mature enough for production use, has lots of cool features and it’s open and free and has an exceptional performance. In fact, hypervisor-based virtualization had been in used by IBM’s POWER architecture for quite some time — IIRC starting with VM/370 — so no innovation here.

9. PowerShell:

   Although a nice addition to Windows, UNIX has always had powerful shells like tcsh, zsh, bash or even PERL for years and also new scripting languages like Python or Ruby are starting to be used to develop more complex scripts.

   A powerful shell scripting environment has proven useful for automating tasks and administering systems. PowerShell sounds like the same tune to me.

10. Server Core:

    Finally, it seems Microsoft starts to understand that useless and insecure features, like Windows Media Player, are only meant for desktop systems and not for servers. They cripple the operating system, increase complexity and bring nothing, so removing them completely is a good idea. It’s a pity, however, that it took Microsoft more than 10 years to realize what UNIX already had 40 years ago.

Conclusion

While many of the ideas and features commented here are good, I can’t seem to see much innovation here. I’m wondering how many patents did Microsoft file for such simple, tried-and-true ideas.