Enhancements in the long-awaited Win2K

Over the years, users have questioned Microsoft's direction in developing Windows NT's file systems and disk management features. I've heard users ask why NT maintains such a large file cache for so few files and why they can't extend partitions using the OS. They also want to know where NT's defragmentation utility is and when NTFS will offer data encryption. Finally, users ask implementation-oriented questions about why you must use NTFS to take advantage of NT's important security features and what the differences are between FAT, Virtual FAT (VFAT), and FAT32.

After much anticipation, Microsoft might finally release Windows 2000 (Win2K--formerly NT 5.0) next year. When the new OS comes out, you can bet it'll be a hot commodity. Everybody wants to get his or her hands on the new Microsoft Management Console (MMC), Active Directory (AD) Services, and Plug and Play (PNP), among other things.

Aside from these new features, you'll want to know what Microsoft has in store for the overall I/O and file system structure. Let's take a look at the new OS to determine whether these changes are technical, or merely semantics.

Changes to Read-Write File Systems
FAT32 is the first significant change to NT's file system since Microsoft redesigned FAT to support partition sizes greater than 32MB. Microsoft first introduced FAT32 in the OSR2 version of Windows 95.

One of FAT32's most important features is that it breaks the 2GB partition barrier and supports drives up to 2TB (2048GB). Although this support is useful to administrators who want to create one large partition for large (i.e., several gigabyte) drives, it is not useful in unattended installations unless the clients are running Win95 OSR2 or Windows 98. In the traditional NT 4.0 installation instances, you must create and format the target partitions before you perform the unattended installations. You typically create the partitions using the DOS command FDISK and use the DOS FORMAT command to format the partition using the FAT file system, which doesn't let you go beyond 2GB when you format the drive from DOS. Supporting FAT32 may be beneficial to NT partitions that you create from scratch during an interactive installation of NT or after you've installed Win95 OSR2 or Win98. Microsoft might adjust installation in the final release to accommodate FAT32 for unattended installations. Alternatively, if you can't get an unattended installation to do your bidding, you can always use disk image cloning software such as Ghost Software's GHOST (http://www.ghostsoft.com) or PowerQuest's Drive Image software (http://www.powerquest.com). Don't confuse FAT32 with VFAT. VFAT is not a file system but an enhanced file system driver for use with NT and Win95. It functions as an overlay to the 16-bit FAT file system commonly found in DOS. If you plan to dual-boot NT 4.0 with any current beta release of Win2K, you can obtain a read-only FAT32 driver for NT 4.0 from Systems Internals (http://www.sysinternals.com).

Another important FAT32 feature that NTFS already implements is that it uses smaller allocation unit sizes and a smaller cluster size than FAT. This configuration results in 15 percent more efficient use of disk space. Slack space occurs when an allocation unit has part of a file or a whole file that doesn't take up the full allocation unit and the remaining portion isn't available for storage. Some OSs, such as NetWare, work around this slack-space problem through block suballocation.

FAT32 provides more file system stability than FAT. Microsoft expanded the boot record on FAT32 drives to include a backup of critical data structures. FAT32 can also maintain multiple copies of its File Allocation Table. In the event that the first table is corrupt or unavailable, FAT32 relocates the root directory and uses a backup copy of the File Allocation Table. FAT32 doesn't restrict the number of entries that can appear in the root directory. However, 16-bit FAT limited the root directory to containing 512 entries of directories and files. This limitation led to Out of disk space errors when a user attempted to copy a file to the root directory, even if the directory had 800MB of disk space available.

Legacy compatibility will be the biggest concern with FAT32. Microsoft said certain APIs might not work with FAT32; therefore, vendors have to modify common programs such as Scandisk, Defrag, and fdisk. Vendors also have modified third-party products such as Partition Magic to support FAT32. In addition, hardware vendors will have to update certain file and filter device drivers (such as those used by virus scanners, and Microsoft will update its file system drivers to support FAT32. In beta testing Win2K, many users reported volume corruption on boot and system volumes formatted using FAT32 and configured to dual-boot Win2K with Win98. At press time, no workaround was available. You should never install a beta on a production system.

NTFS 5.0. Many users thought the onset of FAT32 would bring an end to NTFS. Rumors of the demise of NTFS are untrue, and Win2K confirms this claim. In fact, Win2K adds enhancements, including file system improvements plus inherent features in the file system, to NTFS that the OS formerly wasn't implementing. Microsoft added an updated NTFS 5 driver for NT 4.0 in Service Pack 3 (SP3) in anticipation of a possible dual-boot configuration between NT 4.0 and Win2K.

A welcome improvement will be NTFS 5's per-volume, per-user, per-disk quota management, which the file system supports based on the security ID (SID). You use quotas to limit the disk space you allot to users and groups and place size limitations on particular directories. This feature should be standard in any enterprise network environment.

Other interesting features in NTFS 5 include support of Sparse Files and better transactional tracking. Sparse Files handle disk usage through demand allocation. For example, if an application wants to create or access a 10GB file in size but you commit data only to the first 1MB and the last 1MB, NTFS will allocate physical disk space only to the portions of the file the application is writing to. In other words, the physical file is actually smaller than the 10GB the system allocated to the application. This feature lets the application create large files, but consume disk space only as needed.

NTFS 5's inherent transactional tracking mechanism will support a volumewide transactional tracking log that's better for recording larger amounts of changes over several reboots. Before NTFS 5, records only lasted until the next successful reboot, which limited the extent of NTFS recoverability. The new mechanism will track all changes to files and directories over a longer period, on a per-file basis. Applications can use this feature for I/O analysis, disk recovery, and database transaction recovery. The transactional tracking mechanism is also one of the reasons that all Win2K domain controllers or directory servers must use an NTFS 5.0 partition as the system volume.

One of NTFS 5's most interesting new features is Junction Points (also called Reparse Points). NTFS Junction Points merge a target directory onto an empty NTFS 5 directory in the local computer's file system namespace. The target can be any local or remote (Uniform Naming Convention­UNC) valid NT pathname. NTFS Junction Points are transparent to users, shells, and applications unless the application developer designs the application to be aware of the Junction Point. This means you can transparently reroute applications or users accessing a local NTFS directory to any other directory. For developers, Junction Points are extensible, meaning that developers can incorporate special mechanisms of reparsing NTFS mount points into a file system driver. Mount points in this context are similar to those found in UNIX environments. Unlike the Junction Points in Distributed File System (DFS) 4.1 for NT, NTFS Junction Points are incorporated into the file system, rather than through a service.

Read-Only File System Support
Win2K enhances its support for optical media. The new version will support Digital Versatile Disc (DVD), an optical technology that can store nearly 5GB of data on a single disc. Most DVD drives remain backward compatible with existing CD-ROM technology. DVD also can store information in dual-layer format, and the disk can have two sides. Therefore, a double-sided, dual-layer DVD can hold 17GB.

Win2K will also support the Universal Disk Format (UDF) standard. This support corresponds to the International Organization for Standardization (ISO) 13346 file system that was designed for cross-platform data interchange.

Clustering and DFS. Win2K will support Microsoft Cluster Server (MSCS). This functionality lets you configure two NT servers as a cluster of resources providing a mechanism of server fault tolerance. You can also use MSCS to adjust workloads and load balance I/O and processing loads. Administrators can even perform maintenance without taking the applications and data offline.

Microsoft also plans to incorporate its DFS into the commercial release of Win2K. DFS lets users avoid having to keep track of server names. It organizes file system structures that span multiple computers into a hierarchical uniform directory structure in the same manner that NFS and Andrew File System (AFS) allow UNIX users to do this. This feature brings relief to enterprise administrators who want a software solution for a centralized hierarchical file system namespace. DFS lets administrators transparently link server share resources into a single logical directory structure. It is an excellent tool for creating an environment where you can add storage whenever needed. The size of a disk is no longer a crucial issue, and server managers can load balance their data more practically than previous options permitted.

Users won't have to connect to so many shares all over the network. They can go to one logical directory structure, obtain all their resources, and never know that they are bouncing from server to server rather than traversing a single directory. DFS 4.1 is currently available for free to use with NT 4.0. It is a available from the Microsoft page (http://www.microsoft.com/windows/downloads/contents/Other/NTSDistrFile/). All the features of DFS 4.1 will be implemented in Win2K with the addition of fault-tolerant DFS roots and leafs, support for DNS naming, support for AD Services, and the capability of using NTFS 5 Junction Points.

Changer Media Services. With Win2K, administrators will be able to oversee various removable storage devices, including tape, CD-ROM and optical drives, and robotic libraries, using the NT Media Service (NTMS) Changer interface. NTMS will help NT become an enterprise-level tool by facilitating an inherent means of Hierarchical Storage Management (HSM) and remote control and shared storage of magnetic and optical devices.

Useful features of this new interface include a task scheduler, an interactive file recall notifier, and a uniform gateway to all devices through which you can back up and restore. NTMSs will give users an intelligent disaster recovery component that is currently lacking in NT 4.0's Emergency Repair Disk and Directory process. Through NT 4.0, users have had to rely on third-party solutions to facilitate this support.

Disk Defragmentation
Win2K will include a manual version of Executive Software's Diskeeper defragmentation software. The automated version of Diskeeper will still be available as a third-party utility. Windows 2000 Professional (Win2K Pro--formerly NT Workstation 5.0) and Windows 2000 Server (Win2K Server--formerly NT Server 5.0) will let you defragment all supported read-write file systems.

File System Encryption
Win2K is more secure than the previous version with the Encryption File System (EFS) component that works with NTFS. The encryption technology is public key-based and runs as an NT service. It runs constantly outside the current user's security context so that the encryption is maintained even if other users are logged on, and it remains in place when the user is logged off. Users must have the private key to access an encrypted NTFS file. With the key, the user can work with the file as usual.

In EFS, Win2K treats encryption as an extended attribute like other NTFS security descriptors. EFS works with applications known for using dirty data and a lot of temporary files. If all files are on an NTFS volume, the attributes from the original file are copied to those temporary files, so that they are protected even if the application fails to delete them because of an application failure or a system crash. Per-File Disk Encryption provides another level of security in Win2K. The encryption and decryption is at runtime, so information is adjusted through the interface in the same way Win2K handles file compression. Win2K protects the data even when the user shuts down Win2K, because the file remains encrypted while closed.

Disk Administration
With Win2K, you will be able to perform more online disk management changes, including creating, extending, and mirroring a volume without rebooting the system. You will also be able to add hard disks without rebooting.

Win2K's Remote Disk Administration will let you access any remote computer that runs NT 4.0 or Win2K. Keep in mind that third-party solutions can provide this remote access today. If you use these applications properly, Win2K usually doesn't have to shut down when you make changes using Win2K's Disk Administrator.

New Concepts: Basic and Dynamic Storage
Win2K disk management deals with two basic concepts of storage organization: basic storage and dynamic storage. Basic storage refers to NT 4.0's existing partition-oriented scheme. Basic storage is important for the upgrade process because Win2K automatically sets up partitioned disks as basic storage to provide backward compatibility.

Dynamic storage uses a volume-oriented scheme for disk organization. It isn't backward compatible with disk partitions that you create under NT 4.0. Microsoft designed dynamic storage for fault tolerance drives and special drives such as volume-spanning and generic striping. Through NT 5.0 beta 2, disk management will not default to allowing boot volumes on a dynamic disk. Microsoft recommends that the first disk always remain a basic disk with NT located in the active partition if there are multiple partitions. You can initialize new or empty disks as basic or dynamic after installation.

Not Everything Has Changed
In the early beta of Win2K, Microsoft's mechanism of software-based fault tolerance (RAID Support) is technically the same as the mechanisms in NT 4.0. Although Microsoft has incorporated features such as on-the-fly management, the heavy reliance on good disk I/O and OS level caching might make a server manager consider a hardware-based RAID solution. Nonetheless, strong new features such as NTFS 5, Per-File Encryption, DVD support, and FAT32 make Win2K a significant improvement in the area of secondary storage. If you desire support for dual-booting Win2K with Win98, stronger security, additional media support, and a more integrated DFS, Win2K may be worth the upgrade.