The Future of Windows NT
The future of Windows NT has not yet been written. Although Microsoft has internal design plans and lists of desired features that should be added to Windows NT, only the future will determine exactly what happens. In an attempt to keep interest and to show that the future of Windows NT is bright, Microsoft shares some of these plans with the public. On the other hand, some of the plans are not discussed and shrouded in mystery. In any case, until future products ship, no one, not even Microsoft, can tell what will be included.
Despite some of the speculation that is involved in predicting the course of the future, we can base our projections on what we know today. For instance, we know where some of NT's weaknesses lie, and it is almost certain that Microsoft will address these concerns and strengthen NT's armor. Additionally, we know some of the strategic plans for Windows NT that make the foundation of Microsoft's operating system strategy.
When the name Cairo first surfaced in the media, it was billed as a future version of Windows NT, which would come after NT 3.5x. For most people, this meant that Windows NT 4.0 would be Cairo. Microsoft had even disclosed a number of features that they expected to ship with the Cairo release.
The problem is that some of the Cairo features proved more time-consuming and troublesome than others. In order to provide timely updates to Windows NT, Microsoft changed Cairo from a product-focused concept to a technology-focused concept. This also represented a change in Microsoft's paradigm for shipping and designing operating systems. Although NT 4.0 is not Cairo, it includes a number of the Cairo features, such as the user interface that first shipped with Windows 95. Also, Exchange Server includes some Cairo technologies, specifically the extension of the NT Directory Services, to include additional user information and lay the foundation for full-scale integration with X.500 directory services.
However, some of the Cairo technologies are still under development and will ship with future versions of NT, or as incremental upgrades.
Three of the most significant future Cairo technologies are
- Expanded Directory Services
- Object-Based File System
- Expanded integration with DNS
Expanded Directory Services
The current directory service in Windows NT, called the NT Directory Service (NTDS), provides a two-level hierarchy and is used primarily for administration. However, the future of Microsoft's directory service is an infinitely hierarchical object-based directory service built on Microsoft's Object Linking and Embedding (OLE).
The goal of the next generation directory service is to provide more than just administrative information in the directory. There are many pieces that fit into the puzzle to create this expanded directory service. The two major pieces that are yet to come are
- Object-Based File System: The next generation of NTFS will act as the object repository for the expanded, object-based directory services. This is an integral piece of the directory services strategy.
- ODSI and OLE DS: Open Directory Service Interfaces (ODSI) is one of the fundamental Windows Open Services Architecture (WOSA) technologies. ODSI is related to directory services the same way ODBC is related to databases, or MAPI to messaging services. Object Linking and Embedding (OLE) Directory Services (DS) is one of the core pieces of ODSI that lets developers enable their applications with directory services features. OLE DS is an extension to the core OLE services that supports multiple directory services by abstracting the differences and presenting a common interface to the application developer and the administrator.
Object-Based File System
The Object-Based File System was originally thought to be a completely new file system, called OFS. However, the latest news, and a recent interview with Jim Allchin, who was the originator of the OFS concept at Microsoft, indicates that the new file system will be an extension to the existing NTFS file system.
The major opinion so far seems to be that the decision to embrace OFS as an extension to NTFS was a wise choice. By simply extending the existing file system, it is less likely to cause incompatibilities with existing services and third-party programs. Additionally, it is a testament to the robust design of NTFS that it is capable of supporting such a major upgrade with relative ease.
Some of the improvements to the file system will be extended file attributes, such as author and content information. Additionally, these enhancements will allow content indexing in the file system structure.
The real power of the new version of NTFS will be in its support of the extended directory services feature, which attempts to unify the storage of all data. Currently when you want to look up a file, you must know what drive it's on, and what directory, or directories it is in. With the object file system, you would be able to locate a document based on its author, content, or actual pieces of text within the document itself, easier and quicker than you could today.
It is this rich-text sorting and searching that will give the new file system its real power.
Expanded Integration with DNS
With the tremendous growth of the Internet, which uses DNS as the foundation for name resolution, Microsoft has shifted its focus with regards to DNS. Although Microsoft has recognized that interaction with DNS was a necessary part of support for open systems, the expansion of the Internet has changed Microsoft's stance from one of compliance to one of active participation.
The inclusion of a full DNS to WINS gateway in NT Server 4.0 was simply the first step in the inclusion of DNS into the naming and directory services strategy. Future releases of NT will increase the ability of Microsoft networking clients to use DNS names as a universal naming system.
This system would leverage some of the X.500-compliant features that have already been included in Exchange. Additionally, Microsoft is working with the Internet Engineering Task Force (IETF), the group responsible for setting standards on the Internet, to develop standards for both secure and dynamic DNS.
Clustering is a strategy for Microsoft that plays an important role for the future of Windows NT but is not part of the Cairo technologies.
So what is clustering? A cluster is a group of servers that is addressed and managed as a single, logical entity. Instead of accessing a single server, a client would access the cluster as a whole.
There are third-party clustering solutions for NT, such as Digital's Clusters for Windows NT, in place today. However, these solutions are based on proprietary hardware interfaces. Microsoft's forthcoming clustering system, code named Wolfpack, will provide hardware- and vendor-independent clustering for all systems running Windows NT.
Clustering was first popularized by the OpenVMS VAXcluster model, which was used to distribute symmetric applications across multiple systems, providing increased performance and reliability.
There are two main advantages for clustering:
- Reliability and fault-tolerance
Clustering for Scalability
Although Windows NT's symmetric multiprocessing (SMP) allows you to improve server performance by adding additional processors, clustering is the capability to improve performance by adding more servers. Sometimes this is referred to as bringing scalability out of the box. Following this, adding processors is scaling up and adding servers is scaling out.
It's still too early to tell, but for applications written to take advantage of clustering, performance per dollar should be significantly better than large SMP boxes. For instance, it is significantly cheaper to install two four-processor computers based on the Intel Pentium Pro than a single eight-processor computer based on the same processor. Using clustering the performance for the two machines would be comparable[md]depending on the application. Additionally, the two four-processor computers using clustering would provide 100 percent redundancy in the event of a hardware failure, while the single eight-processor computer would not.
Clustering allows you to leverage commodity hardware to improve performance, rather than adding additional, and often expensive hardware. It also allows you to buy enough capacity to meet your needs today, and then incrementally grow the size and power of your cluster by simply adding additional servers.
Clustering for Reliability and Fault-Tolerance
The compelling feature behind clustering is the level of increased reliability that it can provide. The theory behind the cluster is that if one or more systems within cluster fails, the cluster will continue to work, but with a reduced capacity.
The advantage of clustering here is that it provides complete, multi-path redundancy. Providing clustering as a fundamental technology within Windows NT will allow NT to fully compete in an environment when 100 percent system availability is critical to corporate success.
This chapter looked briefly at the future of Windows NT. Although no crystal ball is able to tell us exactly where NT is going, there are a few features that are in high enough demand and are already in the works from Microsoft. These features, such as expanded directory services through OLE DS and OSDI, the Object File System (OFS), and increased, native support for Domain Name System (DNS) standard for network naming, and clustering for increased performance and availability, are the critical features that are needed for NT to truly meet its promise of being an enterprise network operating system.