Improved File System Infrastructure

   

---

The Virtual Disk and Volume Shadow Copy services are key elements of the improved file system infrastructure in Windows Server 2003.

Virtual Disk Service

Whereas the Volume Shadow Copy service, described in the next section, provides an important building block for managing data on disks, the Virtual Disk service (VDS) provides an important new set of APIs for managing the disks themselves .

In Windows 2000, each storage area network (SAN) hardware vendor provided its own proprietary set of APIs for managing its hardware. This makes it challenging to develop uniform SAN-management software. Windows Server 2003 addresses this issue with VDS. VDS implements a single uniform interface for managing disks. Each hardware vendor writes a VDS provider that translates the general-purpose VDS APIs into specific instructions for its hardware. With this abstraction layer provided by VDS, Windows Server 2003 gives customers a more robust set of solutions, including greater flexibility for making long- term investment decisions regarding SANs and other storage options.

Management applications no longer need to take into account the specific hardware being targeted . Instead, by developing applications that target VDS, the next generation of Windows disk management applications will be able to manage any hardware that has a VDS provider.

There is an immediate benefit from the VDS architecture: Microsoft has implemented VDS providers for basic and dynamic disks. This brings functionality to basic disks, such as online growth, that formerly applied only to dynamic disks. See Figure 11-1 for an illustration of the Virtual Disk service.

Figure 11-1. Virtual Disk service.

Volume Shadow Copy Service

The Volume Shadow Copy service is a general infrastructure for creating point-in-time copies of data on a volume. The goal of Volume Shadow Copy is to provide an efficient, robust, and useful mechanism for the next generation of data management applications.

Shadow Copy Restore, described later in this chapter, is the most visible application of Volume Shadow Copy in Windows Server 2003. Another set of applications that use Volume Shadow Copy is backup applications. Using Windows 2000 Server, you had to either stop activity on your server during the backup process or live with the side effects of an online backup: inconsistent data and open files that could not be backed up. With Windows Server 2003, you get the best of both worlds : you are able to do an online backup that results in consistent data, and you don't have to worry about open files.

Volume Shadow Copy facilitates online backups that result in consistent data by formalizing the relationship among three important entities in the data management process:

• Requestors.

  These are applications, such as backup applications, that perform storage management tasks .

• Writers.

  These are applications that generate data.

• Providers.

  This is hardware or software that is able to create a point-in-time copy of a disk.

Figure 11-2 shows how requestors, writers, and providers interact.

Note

The built-in Windows shadow copy provider uses a copy-on-write algorithm to provide shadow copies in the most space-efficient way possible. For example, if you have 3 GB of data on a volume, Windows Server 2003 does not copy all 3 GB when you create a shadow copy. Instead, Windows Server 2003 simply makes copies of the original data on the volume as applications make changes.

Figure 11-2. Volume Shadow Copy interacts with requesters, writers, and providers.

Distributed File System

Windows Server 2003 contains improvements to the Distributed File System (DFS). DFS is a powerful feature for managing federated file servers. It provides the following capabilities:

• Service layer.

  DFS provides a service layer by separating the names that clients use to access files on the network from the names of the servers that actually host the files.

• Scalability.

  DFS improves scalability by letting you balance the client workload among multiple servers.

• Reliability.

  DFS improves reliability by transparently redirecting client requests to a different server in the event of a failure. DFS uses the File Replication service (FRS) to automatically synchronize replicas.

• DFS roots and DFS junctions.

  The fundamental concepts of DFS are DFS roots and DFS junctions:

  • A DFS root is a server, or a set of servers, that clients go to first when trying to access a file.

  • A DFS junction is a reference from the DFS root to the server, or set of servers (replicas), that can handle the file request.

Windows Server 2003 includes an important reliability enhancement for DFS ”the capability of a single server to host multiple DFS roots. In Windows 2000, you could not have multiple DFS roots on a single server. As a consequence, you could have at most one DFS root in a server cluster. If you had more than one DFS root in the cluster, the system would throw an exception if cluster failover required one of the cluster nodes to host both DFS roots at the same time. Because this restriction has been removed in Windows Server 2003, you can use clustering to efficiently improve the reliability of your DFS roots.

Windows Server 2003 improves the manageability of DFS by allowing delegation of administration. In Windows 2000, an administrator for any part of the DFS namespace needs to be an administrator for the entire DFS namespace. This situation presents problems for large corporations that want to build a companywide DFS namespace because Windows 2000 offers no way to assign permissions to administrators in a subsidiary for managing only the DFS namespace affecting that subsidiary.

In Windows Server 2003Server 2003, you can give a principal the rights to administer a specific portion of the DFS namespace.

Windows Server 2003 also improves the behavior of DFS when you set up replicas that span multiple sites. In Windows 2000, DFS would give priority to replicas within the same site as the client, and if no replicas were within the same site, DFS would pick any available replica at any other site for the client's request. This is not the most efficient algorithm.

For example, if your company has sites in Redmond, Washington; Silicon Valley; and Tasmania, and the replica at the Silicon Valley site goes down, DFS in Windows 2000 will send your requests, with equal probability, to either Redmond or Tasmania. Windows Server 2003, on the other hand, uses the site-costing information in Active Directory to choose from among off-site replicas to satisfy a client request. In this example, DFS would see from Active Directory that it is more expensive to communicate from Silicon Valley to Tasmania than from Silicon Valley to Redmond, and it would correspondingly redirect the Silicon Valley client to the Redmond replica.

Other File Serving Improvements

Windows Server 2003 contains many other file serving improvements:

• Improved Common Internet File System (CIFS) throughput ( especially on multiprocessor servers)

• Increased scalability for DFS (including a reduced memory footprint and faster startup/configuration)

• Better diagnostic tools for FRS

• Better disk space and network bandwidth management for FRS

---

Top