The Ultimate Windows Server 1002003 System Administrators Guide [Electronic resources] نسخه متنی

NETWORKING BASICS

Communication among network devices like computers assumes the existence of mutually understood protocols that comprise a set of rules and structural components. Computers must use a common protocol in order to communicate. Underlying communication in Windows Server 2003 enterprise is the default Transmission Control Protocol and Internet Protocol, otherwise known as TCP/IP.

Network Architecture

The Open Systems Interconnect (OSI) model defines network communication in a sequential and hierarchical fashion. As shown in Figure 12.1, it consists of seven layers, a brief explanation of which should provide a better understanding of its conceptual underpinnings. Protocols like TCP/IP embrace only a portion of the total conceptual model.

Physical layer.
The physical layer loosely refers to just about any characteristic of the hardware, such as signal voltages and cable and connector specifications. It transmits and receives data in the raw bit form of 1s and 0s. Standards such as Ethernet 802.3, RS232C, and X.25 dictate the requirements of this layer.

Data link layer.
The data link layer changes the raw bit stream presented by the physical layer into data frames or blocks. TCP/IP typically assumes the use of other standards to specify the data link layer's characteristics.

Network layer.
The network layer is the first level in which TCP/IP directly relates to the OSI model. It is where the Internet Protocol manages communication between the application layer and the lower-level physical and data link layers. IP provides the Internet addressing scheme that defines a common structure or format for datagrams. These are packets of data that include information such as destination, type, source, and size. The Internet protocol defines how the datagrams are interpreted, as well as how data is routed between Internet networks and addresses, especially on segmented or subdivided networks.

Transport layer.
The transport layer communicates with the application layer. When errors are detected, it may request the retransmission of bad or lost packets. IP traffic typically uses either the Transmission Control Protocol (TCP) or the User Datagram Protocol (UDP). Each transport connection is identified by a port number.

Session layer.
The session layer refers to the connectivity and management of network applications. TCP/IP does not directly map this OSI layer.

Presentation layer.
The presentation layer establishes the data format prior to passing it along to the network application's interface. TCP/IP networks perform this task at the application layer.

Application layer.
The application layer processes data received or sent through the network.

A MODIFIED NETWORK MODEL

There is some debate as to how well TCP/IP maps to the OSI conceptual model. However, for the sake of a simple overview, it may be said that it does conceptually correspond (although not identically) to the OSI model with a four-layer hybrid model having an application layer, a transport layer, a network layer, and a physical/data link layer (Figure 12.2).

IP Addressing

The IP address identifies and addresses the network and devices or nodes such as servers, desktop computers, and routing devices. It consists of a 32-bit value that is separated into four octets. When data segments pass from the transport layer to the network layer, the IP appends its own header information and the new grouping becomes an IP datagram (Figure 12.3). The most important fields in this datagram are the destination and the source addresses.

DISSECTING IP ADDRESSING

There are three classes of IP address. An address class is determined by the following rules:

Class A.
In a Class A address, the first number is in the range of 0 through127, with the 0 value reserved for the default route and the 127 value reserved for identifying the local host. This means that only 126 (1126) networks can be defined as Class A; the other bits in the address are used to assign specific node IP addresses. Although the number of Class A networks is limited, and they are difficult to obtain, the number of nodes they can support is extensive. (x represents a fictitious decimal digit in the following tables.)

Class B.
A Class B address uses the first value range from 128 through 191 in the first byte. It also uses the second octal byte set for defining the network. The result is that there can be many more networks defined by Class B addresses. The remaining octal sets are used to assign specific node addresses.

Class C.
Class C addresses use the value range of 192 through 223 in the first byte, and the next two octal sets are used to define the network as well. With three octal sets defining networks, many more networks can be established and identified. However, since only one octal set is available to define the IP address of a specific node, the network is smaller, with a potential to provide up to 254 hosts or node addresses (the 0 value identifies the network and the 255 value is reserved for broadcasting).

SUBNET MASKS

A subnet mask permits the division of a single IP address into multiple subnetworks. A subnet mask defines how the address space is divided among networks and hosts. It is a 32-bit binary number that designates which bits of the IP address are intended for the network and which determine the host. The network portion is identified by 1s, and the host by 0s. Thus, for the previously discussed Class A, B, and C addresses, the subnet masks in the following table would be used.

A Class C address could designate the first 24 bits as network space and the last 8 bits as host space on the network. Addressing at the data link layer is handled with Media Access Control (MAC) addresses. Discussing address frame delivery at the data link layer is beyond the scope of this book. However, once a packet is forwarded to the segment with the destination node, the data link media is responsible for frame delivery. In order for this to be accomplished, a node must map the destination IP address to the destination MAC address.

THE ADDRESS RESOLUTION PROTOCOL

A computer system or printer may establish connectivity via a serial port, phone line, interface card, or other device. However, most connectivity strategies for local area networks involve an interface card to one of the more popular media types, such as Ethernet, Token Ring, or FDDI, which implement elements from both the physical and data link layers of the OSI model. The interface card uses these hardware definitions and data link protocols to encapsulate network-level packets into media frames that are then delivered to a destinationanother interface card.

Media Access Control addresses identify and address the interface cards. Most consist of six hexadecimal numbers separated by colons (e.g., 00:53:45:00:00:00). The encapsulated network-level IP packet has its own addressing scheme, which must then be mapped onto the MAC address in order to communicate with systems via their interface cards. The Address Resolution Protocol (ARP) maps between MAC and IP addresses on the local network segment.

NOTEWindows Server 2003 has improved device driver support for network cards and equipment. This feature adds network device drivers that are common in home networking and removes earlier device drivers. The feature is accessible from Start Control Panel Performance and Maintenance System Hardware Device Manager. It also includes improving the quality of networking drivers. Driver categories include:

Local Area Network (LAN) Drivers including 10/100 Network Interface Cards (NICs), IEEE 802.11, and Home Phoneline Networking Alliance (HomePNA).

Broadband including cable modems, Asynchronous Digital Subscriber Line (ADSL), and Integrated Services Digital Network (ISDN).

Modems including driver-based and 56 kbps V.90 modems.

Infrared Modems.
IrCOMM Modem is a driver that exposes the IrDA stack to cellular telephones. This feature permits support for older cellular telephones that have IrCOMM virtual serial ports.

For example, if Node A desires to communicate with Node B, it must first map Node B's IP address with its physical MAC address. Once the destination MAC address is resolved, the data link layer is responsible for delivering the frame. Node A broadcasts an ARP request, which contains the destination IP address, to every node on the physical segment (Figure 12.4). Each node reads the request and checks to see if the destination IP address matches its own. Only Node B, owner of the destination IP address, returns an ARP reply to the source MAC address. Node A retrieves the source's MAC address from the ARP reply data link frame and stores the IP-MAC address mapping in its ARP cache. This entry will remain valid in the local volatile ARP cache for a predetermined period of time, usually about 10 minutes. If the address mapping is validated with continued communication, the entry will be refreshed. Once Node A has Node B's IP address and corresponding MAC address, communication may be initiated with direct unicast data link frames between the two.

IP Routing

IP packets are forwarded between networks using specialized network equipment known as routers or by multihomed hosts (i.e., systems with multiple network adapters) with routing software. Upon receiving a packet, the routing device determines which interface produces the quickest route to the destination IP address. The packet is then repackaged and sent out on the correct interface, which the router determines based on its routing table. This table is constructed with manually configured routes, or static routes, and routes obtained via communication with other routers.

NOTEA new feature in Windows Server 2003 is a network bridge that offers the ability to interconnect network segments. A multisegment network system can have multiple adapters such as a wireless adapter, a telephone line adapter, and an Ethernet adapter. By bridging these adapters, computers and devices on each network segment use the bridge to communicate with each other. The New Connection Wizard is used to set up and administer a multisegment home network as follows: Start Control Panel Network and Internet Connections Network Connections Create a new connection.

NOTEMedia Bridge support has also been added in Windows Server 2003. The Network Media Bridge is a Layer 2 bridge that simplifies setup and administration. Typically, a multiple-segment IP network includes assigning each network segment a subnetwork number plus configuring hosts on each subnetwork and packet forwarding between the subnetworks. It provides a single IP subnetwork with no configuration necessary on the part of the user. This feature is available from: Start Control Panel Network and Internet Connections Network Connections Create a new connection. Only one bridge is allowed. The Network Media Bridge also supports IEEE 1394 and IEEE 802.11 devices.

Routers communicate these dynamically changing routes through the Routing Information Protocol (RIP), the Open Shortest Path First (OSPF) protocol, and others. Static routes are configured through the command line using the route command or through the Routing and Remote Access snap-in.

CLIENT-SIDE ROUTING CONFIGURATION

Three parameters are of particular interest when configuring a host for communication on a network. Double-click an interface from Start Settings Network and Dial-up Connections. Select the Internet Protocol (TCP/IP) and click Properties to see them (Figure 12.5):

IP address
unique IP address assigned to the interface.

Subnet mask
determines network and host portions of the IP address.

Default gateway
IP address of network's router interface.

A client's IP address may be manually configured with a static IP address or assigned automatically via the Dynamic Host Configuration Protocol (DHCP), discussed later in the chapter. The subnet mask identifies the portion of the IP address that pertains to the network and host addresses, and is mapped against an outbound packet's destination IP address. If the packet is not bound for the local network segment, the client sends it to the default gateway. The gateway forwards the packet onto the destination network using routing tables, as discussed earlier.

NOTEThe following command can be used to return TCP/IP settings to default values with Windows Server 2003:

Netsh int ip reset logfile

NOTEAn integral feature of Windows Server 2003 is that the TCP receiving window size is determined by the local NIC. This is important because the window size determines the maximum number of bytes that can be unacknowledged but sent at any time. This can be critical on a slower dial-up network connection where the window size is generally equal to the queue depth on the Remote Access Service server. When the queue is filled with packets from one Transmission Control Protocol connection, a new TCP connection cannot be established until all these packets are sent out. Window size adjustments to make the connection speed are now made to improve the Quality of Service (QoS) Packet Scheduler on a system with Internet Connection Sharing (ICS). This reduces the queue depth at the RAS server. It also enables new connections to work better. ICS must be used with this new feature.