What type of network does not require a network operating system in which each computer is considered equal?

The Network Operating System

The NOS is the brain of the server; without it, it's just a computer. It offers applications and utilities that do business faster and better. There are only a few popular choices – Novell, UNIX, Linux, and Windows. The complexity of NOS forces a simple overview of the features and benefits.

Although Linux offers a free download and a plethora of available features to function as a NOS, the most attractive aspect of this Microsoft product is the monolithic support structure and community. All VMS software works on Microsoft OS. Microsoft takes sophisticated tools and applications and makes them usable and affordable, so that any size company can take advantage of a complete suite of server utilities and applications. However, many people have a bias and prefer other systems to Microsoft or simply don't like Microsoft. Years ago, one of my clients chose a Web servicing company that slapped together an assortment of custom and obscure technologies from which they created four unique Web sites. Providing them with a shared Oracle database gave my client a more cost-effective database solution, but negated other features (for security reasons they were told) and locked them into using this particular Web services company. My client was unconcerned about this arrangement until the company found itself tied into multiple approvals and design changes to their Web site at $200 an hour. Arrangements like this can become very expensive. The IT director hated Microsoft and their products with a passion (there's one in every crowd). This prejudice steered the company to technologies that had no immediate support structure set into place. At one point they attempted to internally convert the original NOS, Application Server, and dynamically generated Web sites from one non-Microsoft platform (Linux/Jrun/Java) to another non-Microsoft platform (Novell/Websphere/Java), but they had a problem finding a consultant with the appropriate skill set who was immediately available.

After they spent a month of searching, experiments, and multiple dead ends, I walked into their building armed with a developer copy of the Windows Server NOS and asked for one server machine with a static IP address and access to the Internet. That was at about 1 p.m. and by 5:30 p.m. (after some BIOS upgrading), I had installed and configured the Windows Server NOS, configured a Web Server and Application Server, and installed all four Web sites. The director was bewildered at the rapid deployment, so much so that he asked me to show him the process step by step.

Over 95% of computer users utilize a Microsoft OS, which also means that most of the VMS software (as most all software) is developed for the Microsoft platform. It's an OS that's understood and has become more intuitive and an intricate part of our daily lives; thus it's the better choice for small or mid-sized companies with limited time and resources (human and otherwise).

Typically, a networking environment opens up shared resources such as files, printers, and an Internet connection. The Windows Server NOS provides a configuration wizard that gets the server up and running within minutes. This also includes a few functions that are required for remote viewing of the VMS software such as a Web server or Terminal Services.

NOS provides more features than will ever be used, but there are a few select features beneficial for a DVS deployment.

Administrative Tools

Microsoft Windows, both for workstations or for servers, comes with a set of computer management tools that help you manage the machine. These are very important and should be pinned to the Start Menu. That can be done by navigating to the Administrative Tools folder in the Control Panel:

START > CONTROL PANEL > ADMINISTRATIVE TOOLS

Inside the Administrative Tools folder is an icon named Computer Management. Right-click on that icon and choose Pin to Start Menu.

The Computer Management console is a pre-configured interface with a number of administrative and troubleshooting components to save time. This console can also be accessed by doing the following:

START > RUN and Type COMPMGMT.MSC

VIII.B. Local Area Networks

Popular LAN operating systems are outlined in Table V. In this section we review three of the most popular ones including Microsoft NT Server, Novell NetWare, and Linux.

Table V. Popular LAN Operating Systems

Windows NT Server is a preemptive multitasking and multithreading network operating system. “Preemptive multitasking” means that the operating system allocates processor time to one or more applications that are running. Alternatively, cooperative multitasking allows each application to control how much CPU time they get and to surrender the processor after a certain amount of time. If an application refuses to yield control of the CPU, all other applications stop running. The term “multithreading” means that programs can start subtasks that are then executed by the operating system in the background. Applications that utilize multithreading can be much more responsive to the user's needs. Windows NT Server provides an application server foundation with tightly integrated file and print sharing, backup and recovery management, and a high level of security by using the NT file system (NTFS) allowing permissions to be set on a file and directory basis.

Novell NetWare has the largest installed base of any LAN operating system on the market. NetWare possesses a number of important security features, including a technology for data encryption that prevents computer hackers from extracting password and other ID information from network traffic. NetWare also features an auditor function that prevents even systems administrators from changing the audit logs, resulting in greater data integrity and administrator confidence in the activity logs. As with Microsoft NT, a major benefit of Novell NetWare is that it works on multiple client operating systems.

Linux (pronounced “lih-nucks”) is a free or very low-cost operating system comparable to traditional and usually more expensive Unix systems. It is a full-featured, multiuser, multitasking operating system that runs on 80386 processors and beyond. Linux comes in versions for all the major microprocessor platforms including the Intel, PowerPC, Sparc, and Alpha platforms. Linux is a complete operating system, including a graphical user interface, X Window System, TCP/IP compatibility, the Emacs editor, and other components commonly found in more comprehensive operating systems. Development of the Linux OS began as a postgraduate project by Linus Torvalds, a student at the University of Helsinki in Finland. The first official version of the Linux OS was released in October 1991, and with the improvement efforts of hundreds of Unix programmers' worldwide, new features have been added every day. Despite its power, Linux is still simple enough for end users that need an inexpensive, efficient OS for accounting, word processing, or Internet browsing.

Network Operating Systems

Modern operating systems have networking capabilities built in. Early PC operating systems such as DOS (and the Windows shell that ran on it) did not; it wasn't until Version 3.11, with Windows for Workgroups, that Microsoft included networking components. As the name implied, that version of Windows was designed to function in a small peer-to-peer local network. Windows NT added authentication server functionality (Microsoft called the authentication server a domain controller), but with the early versions of NT, the focus was still on the LAN, not the WAN. At that time, Microsoft operating systems were not considered scalable enough for enterprise networking, and most Web servers on the Internet were UNIX machines. With Windows 95, it became easier for users to connect to the Internet, and NT 4.0 supported Web services (IIS) that made it easy to host Web sites on the Internet or intranets. Windows 2000 built more heavily on Internet connectivity and added features to the server products that made it more suitable for enterprise-level computing, including a robust directory service (Active Directory), industry-standard security protocols such as Kerberos and IPSec, and load-balancing and clustering support. The next generation of Windows servers, 2003 and 2008, continued this trend and embrace the idea that “the network is the computer” to a larger extent than ever.

The term network operating system (NOS) is used in three different ways:

It is sometimes used to refer to any computer operating system that has built-in networking components, as do all of today's popular PC operating systems. Thus, Windows 9x, NT, 200x, XP, and Vista, along with most distros of Linux, UNIX, and Macintosh, are considered NOSes, whereas MS-DOS and Windows 3.1 and earlier are not.

It is sometimes used to refer to the components of the operating system that make networking possible. For example, today's Windows operating systems include file and print sharing services, which allow the computer to act as a server and share its resources with other systems, and the Client for Microsoft Networks (known as the Workstation service in NT) which allows the computer to connect to and access the shared resources of other systems. These components, along with the protocol stacks on which the network operates, are sometimes referred to as the NOS.

It is sometimes used to refer to the server operating system software—such as Windows NT Server, Windows 200x Server, UNIX, Apple OS X Server, or NetWare—especially when functioning as an authentication server that maintains a security accounts database for the network.

In the following sections, we look at how client/server computing works and discuss both the server software and the client software that work together to enable network communications. We will also take a look at network file systems and how they differ from local file systems as well as the protocols that govern the network communication process.

Client Software

Most modern operating systems can also function as network clients. For example, if you were running Windows 2008 on your computer, you could log on to the network as a user, run programs, and use it as you would Windows Vista. With the exception of NetWare, this is common among many server operating systems. However, it would be inefficient and costly to run Windows Server 2008, for example, as a desktop client as it costs considerably more than the desktop operating system. UNIX is most often used as a server, but Linux has grown in popularity as a desktop/client OS. Mac OS X comes in both client and server forms. Novell doesn't make a client OS of its own; NetWare clients generally run Windows or UNIX operating systems with NetWare client software installed.

This brings up an important point: Client machines don't necessarily have to run an operating system made by the vendor of the network's server software. Macintosh and UNIX-based clients can access Windows servers, Windows and Macintosh clients can access UNIX servers, and so forth. As shown in the Figure 4.16, the Novell client for Windows is used to supply a username and password, which is then sent to a Novell server. The Novell server then uses eDirectory to authenticate the user and to determine what the user is permitted to access, and may access a script to map drives to locations on the network. As a result, the user will see a variety of new drive letters, which allow the user to store files on network servers.

Figure 4.16. The Novell Client

Network Operating Systems

Although we will discuss network operating systems fucntions in some depth in Part Three, at this point you should at least be familiar with the names of the software that manages networking. Those that you are likely to encounter include:

Novell NetWare: Novell NetWare was one of the first network operating systems. It made possible the networking of computers running MS-DOS. NetWare used DOS to boot the server and then installed itself as an alternative operating system. Although today Novell NetWare uses TCP/IP, its original file transfer protocol was IPX. Novell NetWare requires a server runing the server software and client software on all machines. Novell client software is included in recent Windows releases but must be purchased separately for other operating systems. Novell NetWare is the least commonly used of the major NOSs today and is the least likely to be installed in a new, small network.

Microsoft Windows: Current desktop releases of Windows support peer-to-peer networking. In addition Windows server software provides a full range of network services, many of which are discussed throughout this book.

Mac OS X Server: Like desktop versions of Windows, Mac OS X supports a variety of peer-to-peer networking services. However, if you want to use a Mac OS X server, you will need the separate server software.

UNIX: The many variations of the UNIX operating system incorporate TCP/IP as their networking foundation. In this book, we will look at Linux, the open source version of UNIX that is the most commonly used UNIX in small offices.

Note: Linux networking is generally more complex to implement and manage than Windows and Mac OS X, in particular, because Linux has no single graphic user interface (GUI); the tools that you have to manage networking depend on the Linux distribution you purchase. Therefore, the only way to talk about networking that can be certain to apply to all distributions of Linux is look at the command line.

33.1 Introduction

Windows NT has provided an excellent network operating system. It communicates directly with many different types of networks, protocols and computer architectures. Windows NT and Windows 95 have the great advantage of other operating systems in that they have integrated network support. Operating systems now use networks to make peer-to-peer connections and also connections to servers for access to file systems and print servers. The three most widely used operating systems are MS-DOS, Microsoft Windows and UNIX. Microsoft Windows comes in many flavours; the main versions are outlined below and Table 33.1 lists some of their attributes.

Table 33.1. Windows comparisons

Microsoft Windows 3.xx –16-bit PC-based operating system with limited multitasking. It runs from MS-DOS and thus still uses MS-DOS functionality and file system structure.

Microsoft Windows 95 – robust 32-bit multitasking operating system (although there are some 16-bit parts in it) which can run MS-DOS applications, Microsoft Windows 3.xx applications and 32-bit applications.

Microsoft Windows NT – robust 32-bit multitasking operating system with integrated networking. Networks are built with NT servers and clients. As with Microsoft Windows 95 it can run MS-DOS, Microsoft Windows 3.x applications and 32-bit applications.

Novell NetWare

Novell NetWare was the first widely used network operating system. Early releases made it possible to network computers that ran MS-DOS, a single-user operating system that had absolutely no native networking capabilities.

Versions 3.x and 4.x, which are still widely used, rely on the proprietary Novell NetWare protocol stack. However, NetWare 5—the version released about the time this book was written—has taken a different approach. Novell recognized the overwhelming acceptance of TCP/IP and has switched its primary protocol support to the Internet protocols. However, NetWare 5 still provides support for the Novell protocol stack to ensure backward compatibility with existing installations.

As with any NOS, NetWare requires two types of software: server software installed on all servers on the network, and client software installed on all workstations.

7.4 Programmable protocol stack

The concepts of a programmable protocol stack and a wireless network operating system are additional important concepts in the context of network virtualization and virtualization of network functions. Legacy network protocols, as they are, have become increasingly less effective and less efficient in satisfying certain QoS conditions (such as latency), especially when considering the path toward future generation networks. The satisfaction and prediction of specific QoS levels have become increasingly difficult, especially when considering combinations of multiple protocols. Coordination across different protocols has similarly increased in difficulty, especially given the lack of a unified architecture. The upcoming realization of a unique SDN-NFV architecture represents the fertile ground for the growing emphasis on developing reconfigurable protocol stacks.

This trend leads to the question of What exactly is a programmable protocol stack? This paradigm represents the implementation of a software-based environment that supports flexible and adaptive management of protocols and network layers. Reconfiguration refers to actions (such as parameter reassignment, service updating, and replacement of functionalities) according to user/network/environmental requirements.

The idea of programmable protocols, and consequently programmable stacks, is derived from preliminary works at the end of the last century [152]. Given the emergence of applications for multimedia content distribution through networks, the research community started thinking of adaptive/programmable transport protocols, which could have better answered to the requirements – in terms of greater QoS – of end users. This protovirtual system featured an abstraction layer to manage and remotely control the signaling system.

In the very beginning of the current millennium, society enormously increased its mobile internet population together with great evolution of wireless networks. These historical factors increased the degree of heterogeneity along several dimensions, such as the access technology, network model, device, and application requirements. Such an increasingly complex context enforced the idea behind the need of different nature of a network protocol stack, capable to adapt its different layers dynamically to the varying operating environment. A solution called AdaptNet [153] proposed a network protocol stack where different layers (such as application, transport, and link layers) contained adaptive protocols. The system proposal only left the network layer with IP unchanged to facilitate easy deployment while maintaining existing routing infrastructure.

Fig. 7.7 illustrates the adaptive architecture proposed in [153]. The blocks in the stack with gray background are the software-based ones, which can adapt according to network/application variations. The link layer includes an adaptive MAC to seamlessly change MAC characteristics without requiring any additional changes in the existing network infrastructure. The transport layer has an adaptive structure according to mobile hosts. Moreover, this layer includes an adaptive congestion control algorithm, adjusting according to specific operational environments. The application layer at the time of the specification was focused on supporting real-time video streaming. Thus the architecture includes source- and channel-adaptive coding to effectively handle the data and bit error rate fluctuations of the wireless channel.

After the first decade of the 21st century, network virtualization implementations became a reality and actually deployable. Subsequently, the idea of a protocol stack virtualization was able to achieve a higher level of generalization. In 2012 a model-driven framework for reconfigurable protocol stacks was proposed [154]. The realization of a programmable stack capable of supporting various kinds of real-time applications and protocols required the characterization and modeling of the complete system structure, including its traffic classification and constraints. The design includes interfaces for data to real-time applications, effectively providing a real-time intertask communication channel, which is capable to carry reconfigurations of protocols and layer logic blocks. Between 2017 and 2018, SDN and NFV paradigms started to become mature, which influenced the evolution of architectures and practical implementations of fully virtualized protocol stacks, as it is possible to see in newer proposals [155,156].

Fig. 7.8 depicts the logical architecture of Wireless Network Operating System (WNOS). The majority of research community and industry efforts have focused on SDN and NFV since the main interest has been virtualization of routing, network resources, and network functions. Nevertheless, the control problems in wireless networks can require further elements to be considered to optimally allocate resources. Specifically, allocation decisions should take the multilayer characteristics of the network protocol stack into account.

The network abstraction framework is the interface through which the targeted network control problem can be designed according to specific aims of the end-to-end applications. This logic block provides the characterization of network behavior and the centralized definition of the network control problem. The objectives can be defined via APIs to target throughput maximization, low latency, and so on. Furthermore, constraints should be included as well to take the characteristics of the physical network into account.

Next, the automated network control problem decomposition considers the definition of wireless network behavior to divide the targeted network control problem into distributed subproblems with their specific characteristics. According to the network and network control problem structures, decomposition can take different forms and can imply different overhead and complexity. It is important to mention that a network control problem and its decomposition can involve subsets of network layers and protocols, without necessarily applying modifications to all the layers of the stack.

Finally, the programmable protocol stack is a software-based stack that inputs information from the higher logical layers and configures various parameters at each layer, repeating the procedure at each network device. This update of parameters is dynamic according to network changes and end-to-end service requirements. The adaptive nature of the programmable protocol stack also involves the physical layer via the deployment of SDRs to optimize spectrum allocation and wireless resource management.

Fig. 7.9 illustrates the system layout of the Software-Defined Protocol (SDP) system. This system consists of SDP controllers and servers, which contain SDP blocks. In particular, SDP blocks perform processing of the paths of packets. New connections send SDP requests to the SDP controller, which establishes all the functionalities and characteristics (and eventually aggregations of multiple data flows) to release the on-demand protocol stack to satisfy the required QoS. In fact, an SDP controller maps the existing SDP requests onto the available SDP servers. The number of functional blocks involved in the processing path mainly depends on the end-to-end latency requirements. Furthermore, an SDP controller makes decisions on processing procedures for specific traffic flows, on configuration of flow table in the switches and on function blocks in SDP servers.

Fig. 7.9 also showcases the internal logic structure of an SDP server. This entity consists of four main logical blocks: i) control agent, ii) SDP block pool, iii) switch module, and iv) lower-layer interfaces. The control agent receives the control commands from the SDP controller, which are translated into rules for the functional blocks in the switch module. Moreover, the controller also updates the flow tables to assign the packets to their specific flows. The SDP block pool contains different kinds of functions to be performed. The processing carried out by these functions can also be subject to decomposition (flow tables are designed to support this feature).

Next, the packets belonging to a flow are sent to the necessary flow tables they request: i) the main flow table, ii) the user layer flow table, iii) the logical link layer flow table, and iv) the physical layer flow table. The main flow table sorts incoming data packets. Next, the user layer flow table and logical link flow table, respectively, categorize data packets according to the different users and logical links/services/applications they belong to. Finally, the physical layer flow table aims at forwarding the packets to their specific physical channels and interfaces. In Fig. 7.9 the considered interfaces are Ethernet (Eth), Common Public Radio Interface (CPRI), and IP. In this way, SDP servers can connect to other servers or controllers via different kinds of physical channels and also via RRHs.

17.4.2 The application layer

Having discussed and justified the high-level map abstraction and netOS approach, we now discuss the application layer that is above them that makes use of the network API. In SDN, the control plane is not intended to provide comprehensive functionality as the term often denotes in other contexts. Rather, it provides a platform to expose information about and access to the underlying network. The network API provided by the control plane exposes the topology and connectivity of the network as well as a set of services that allow programmatic responses to events in the network.

The applications are the consumers of the services provided by the controller platform. As the NBI lacks a standard in most SDN developments, the degree of granularity of network events communicated from the controller to applications varies.

At one extreme, you might imagine that specific packet events that are visible to the controller could be exposed to applications. For example, if spanning tree were written as an SDN application, actual Bridge Protocol Data Units (BPDUs) along with the metadata of the switch and port could be communicated to the application.

At another extreme, the controller platform might be responsible for maintaining the state information of hosts present in the network and their locations in the network, and for calculating paths between pairs of hosts. The application might only see notification of the arrival of a new host. It could communicate policy for allowing connectivity between host pairs, and the control plane would be responsible for programming the forwarding plane state appropriately to implement this policy.

Alternative Protocol Stacks

You should be aware of two alternative protocol stacks that often run on Ethernet hardware: IPX/SPX and AppleTalk.

IPX/SPX was developed by the Novell Corporation for use by its Novell Netware network operating system. Although Netware now uses TCP/IP, you may encounter a legacy network that uses IPX/SPX. The IPX protocol is roughly analogous to TCP/IP's UDP protocol; SPX, which comes from the OSI protocol model, is roughly equivalent to TCP. For more information on IPX/SPX see http://en.wikipedia.org/wiki/IPX/SPX.

AppleTalk is a protocol stack developed by Apple Computer for Macintosh networks. Based on the OSI reference model, it originally used its own hardware (LocalTalk) but now runs on Ethernet. Most Macintoshes use TCP/IP today, but AppleTalk is also still widely used for printing. Implementations of the AppleTalk protocol stack are available for both Windows and Linux distributions. For more information on AppleTalk, see http://en.wikipedia.orglwikilAppleTalk.

LDAP

Lightweight directory access protocol (LDAP) provides a common open protocol for interfacing and querying directory service information provided by network operating systems. LDAP is widely used for the overwhelming majority of internal identity services including, most notably, Active Directory. Directory services play a key role in many applications by exposing key user, computer, services, and other objects to be queried via LDAP.

LDAP is an application layer protocol that uses port 389 via TCP or user datagram protocol (UDP). LDAP queries can be transmitted in cleartext and, depending upon configuration, can allow for some or all data to be queried anonymously. Naturally, LDAP does support authenticated connections and also secure communication channels leveraging TLS.