◙Cloud◙ Computing
The Metcalfe’s Paradigm or the Heart of the Cloud
In the Land of the LANs, the Ethernet is the King
Introduction
In many networking diagrams, the Internet is depicted with an icon of a cloud, as representing perhaps its complex and dense structure. I really do not know how to put it: A networked cloud or Clouded internetwork, anyways the Internet is made out of many other networks. As we “traveled” into the “cloudnet”, we can see that is made out of other interconnected clouds or units, we have some denominations for them. From Global Area Network [GAN] (internet, in this sense all GANs or internets form the one and only one: Internet) (Panko, 2005), we continue our voyage, passing through: The Wide Area Networks [WANs]; Metropolitan Area Networks [MAN]; Campus Area Networks [CAM]; until we outreach the Personal Area Networks [PANs], or Body Area Networks, [BANs] and Power Line Are Networks [PLANs]. (the object of PANs, or BANs, is to peruse the human body to transmit data from people to people by the simple means of touching each other). (Tomasi, 2005, p. 34-38) Notwithstanding, as soon as we look the Internet’s architecture, we by simply inspection understand that within all its main portions or networks, the Local Area Networks [LAN] are the more numerous structures, and in the “Land of the LANs the Ethernet standard is the king”.
Incidentally, Ethernet the subject of this paper, this is a brief study of the development of Ethernet LANs [E-LANs], a standard which in spite of being more than 30 years used by experts in the networking industry, and consumed by the enterprises and homes, Ethernet still prevalent, but especially for companies that as the Great Catalogues Inc. [GCI], are doing or thinking on doing business over the Web (which is just a part of the Internet perhaps the biggest interactive part of it) and thus participate actively in the eCommerce environment. This paper, which has been prepared for the top management of GCI, documents the conducted researching efforts made by the GCI information systems security [GISS] department, for explaining the importance, shortcomings and prevalence that Ethernet plays for the accomplishment of GCI’s eBusiness goals and objectives (Course, 2008).
The Ethereal Ethernet
The ether is here to state, this concept extracted by Robert “Bob” Metcalfe, from the field of classic theoretical physics, to name his network standard, is still very popular among notable researchers, as its definition is being reformulated, once and again, by investigators of the caliber of such physicists as the Nobel Prize laureate, Frank Wilczek of the Massachusetts Institute of Technology [MIT] and by the latest experiments involving Quantum Chromodynamics or QCD and broken symmetries, done at the 2.4 mile Relativistic Heavy Ion Collider [RHIC] of the Brookhaven Laboratory in New York. Nonetheless, Ethernet networks are what we find today in almost every network, whether be at the enterprises or at homes, it is predominantly and extensively utilized. Metcalfe, and his assistant Boggs, were able to develop a system at Xerox Palo Alto Research Center [PARC] to interconnect their computers with those of the then famous, minicomputer manufacturer, Digital Equipment Corporation [DEC] without the use of a mainframe network (Russell, chap. 5 p. 65-73).
Metcalfe & Boggs (1997) published, in 1976, a seminal paper in the journal “Communications” of the Association for Computing Machinery [ACM] with the title “Ethernet: Distributed Packet Switching for Local Computers Networks” in which they defined what really is Ethernet, or the U.S Patent number 4,063,220, “Multipoint data communication with collision detection” issued by Xerox Corporation on December 13, 1976 (TechFest, 1999). Here is part of the original text: “Ethernet is a branching broadcast communication system for carrying digital data packet among locally distributed computing stations. …” (p. 1). Let us how all started …
The Development of the Ethernet
Ethernet networks located itself, at the center of two extremes of the computing spectrum of the late 70’s. On one hand, they had those “big-irons”, enormous mainframes of centralized connections from remote networks of stations, i.e., dumb-terminals (Terminal Teletypes [TTYs]), with the only computability power of a transceiver with a keyboard plus a monochrome screen of lower resolution managed by an Command line Interface, perhaps Multics or MVS. On the other hand, those isolated main frames, with the parallelism offered by its multiprocessing capabilities. However, Ethernet is distributed and for such decentralized (Metcalfe & Boggs, 1976) perhaps, the main ideas behind Ethernet came from the Arpanet’s design which from the beginning was intensively focused on decentralization for resiliency. LAN technology has facilitated our ability to created distributed networks. The figure 1 shows two diagrams that depicts the evolutionary differences between those early time-sharing system of the 60s; and the distributed networks that characterizes most of the currents networks and internetworks of the present time, most of what we owe to the Metcalfe’s paradigm, i.e., the Ethernet.
Figure 1
In fact, as Metcalfe and Boggs (1976) stated, that Ethernet derives from the telecommunications terminal-computer communication era, as depicted in figure 1. The whole point, of these types of networks, was to connect those dumb terminals to a centralized computing facility. The facility consisted of several separated units: The computer, the front-end processor, the file storage unit, et al. So the challenge that Metcalfe faced at the time, and was able to met rather successfully, was the need for computer-to-computer communication. in which computers were used as a packet switches and for resource sharing, all of what of course, and again, was developed by the direct initiative of ARPA, and names like Bob Taylor, Leo Roberts, Vincent Cert, et al.
Figure 2
Metcalfe op. cit., pointed out in his paper how Ethernet was developed since the times of the Aloha networks’ Menehune, (Figure 2) the Hawaiian version of the Arpanet’s central processor called the Interface Message Preprocessor [IMP] (A packet switching appliance that can be considered as the first router ever, it was designed and built by BBN in Boston, Massachusetts, after winning among 140 companies, a Request For Quotation [RFQ] released from the then ARPA’s Director, Bob Taylor, to built the Arpanet). (p.1-2) Tomasi (2005) states that Metcalfe called the first Ethernet the Alto Aloha Network, changing the name later to Ethernet to making a point that his standard could be used by any computer not just the Xerox’s Alto. Tomasi (idem) added that Metcalfe had had chosen “ether” as meaning of air, atmosphere or heavens. (Chap. 18, p. 572) By reading his paper, I think that what he meant by “ether” was the vital force that connects the computers and transports the signal, the cable, he refers constantly to the “ether” whenever and wherever there is the suggestion that should be a cable instead.
In the beginning, Ethernet (See Appendix A – Figure 3) shared many objectives with other local networks prototypes; such as, MITRE’s Mitrex, Bell Telephone Laboratory’s Spider, and the University of California Irvine’s Distributed Computing System [DCS]. (Metcalfe & Boggs, 1976) However, two years after Ethernet was patented, DEC, Intel and Xerox combined efforts with the objective to standardized an “Ethernet system that any company could use.”
The products of this industrious collaboration appeared a year later, in September 1980, under the released of Ethernet version 1.0, the first specification, labeled “Ethernet Blue Book” or DIX (from the initial of the aforementioned corporations). Version 1.0 specified the application of the coaxial cable 10Base5, aka “thick”, 10 Megabits per second Mb/s Carrier Sense Multiple Access with Collision Detection CSMA/CD protocol, in addition, the networks connected with 10Base5 were called thicknet, the final version of DIX standard was released in 1982, as version 2.0.
From the DIX era, we enter, in 1983, the Institute of Electrical & Electronic Engineers [IEEE] period, with its first release, an Ethernet improved standard developed by the “802.3 Working Group” of the “802 Committee”, titled, “CSMA/CD Access Method and Physical Layer Specifications”. This specification was geared to add hardware interoperability between to different standards and now they are encapsulated and known by the IEEE Standard, Std-802.3 Ethernet.
The IEEE also had continued the improvement of the Ethernet system by creating the standard 802.2, which involved the packetizing of data and the identification of the protocol structure, whereas 802.3 defines the standard used to prevent multiple computers from sending data at the same time with reduces the likelihood of collisions. From 1980 Ethernet standard has been evolving all along in a steady pace and an incremental and significant ways, as presented below by the table 1 (Russell, 2000, TechFest, 1999, Tomasi, 2005, Panko, 2005):
| Ethernet’s Milestones |
| Date | Who | What | Specification | Detail |
| Early 1970s | Metcalfe & Boggs Xerox PARC | Experimenting | Xerox Alto connected to a printer at 2.94 Mb/s |
|
| July 1976 | Metcalfe & Boggs ACM’s Communications Journal | Ethernet: Distributed packet Switching for local Computer Networks |
|
|
| 12/13/1977 | Xerox Corp | Multipoint data communications with collision Detection | U.S. Patent |
|
| 1979-80 | DIX | Ethernet System that anyone could use | Version 1.0 or Ethernet Blue Book | CSMA/CD – 10Base5 – 10Mb/s Thick coaxial cable |
| 1982 | DIX | 1st Ethernet Controller DIX Final version | Version 2.0 |
|
| 1983 | IEEE Working Group of The IEEE 802.3 | 802.3 CSMA/CD | Access Method and Physical Layer Specification |
|
| 1985 | IEEE | 2nd version IEEE Ethernet called “thinnet” | 802.3a | Simply cable and cheaper cable |
| 1987 | IEEE | Fiber Optic Inter-Repeater Link (FOIRL) | [1] 802.3d [2]802.3e | [1] Extend maximum distance between 10Mb/s Ethernet repeaters [2] Based on 1Mb/s twisted pair wiring. |
| 1990 | IEEE | Major advance | 802.3i | 10-Base-T standard Operated over Category [CAT] 3 Unshielded Twisted Pair. Wired in a Star topology fashion. |
| Date | Who | What | Specification | Detail |
| 1993 | IEEE | Attachments over longer distances, up to 2000 meters | 802.3j | 10 Base F (FP, FB & FL) expanded the FOIRL |
| 1995 | IEEE | Performance Improved ten (10) times. | 802.3u | 100Base-T aka Fastnet 100Base-Tx 100Base-T4 100Base-Fx |
| 1997 | IEEE | Full-duplex Ethernet allows concurrency of transmissions from both stations. | [1] 802.3x [2] 802.3y | [1] 100 Mb/s and beyond [2] 100Base-T2 standard for 100 Mb/s CAT 3 |
| 1998 | IEEE | Improved 10 times performance | [1] 802.3z [2] 802.3ac | [1] Gigabit Ethernet 1000Base-SX 850mm laser over multi-mode fiber, 2 10000Base-CX – short haul copper “twinax” Shielded Twisted Pair [STP] [2] Support Virtual LAN VLAN tagging on Ethernet networks |
| 1999 | IEEE | Operation over four pairs of CAT 5 UTP | 802.3ab | 1000Base-T I Gb/s |
| July 2001 | IEEE document | 1562 pages Long document condensing all 802.3 standards | Std 802.3-2000 [1]802.3ad | [1] Link Aggregation |
| 2002 | IEEE |
| 802.3ae | 10 Gb/s |
| 2003 | IEEE |
| Std 802.3af | DTE Power via MDI |
| 2004 | IEEE |
| 802.3ah | Ethernet in the First Mile |
| 2004 | IEEE |
| 802.3ak | 10GBASE-CX4. |
| 2006 | IEEE |
| 802.3an | 10GBASE-T |
| 2006 | IEEE | Creation |
| 802.3 Higher Speed Study Group [HSSG] |
| 2006 | IEEE |
| 802.3aq |
|
| 2007 | IEEE |
| 802.3ap | Backplane Ethernet |
| 2007 | IEEE Project Authorization Request [PAR] |
| 802.3ba PAR status | 40 Gb/s & 100 Gb/s |
Ethernet Security
Still Ethernet has breaches and is not secure as other technologies; this means that many security features are needed to be set in place for obtaining a decent effective Ethernet security protection. It is curious that Metcalfe itself would prognosticate much of the security issues that we are facing today, and he did it much before he developed the Ethernet, in his RFC 602 “The Stockings Were Hung by the Chimney with Care” that he in1973 about weak passwords.
One of the major problems that I have is with IP spoofing. Well, it is TCP/IP embedded in the architecture of Ethernet, since an Ethernet frame in its header contains or encapsulates an IP packet. Even though Ethernet operates at the physical and data link layer, divides this layer in two sub-layers the MAC-Media Access Control and LLC – Logical Link Control, it is at the LLC that many thing could happen and evenly at the MAC level.
At the AT&T Bell labs, Bellovin (1989) wrote a paper on security problems of TCP/IP protocols and IP address spoofing attacks, telling us that IP spoofing were already introduced circa 1980s by attackers for hide their true identities (p. 1-17). The reader might already thinking, “wait a minute” you can only spoofed an IP packet, so these attacks are only activities done at the network and transport layers, but the response could be that the Ethernet address or MAC address (the 48-bits) of the sending machine can be spoofed also. It is a matter of fact, that it is spoofed most often that we can normally think is possible. There many problems and nothing but nothing is 100% completely secure out there in the Internet. Another example of Ethernet security are the Address resolution Protocol [ARP] attacks, heretofore, Ethernet attacks protection is a field by itself in information security.
Conclusion
As Ethernet is being upgraded and reviewed, the case of the IEEE 802.11 Wi-Fi standard is becoming more and more fashionable and it is used at homes, as by the enterprises. However, and spite of Wi-Fi’s advantages, for instance, mobility, the experts are talking about already, M-Commerce. They want to define the transactions done over smart wireless phones and Personal Digital Assistants [PDAs]. Moreover, another advantage of Wireless networks is that there is not need of the expensive wiring costs and procedures or as Metcalfe & Boggs (1976) would put it, in a wireless network, you do not need to worry about the “ether in the Ethernet” anymore; because, the connection is, casually, the mere air. Anyways, this type of network juxtaposes serious security challenges as well. At may firm we prefer to use a combination, like some kind of Pareto networking strategy, i.e., 80% Ethernet network at the border and within our perimeter to support the 20% or little more of wireless or mobile secure connections.
