外文翻譯---校園網(wǎng)絡(luò)多媒體的演變_第1頁
已閱讀1頁,還剩9頁未讀, 繼續(xù)免費閱讀

下載本文檔

版權(quán)說明:本文檔由用戶提供并上傳,收益歸屬內(nèi)容提供方,若內(nèi)容存在侵權(quán),請進(jìn)行舉報或認(rèn)領(lǐng)

文檔簡介

1、<p><b>  一、英文原文:</b></p><p>  The Evolution of Campus Networks Towards Multimedia</p><p>  Ciro A. Noronha Jr. Fouad A. Tobagi</p><p>  4 The Backbone</p>

2、<p>  The campus backbone carries the aggregate traffic from the sub networks. From a theoretical point, of view, the design of the backbone is no different, from the design of the sub networks themselves; the steps

3、 taken in section 3 can be repeated here, with the difference that the network being designed is the backbone, and its "users" are the sub networks. The difference is one of scale. In this section, we will brie

4、fly comment on the effect of the backbone traffic on the aggregate throughput of </p><p>  4.1 Connecting the Sub network to the Backbone</p><p>  The main difference between the traffic to/from

5、 the backbone and the internal sub network traffic is the fact that former is concentrated in a single point (the backbone connection); in other words, all the traffic from the the sub network to the backbone is directed

6、 to the backbone connection, and all the traffic from the backbone to the sub network originates in the same location. In the other hand, the sources and destinations of the intra-sub network traffic are spread over the

7、sub network.</p><p>  The effect of adding the backbone connection in the aggregate sub network bandwidth' depends on its topology. In sub networks organized as trees, the congestion in the branches near

8、 the location of the backbone connection might limit the aggregate throughput to a value lower than what is attained when no backbone traffic is present. In the star configuration, however, the high-speed channel is the

9、ideal location for attaching the backbone connection, and the aggregate throughput in this case is </p><p>  The actual increase or decrease in the aggregate bandwidth will be a function of a, the fraction o

10、f the traffic that, is intra-sub network, and A, the ratio between the traffic from the backbone to the traffic into the backbone (defined in section 2.).The increase is higher for small values of a (most of the traffic

11、is inter sub network).</p><p>  4.2 Structure of the Backbone</p><p>  In this section, we will give a number of scenarios that illustrate how the structure of the backbone can change as a funct

12、ion of the size of the campus network; we will use figure 3 to determine the required capacity. The results are shown in table 1; note that the solutions for the structure of the backbone are not unique, and we have list

13、ed one of such solutions for each case.</p><p>  In figure 5 we show one possible backbone topology which has all the main elements we have been discussing. In the figure we show the sub networks being serve

14、d by switching hubs. The switching hubs serving sub networks with little external traffic can be grouped by an ATM multiplexer before reaching the ATM switch; this multiplexer can perform some local switching or rely com

15、pletely in the central switch. Hubs generating larger traffic can be connected directly to the ATM switch. The switch is a</p><p>  5 Adding multimedia to the other OSI layers</p><p>  In this s

16、ection we discuss the issues raised by the addition of multimedia services on the other layers of the OSI model.</p><p>  5.1 The Medium Access Control (MAC) Layer</p><p>  The synchronous natur

17、e of audio and video traffic dictates that data be delivered to its destination within strict timing constraints. Failure to deliver data on time results in a discontinuity of the video or a degradation in the quality. T

18、he busty nature of data traffic, on the other hand, means that a station transmits data in an unpredictable fashion, and the amount of data that the station transmits in a burst is also random.</p><p>  Both

19、 types of traffic are to be supported by the same local network. Indeed, multimedia applications naturally involve both types of traffic simultaneously; existing local area networks are expected to carry audio and video

20、traffic alongside existing data applications. Mixing the two types of traffic on the same network requires special attention, particularly where shared resources are contended to by both types. The bandwidth available on

21、 local area network segments is one such shared resource</p><p>  The CSMA/CD protocol (IEEE 802.3) does not differentiate among the different traffic types. In essence, it operates as follows. A station wit

22、h a packet ready for transmission transmits the packet as soon as the channel is sensed idle. If it collides with another packet, then it attempts again but after incurring a random rescheduling delay. The mean reschedul

23、ing delay is doubled with each collision incurred. This is clearly inadequate for real-time traffic. Not only is the delay incurred by pack</p><p>  IEEE 802.5 is a token protocol devised for a ring network.

24、 It uses a single token, in the sense that a station that has completed transmission will not issue a new token until the busy token returns to it. Since in a ring network the connection to the medium is active, a priori

25、ty scheme can be implemented by assigning dynamically a priority value to the token, and by restricting access to the ring to packets of priority equal or higher than the currently assigned value. This scheme can be appr

26、op</p><p>  In FDDI (ANSI X3T9.5), another scheme with multiple priority levels is employed, by the use of a set of timers, that regulate for how long the station can transmit traffic of each priority.</p

27、><p>  In summary: the assignment of different types of service for real-time and data traffic at the MAC layer is very important when communicating multimedia information. This is not offered in the widely dep

28、loyed IEEE 802.3 Other networks, such as the IEEE 802.5 token ring and FDDI can provide this functionality, although many vendors have chosen not to implement it. Given the large base of deployed networks, it is not reas

29、onable to expect that changes in the MAC layer to support multimedia would be</p><p>  5.2 The Network Layer</p><p>  As the changes described before take place in the network, the route to use

30、when moving information from segment to segment becomes increasingly important. In simple topologies, such as star with the switching element in the center, routing is trivial because there is only one route from the sou

31、rce to the destination. However, in practice multiple routes might be provided between a given pair of stations, for reliability and/or increased.</p><p>  Currently, the path taken by a packet from its sour

32、ce to its destination through the network is determined by the bridges and routers that define the network topology. Transparent bridges use the spanning tree algorithm, i.e., the bridges, working together, identify a su

33、bset of the network topology that constitutes a spanning tree, and direct all traffic through this spanning tree. Links not on the spanning tree are not used and kept in reserve, to be activated in case of failure of a l

34、ink in th</p><p>  Routers, on the other hand, have the ability to use multiple spanning trees; as a matter of fact, each router forwards the packets over the spanning tree having itself as the root. Many al

35、gorithms exist to compute those routes in a distributed fashion (RIP,OSPF, IS-IS, etc), but they all employ the cost of the link as the metric used when computing the spanning tree (the cost of a link is an arbitrary fun

36、ction, inversely proportional to the link’s bandwidth). Routers can recover from network fai</p><p>  Of special importance is the routing of multicast traffic (video-conference is an example of a situation

37、where this kind of traffic is generated). For multicast traffic, two kinds of routes can be identified: minimum delay routes and minimum cost routes. Current multicast routers use the minimum delay criterion to compute t

38、he routes, but this might lead to unacceptable network loading. Therefore, new routing algorithms with the following characteristics are needed:</p><p>  ?Use of multiple spanning trees.</p><p>

39、  ?Use of multiple routes between sources and destinations.</p><p>  ?Use of different criteria for routing the streams(delay, cost or a combination thereof).</p><p>  ?Use of a system of priori

40、ties that takes into account the real-time nature of the traffic, and not only different link costs for each priority.</p><p>  5.3 Transport Protocols</p><p>  Existing transport protocols have

41、 been designed and implemented to support data applications in which the traffic is usually busty. Flow control procedures are embedded in the protocols to properly pace the flow between end users and achieve efficient u

42、tilization of network resources. For data applications, reliability is an absolute goal and is achieved by an error detection and retransmission scheme. Such transport protocols, however, are not appropriate for the deli

43、very of video data which is</p><p>  5.4 Higher Layers</p><p>  For video services, the application layer plays an important role in managing video communication among nodes on the network, and

44、in securing the necessary network resources to support effectively that communication. Accordingly, the development of video specific application layer protocols is a clear necessity. In this respect, at the present time

45、, little has been done in that area. The desired functionality includes:</p><p>  ?Conference control and management.</p><p>  ?Managing different encoding techniques.</p><p>  ?Ide

46、ntification and use of resources.</p><p>  ?Encryption.</p><p>  6 Measurements in the Stanford Campus Network</p><p>  In this section we present a series of measurements performed

47、 in the Stanford Campus Network (SUNET), to characterize its current performance and illustrate some of the issues discussed in section 5.</p><p>  6.1 Description of the Stanford Campus Network (SUNET)</

48、p><p>  The Stanford Campus Network (SUNET) is composed of a set of IP subnetworks3, interconnected by a combination of FDDI and Ethernet backbones. Figure 6 shows the organization of the backbone. There is a b

49、asic Ethernet backbone, which is composed by a number of segments interconnected by bridges, forming a mesh topology. This backbone connects to all sub networks, and can be used to carry the traffic in event of a failure

50、 of the main FDDI backbones. There are two FDDI networks in the backbone. The </p><p>  There are 124 IP sub networks on campus, each one with an average of 119 addresses. Figure 7 shows the histogram of the

51、 number of addresses in the sub networks. Note that some nodes, such as AppleTalk gateways, have multiple IP addresses, which are dynamically allocated to other nodes using this protocol.</p><p>  6.2 Measur

52、ements</p><p>  We have conducted a number of measurements over the Stanford Campus Network to determine what ranges of throughputs are achievable in practice. For all the measurements described in this sect

53、ion, we will consider the subset of the Stanford Campus Network depicted in figure 8. </p><p>  Figure 9 shows the result of throughput measurements, under TCP/IP, on the Stanford Campus Net- Figure 8: The s

54、ubset of the Stanford Network used in the measurements work. The labels in the plot correspond to the hosts indicated in figure 8. All measurements consisted of opening a TCP connection, transmitting 4 Mbytes of random d

55、ata, and closing the TCP connection. The plots give the throughput as a function of the TCP transmit buffer size. The measurements were taken on a weekday afternoon.</p><p>  Curve 1 was represents communica

56、tion from F to H using a two-node completely unloaded FDDI ring. The maximum throughput was in the order of 17 Mb/s, and was limited by the protocol processing in the hosts. Curves 2 and 3 represent the throughput betwee

57、n two DEC 5000/240 workstations using Ethernet. For curve 2, the workstations were in a short unloaded Ethernet segment; for curve 3, they communicated across a bridge. We can see that the maximum throughput is quite hig

58、h (9 Mb/s for curve 2, and 8</p><p>  Curves 4 and 5 represent communication between system A and systems C and E. Since systems C and E have slower CPUs, a lower throughput is achieved due to protocol proce

59、ssing. However, the two curves are virtually indistinguishable, indicating that the bridge really did not have any impact on performance, as far as throughput is concerned.</p><p>  Curve 6 represents the th

60、roughput from F to G, while curve 8 has the throughput from A to G. This pair of curves shows the effect of processing power in the router. Since the source and destination systems are of the same kind in both cases, and

61、 topologically are placed in similar positions in the network, any difference in performance has to be explained by the connection to the backbone. And, indeed, figure 8 indicates that router Rl, which provides connectio

62、n to the backbone for node A, is a </p><p>  In summary, our measurements have indicated that:</p><p>  ? The bridges in use do not introduce any degradation in the available throughput.</p&g

63、t;<p>  ? Some of the current routers pose severe limitations to the throughput achievable in practice, due to their limited processing capabilities.</p><p>  ? Protocol processing in the hosts is sti

64、ll an important factor in determining the throughput achieved by the less-powerful machines.</p><p>  ? In spite of the data traffic, large end-to-end throughputs can be obtained, it the hosts have fast enou

65、gh CPUs to do the protocol processing. For high-end machines, TCP/IP processing is not a bottleneck when communicating over Ethernet, but it still is the determining factor when using FDDI.</p><p><b> 

66、 二、英文翻譯:</b></p><p>  校園網(wǎng)絡(luò)多媒體的演變</p><p><b>  4 骨干網(wǎng)</b></p><p>  校園骨干網(wǎng)傳送從子網(wǎng)匯集來的通信量。從理論上來講,與子網(wǎng)相比,骨干網(wǎng)的設(shè)計沒有什么特別;第三部分的那些步驟可以在這里重復(fù),只是被設(shè)計的網(wǎng)絡(luò)變成骨干網(wǎng),而用戶變成了子網(wǎng)。在比例上有所不同而已。在這一部

67、分,我們將簡短地討論骨干網(wǎng)匯集子網(wǎng)吞吐量后進(jìn)行通信的效果(這個在第三部分被忽略了),之后我們將計算在不同假設(shè)下骨干網(wǎng)的通信量。</p><p>  4.1 子網(wǎng)和骨干網(wǎng)的連接</p><p>  骨干網(wǎng)通信和內(nèi)部子網(wǎng)通信的主要不同是前者著重于單個點(骨干網(wǎng)連接);也就是說,從子網(wǎng)到骨干網(wǎng)的所有通信量被骨干網(wǎng)的連接所引導(dǎo),并且這些通信量源于相同的地方。另一方面,內(nèi)部子網(wǎng)通信的信源和信宿通過子

68、網(wǎng)進(jìn)行傳播。</p><p>  在匯聚子網(wǎng)帶寬上添加骨干網(wǎng)連接的效果取決于它的網(wǎng)絡(luò)拓?fù)洹.?dāng)子網(wǎng)為樹型拓?fù)鋾r,骨干網(wǎng)連接附近的分支產(chǎn)生的擁塞將可能限制吞吐量的匯聚,這個值將比沒有骨干網(wǎng)通信存在時還低。然而在星型拓?fù)渲校咚傩诺朗沁M(jìn)行骨干網(wǎng)連接的理想地點,并且這種情況下匯聚吞吐量往往高于沒有通信來自骨干網(wǎng)時的值。這個原因很簡單:內(nèi)部子網(wǎng)通信在信源和信宿段都使用帶寬,但是骨干網(wǎng)的通信僅僅在信源或者信宿使用帶寬,這將引

69、起較大的吞吐量。</p><p>  在匯聚帶寬中的實際增長或降低,是內(nèi)部子網(wǎng)通信的分?jǐn)?shù)a和骨干網(wǎng)來去通信量的比A(第二部分有定義)的函數(shù)。這個增長要比a的最低值要高(大多數(shù)的通信都來自于內(nèi)部子網(wǎng))。</p><p>  4.2 骨干網(wǎng)的結(jié)構(gòu)</p><p>  這一部分,我們將給出許多的假設(shè),以解釋骨干網(wǎng)的結(jié)構(gòu)如何能夠變換為校園網(wǎng)大小的函數(shù);我們將使用圖3來決定所

70、需的容量。表格1顯示了結(jié)果;注意對于骨干網(wǎng)結(jié)構(gòu)的解決方法不是唯一的,我們列出了每種情況中的一種。</p><p>  在圖5中,我們給出了一種可能的骨干網(wǎng)拓?fù)?,它具有我們之前討論的所有主要元素。在圖中,我們指出了擁有交換型集線器的子網(wǎng)。在到達(dá)異步轉(zhuǎn)移模式交換機之前,交換型集線器使得子網(wǎng)沒有外部通信匯集,而這些通信量可以被異步轉(zhuǎn)移模式多路復(fù)用器所分組;這個復(fù)用器能夠完成本地交換以及完全地依靠中心交換機。產(chǎn)生大量通信

71、的集線器能夠直接連接到異步轉(zhuǎn)移模式交換機。這個交換機對于連接中心共享資源來說是最好的設(shè)備,例如視頻服務(wù)器。這個異步轉(zhuǎn)移模式交換機自己本身能夠依靠通信需求以分級方式工作。</p><p>  5 添加多媒體到OSI(開放式系統(tǒng)互聯(lián)參考模型)的其他層中</p><p>  在這一部分,我們將討論由于OSI模型的其他層中添加了多媒體服務(wù)而引起的問題。</p><p>  

72、5.1 媒體訪問控制層</p><p>  音頻和視頻通信的同步性規(guī)定了數(shù)據(jù)應(yīng)該在嚴(yán)格的時間約束下傳輸?shù)侥康牡兀瑳]有準(zhǔn)時傳輸數(shù)據(jù)將導(dǎo)致視頻的不連續(xù)或者質(zhì)量的降低。另一方面,數(shù)據(jù)傳輸?shù)耐话l(fā)性,意味著一個站點將以一種不可預(yù)知的方式傳輸數(shù)據(jù),并且一個脈沖中站點傳輸?shù)臄?shù)據(jù)數(shù)量也是隨機的。</p><p>  兩種類型的通信都得到了相同的本地網(wǎng)的支持。事實上,多媒體應(yīng)用本來就同時包括了這兩種類型的通

73、信,依靠現(xiàn)有的數(shù)據(jù)應(yīng)用,現(xiàn)有的局域網(wǎng)也有望支持音頻和視頻通信。在同一個網(wǎng)絡(luò)中混合這兩種通信類型需要特別的注意,特別是在它們爭奪共享資源的地方。局域網(wǎng)各段的可用帶寬就是這樣一種共享資源,并且是這一部分的主題。它被所有點產(chǎn)生的全部通信所共享,并且經(jīng)由媒體訪問控制協(xié)議訪問。如果理想化的有一種協(xié)議像媒體訪問控制協(xié)議一樣包含能夠預(yù)存同步通信的帶寬需求,以此來消除突發(fā)通信產(chǎn)生的任何影響,那該多好。</p><p>  帶有沖

74、突檢測的載波偵聽多路存取協(xié)議(IEEE802.3)沒有區(qū)別這兩種不同的通信類型?;旧希侨缦旅嬉粯舆\作的。當(dāng)信道被檢測到空閑時,站就會把準(zhǔn)備好的分組發(fā)送出去。如果和另一個分組相沖突的話,會在引起一個隨機重組延遲后,再一次嘗試發(fā)送。這個平均的重組延遲兩倍于單個沖突引起的。很明顯的不適于實時通信。這個分組引起的延遲不僅是不確定的,而且由于指數(shù)后退算法,變化會非常大。然而,因為在最小負(fù)載的情況下,這個協(xié)議運行良好,在網(wǎng)絡(luò)被分割到確保條件符

75、合的情況下,它仍然可以繼續(xù)使用。</p><p>  IEEE802.5是一個為環(huán)狀網(wǎng)絡(luò)設(shè)計的令牌協(xié)議。它使用單個令牌,在被占用的令牌沒有返回之前,完成了發(fā)送的站點不會傳送新的令牌。因為在一個環(huán)狀網(wǎng)絡(luò)中,介質(zhì)的連接是積極的,一個優(yōu)先權(quán)方案可以通過這樣來實現(xiàn),動態(tài)分配一個優(yōu)先值給令牌,對于優(yōu)先值等于或大于當(dāng)前分配值的分組,限制其對環(huán)的訪問。只需簡單地給予同步通信比突發(fā)數(shù)據(jù)通信更高的優(yōu)先權(quán),這個方案就會很適合用來在同

76、一信道整合兩種類型的通信。</p><p>  在光纖分布式數(shù)據(jù)接口協(xié)議(ANSI X3T9.5)里,因為一批用于校準(zhǔn)站點傳送每個擁有優(yōu)先權(quán)通信時間的計時器的應(yīng)用,另一項有著多倍優(yōu)先權(quán)等級的方案得到了使用。</p><p>  總結(jié):當(dāng)傳送多媒體信息時,位于介質(zhì)訪問控制層的實時以及數(shù)據(jù)通信,它的不同類型服務(wù)的分配有著非常重要的作用。在廣泛應(yīng)用的IEEE802.3協(xié)議中,不存在這樣的問題。其

77、他類型的網(wǎng)絡(luò),諸如IEEE 802.5令牌環(huán)以及光纖分布式數(shù)據(jù)接口,它們能夠提供這樣的功能,盡管許多網(wǎng)絡(luò)商選擇不去實現(xiàn)它們。由于現(xiàn)有網(wǎng)絡(luò)的巨大基礎(chǔ),讓介質(zhì)訪問控制層產(chǎn)生變化以支持多媒體是沒有道理的。因此,必須尋找其他的選項(例如要確定網(wǎng)絡(luò)在足夠低的負(fù)載下運行,也就是說,這個部分討論的問題不會出現(xiàn))。</p><p><b>  5.2 網(wǎng)絡(luò)層</b></p><p>

78、  在之前描述的問題沒有出現(xiàn)在網(wǎng)絡(luò)之前,在段間傳送信息,路由的使用會逐漸變得重要。在簡單的拓?fù)錂C構(gòu)中,例如在中心有著交換部分的星型網(wǎng)絡(luò),因為從信源到信宿只有一條路由,所以路由的作用變得微不足道。然而實際上,為了可靠性和增長了的吞吐量,在已有的一對站點間,復(fù)合路由可能會得到應(yīng)用。</p><p>  通常,網(wǎng)絡(luò)中從信源到信宿的一個分組所傳送的路徑是由那些確定網(wǎng)絡(luò)拓?fù)浣Y(jié)構(gòu)的網(wǎng)橋和路由器決定的。透明網(wǎng)橋使用生成樹算法,

79、也就是,所有的網(wǎng)橋一起工作,將網(wǎng)絡(luò)拓?fù)渲袠?gòu)成一棵生成樹的子集確定出來,然后命令所有的通信都經(jīng)過這顆生成樹。不在生成樹上的鏈接不會生效,保留下來,以便于生成樹上的鏈接失效時能夠得到使用。而且,一個拓?fù)浣Y(jié)構(gòu)確定的生成樹本質(zhì)上是隨機選擇的,并不是在任何情況下都適用的。這樣會產(chǎn)生兩個結(jié)果:首先,多余鏈接上的額外容量從不使用;其次,由于那棵特別的生成樹的使用,某些段會產(chǎn)生通信堵塞,然后網(wǎng)絡(luò)瓶頸就會更加明顯。我們已經(jīng)在第三部分討論過通信堵塞的問題。

80、</p><p>  另一方面,路由器有能力去使用復(fù)合生成樹;事實上,每個路由器可以將它自己視為生成樹的根部,然后通過生成樹將分組轉(zhuǎn)發(fā)出去。許多算法都用來計算分布式結(jié)構(gòu)中的路由,但是當(dāng)計算生成樹時,它們會將使用鏈接所需的損耗作為公制(每條鏈接的損耗都是一個任意的函數(shù),與這個鏈接的帶寬成反比)。路由器可以從網(wǎng)絡(luò)故障中恢復(fù)過來,而且可以利用所有可用的鏈接的帶寬。然而在大多數(shù)情況下,一個信源和信宿之間的所有通信只會在相

81、同的路由中進(jìn)行。再者,通常情況下,當(dāng)計算路由時,路由器不會重視實時的請求(例如延遲),并且,設(shè)置了優(yōu)先權(quán)的系統(tǒng)明顯地不適于多媒體通信。</p><p>  多播通信的路由特別重要(視頻會議就是一種會產(chǎn)生這種類型通信的環(huán)境的例子)。對于多播通信,有兩種路由必須加以區(qū)別:最小延遲路由和最小損耗路由。通常多播通信使用最小延遲標(biāo)準(zhǔn)去計算路由,但是這樣可能會導(dǎo)致不受歡迎的網(wǎng)絡(luò)負(fù)載。因此,需要一些擁有以下特點的新的路由算法:

82、</p><p><b>  ?使用復(fù)合生成樹。</b></p><p>  ?在信源和信宿之間使用復(fù)合生成樹。</p><p>  ?對于發(fā)送數(shù)據(jù)流(延遲,損耗,或者是兩者的混合)使用不同的標(biāo)準(zhǔn)。</p><p>  ?使用一個擁有優(yōu)先權(quán)的系統(tǒng),它不僅對于每個優(yōu)先權(quán)的不同鏈接損耗,而且對于處理通信的實時性都很重視。&l

83、t;/p><p><b>  5.3 傳輸協(xié)議</b></p><p>  現(xiàn)有的傳輸協(xié)議已經(jīng)被設(shè)計和實現(xiàn)用于支持突發(fā)通信比較繁忙處的數(shù)據(jù)應(yīng)用。數(shù)據(jù)流控制程序被嵌入進(jìn)協(xié)議里,使其能適當(dāng)?shù)匕才抛罱K用戶之間的數(shù)據(jù)流,同時能有效地利用網(wǎng)絡(luò)資源。對于數(shù)據(jù)應(yīng)用,可靠性依靠錯誤檢測和中繼方案實現(xiàn),是根本的目標(biāo)。然而,諸如這些傳輸協(xié)議,對于視頻數(shù)據(jù)的傳送都是不合適的,這些視頻數(shù)據(jù)很穩(wěn)定

84、,對時間要求嚴(yán)格,并且大多數(shù)情況下能夠容忍一定程度的數(shù)據(jù)丟失。結(jié)果,新的傳輸協(xié)議,為了支持?jǐn)?shù)據(jù)流通信,必須對現(xiàn)有的協(xié)議進(jìn)行添加和修正。更加精確的是,在傳輸層,具有能將視頻、音頻以及其他數(shù)據(jù)組成數(shù)據(jù)包傳送的方法是很重要的。在其他情況下,比如對于數(shù)據(jù)的邏輯分組(比如一個視頻幀的頭部和尾部)給出指示,以及將媒體資源,媒體編碼和媒體會議復(fù)用起來,都可能變得很重要。同樣很重要的是,在數(shù)據(jù)流里嵌入實時信息的能力,以允許屬于這個數(shù)據(jù)流的數(shù)據(jù)能夠準(zhǔn)時地

85、傳送,并且能夠到達(dá)在信宿多條數(shù)據(jù)流的同步。為了避免信宿和中轉(zhuǎn)路由器的擁塞,信源處的速率控制被證明是極其有用的。最后,當(dāng)通信速率非???,并且用戶桌面不是一個高速的終端工作站時,協(xié)議的效率,較低的處理耗費以及協(xié)議執(zhí)行的方便性是最重要的。</p><p><b>  5.4 更高層</b></p><p>  對于視頻服務(wù),管理網(wǎng)絡(luò)節(jié)點中的視頻通信,以及保護必需的網(wǎng)絡(luò)資源以

86、有效的支持視頻通信,應(yīng)用層起了很重要的作用。因此,視頻特殊應(yīng)用層協(xié)議的發(fā)展是一個明顯的需求。當(dāng)前,在這個方面,這個領(lǐng)域還是空白。需要的功能包括:</p><p><b>  ?會議控制和管理。</b></p><p>  ?管理不同的編碼技術(shù)。</p><p>  ?鑒定以及利用資源。</p><p><b>

87、  ?加密技術(shù)。</b></p><p>  6 斯坦福校園網(wǎng)的一些計算</p><p>  在這一部分,我們進(jìn)行了一系列有關(guān)于斯坦福校園網(wǎng)的計算,以得出其當(dāng)前的性能,同時會闡釋在第五部分討論的一些問題。</p><p>  6.1 斯坦福校園網(wǎng)的介紹</p><p>  斯坦福校園網(wǎng)是由一批IP子網(wǎng)構(gòu)成的,通過光纖分布式數(shù)據(jù)接口

88、以及骨干以太網(wǎng)的混合體互連的。圖6指出了骨干網(wǎng)的集合。由許多網(wǎng)橋連接的段組成的基本骨干以太網(wǎng),形成了網(wǎng)狀的拓?fù)浣Y(jié)構(gòu)。這個骨干網(wǎng)連接了所有的子網(wǎng),當(dāng)主要的光纖分布式數(shù)據(jù)接口骨干網(wǎng)故障時,可以用來傳送通信量。骨干網(wǎng)中有兩個光纖分布式數(shù)據(jù)接口網(wǎng)絡(luò)。第一個,也就是圖6左邊那個,基本上適合骨干以太網(wǎng)平行的;沒有路由器連接到它。以太網(wǎng)到光纖分布式數(shù)據(jù)接口網(wǎng)絡(luò)的網(wǎng)橋用于組成單個的邏輯網(wǎng)絡(luò)。第二個,也就是圖6右邊那個,僅僅只有路由器能連接到它,邏輯上和

89、第一個明顯不同。所有的IP子網(wǎng)通過路由其同骨干以太網(wǎng)相連;在圖6右邊已經(jīng)有一些利用路由器連接到了光纖分布式數(shù)據(jù)接口網(wǎng)絡(luò)。一天平均下來,每個光纖分布式數(shù)據(jù)接口的通信量相對來說比較?。?0kb/s左右。最繁忙小時的最繁忙IP子網(wǎng)平均以1.2Mb/s的速度將信息發(fā)送到骨干網(wǎng)。</p><p>  所做的計算表明最繁忙小時子網(wǎng)間的變化,例如,1992年12月14號,覆蓋主要校園網(wǎng)的子網(wǎng)最繁忙小時是晚上10點到11點,而同

90、一天醫(yī)療中心子網(wǎng)的最繁忙小時卻是從下午5點到6點。</p><p>  這個校園網(wǎng)里有124個IP子網(wǎng),平均每個有119個地址。圖7畫出了子網(wǎng)中地址數(shù)量的柱狀圖。注意那些節(jié)點,比如AppleTalk協(xié)議網(wǎng)關(guān),就有多個IP地址,當(dāng)使用這個協(xié)議時,就會動態(tài)地分配給其他節(jié)點。</p><p><b>  6.2 運算</b></p><p>  我們

91、從斯坦福校園網(wǎng)導(dǎo)出了許多計算,以確定實際中有多大范圍吞吐量可用。對于這部分描述到的所有運算,我們將考慮圖8中描繪的斯坦福校園網(wǎng)的子集。</p><p>  在傳輸控制和網(wǎng)際協(xié)議下,依據(jù)圖8所示的斯坦福校園網(wǎng),圖9給出了吞吐量計算的結(jié)果:運算工作中所用到的斯坦福校園網(wǎng)的子集。圖中的標(biāo)簽對應(yīng)于圖8所指的主機。整個計算過程包括,打開一個傳輸控制協(xié)議連接,發(fā)送一個4M字節(jié)的隨機數(shù)據(jù),然后關(guān)閉傳輸控制協(xié)議連接。這幅圖將吞吐

92、量作為傳輸控制協(xié)議傳輸緩沖大小的函數(shù)。數(shù)據(jù)是在一個平日下午采集的。</p><p>  曲線1表示從F到H兩個節(jié)點完全利用光纖分布式數(shù)據(jù)接口環(huán)狀網(wǎng)的通信情況。正常狀態(tài)下,最大吞吐量能達(dá)到17Mb/s,同時會受到主機協(xié)議處理的限制。曲線2和3代表兩臺美國數(shù)字設(shè)備公司5000/240型號的工作站使用以太網(wǎng)時之間的吞吐量。對于曲線2,工作站較少利用以太網(wǎng)網(wǎng)段;對于曲線3,它們通過網(wǎng)橋進(jìn)行通信。我們可以看到最大吞吐量是相

93、當(dāng)?shù)母撸ㄇ€2達(dá)到9Mb/s,曲線3達(dá)到8Mb/s)。這個不同是由于網(wǎng)絡(luò)N1負(fù)載較小,而網(wǎng)絡(luò)N2則相反。</p><p>  曲線4和5表明了系統(tǒng)A,C,E之間的通信。因為系統(tǒng)C和E的處理器較慢,協(xié)議處理也會變慢,所以會得到較小的吞吐量。然而,這兩條曲線事實上很難區(qū)分,它們表明在網(wǎng)絡(luò)性能上,網(wǎng)橋不起任何作用,更談不上吞吐量。</p><p>  曲線6表明了從F到G的吞吐量,而曲線8則反映

94、了A到G的吞吐量。這對曲線指出了路由器處理能力的效果。因為信源和信宿系統(tǒng)在這兩方面是一樣的類型,網(wǎng)絡(luò)拓?fù)渖弦蔡幱谕瑯拥奈恢茫虼诵阅苌系牟煌坏貌煌ㄟ^骨干網(wǎng)的連接來闡釋。并且,事實上圖8指出了型號為思科CSC-2的路由器R1為節(jié)點A提供了到骨干網(wǎng)的連接,而型號為思科CSC-3路由器R2提供了到節(jié)點F的服務(wù)。曲線7給出了A到F的吞吐量,也進(jìn)一步證實了性能上的不同。</p><p>  總結(jié),上面的計算可以表明:&l

95、t;/p><p>  ?在可用的吞吐量上,網(wǎng)橋不會引起任何消極的作用。</p><p>  ?由于受到限制的處理能力的原因,實際中目前一些路由器會引起吞吐量的限制。</p><p>  ?對于較低性能的機器,主機上的協(xié)議處理仍然是決定吞吐量的重要因素。</p><p>  ?不管數(shù)據(jù)通信如何,當(dāng)主機擁有足夠快的處理器處理協(xié)議時,可以獲得巨大的端到

溫馨提示

  • 1. 本站所有資源如無特殊說明,都需要本地電腦安裝OFFICE2007和PDF閱讀器。圖紙軟件為CAD,CAXA,PROE,UG,SolidWorks等.壓縮文件請下載最新的WinRAR軟件解壓。
  • 2. 本站的文檔不包含任何第三方提供的附件圖紙等,如果需要附件,請聯(lián)系上傳者。文件的所有權(quán)益歸上傳用戶所有。
  • 3. 本站RAR壓縮包中若帶圖紙,網(wǎng)頁內(nèi)容里面會有圖紙預(yù)覽,若沒有圖紙預(yù)覽就沒有圖紙。
  • 4. 未經(jīng)權(quán)益所有人同意不得將文件中的內(nèi)容挪作商業(yè)或盈利用途。
  • 5. 眾賞文庫僅提供信息存儲空間,僅對用戶上傳內(nèi)容的表現(xiàn)方式做保護處理,對用戶上傳分享的文檔內(nèi)容本身不做任何修改或編輯,并不能對任何下載內(nèi)容負(fù)責(zé)。
  • 6. 下載文件中如有侵權(quán)或不適當(dāng)內(nèi)容,請與我們聯(lián)系,我們立即糾正。
  • 7. 本站不保證下載資源的準(zhǔn)確性、安全性和完整性, 同時也不承擔(dān)用戶因使用這些下載資源對自己和他人造成任何形式的傷害或損失。

最新文檔

評論

0/150

提交評論