9 IP route selection
9.1 Introduction Routing is one of the most important features of IP. Figure 9.1 is a simple process for the IP layer processing process. A datagnet that needs to be routed can be generated by the local host or by other hosts. In the latter case, the host must be configured to a router, otherwise the datagram via the network interface, if the destination address is not the local machine, it is to be discarded (for example, being discarded quietly). In Figure 9.1, we also describe a routing daemon (daemon), which is generally a user process. In UNIX systems, most common daemons are routing programs and gateway programs. (The term DAEMON refers to the process running in the background, which performs some operations on behalf of the entire system. Daemon is typically started when the system boots, and there is always a routing protocol that runs on a given host. How to exchange selection information on adjacent routers and how to work, all of which are very complicated, which can be discussed with a whole book. (Interested readers can refer to Document [Perlman 1992] to get more detailed information.) In Chapter 10, we will simply discuss dynamic routing and selection information protocol RIP (Routing Information Protocol). In this chapter, our main purpose is to understand how a single IP layer makes routing decisions. The routing table shown in Figure 9.1 is often accessed by IP (on a busy host, there may be hundreds of times within one second), but it is much lower than that of the routing the counter program (possibly 30 seconds a time). When receiving ICMP, "Redirect" packet, the routing table is also updated, which we will introduce when the route command is discussed in Section 9.5. In this chapter, we will also use the NetStat command to display the routing table.
Figure 9.1 IP layer workflow
9.2 The principle of the route begins to discuss the IP routing, first understand how the kernel maintains the routing table. The information contained in the routing table determines all decisions made by the IP layer. In Section 3.3, we list a few steps of the IP search routing table: 1. Search the matching host address; 2. Search the matching network address; 3. Search the default entry. (The default entry is generally designated as a network entry in the routing table, and its network number is 0.)
Matching the host address steps always happen before matching the network address step. The routing of the IP layer is actually a routing mechanism that searches the routing table and determines which network interface to send groups. This is different from the routing policy, it is just a set of rules that decide which routes put into the routing table. IP executes routing mechanisms, while routing daemons generally provide routing policies.
Simple routing table first let us take a look at some typical host routing tables. On the host SVR4, let's perform the netstat command with the -r parameter lists the routing table, then execute the command again with the -n parameter, print out the IP address in digital format. (We do this because some of the entries in the routing table are network addresses, not host addresses. If there is no -n parameter, the netstat command will search for file / etc / networks and list the network name. This will be with another The form of the form - the network name plus host name confusion.
(See the original book P.113 1)
The first line of description If the destination is 140.252.13.65 (SLIP host), the gateway (router) will forward the packet to 140.252.13.35 (BSDI). This is what we expect because the host SLIP is connected to BSDI via the SLIP link, while BSDI and the host are in the same Ethernet. For a given router, five different flags can be printed: u The route can be used. G The route is to a gateway (router). If this flag is not set, the destination is directly connected. H The route is a host, that is, the destination address is a complete host address. If this flag is not set, the route is to a network, and the destination address is a network address: a network number, or a combination of the network number and the subnet number. D This route is created by the change routing message (9.5). M The route has been changed to the routing message (9.5).
The flag G is very important because it distinguish between indirect routing and direct routing. (For direct route is not seting the flag G.) The difference is that the group of direct routes not only have the IP address indicating the destination, but also has its link layer address (Figure 3.3). When the packet is sent to a indirect route, the IP address is indicated by the final destination, but the link layer address indicates the gateway (ie the next station router). We have seen such an example in Figure 3.4. In a routing table example, we have a indirect route (set the flag G), so the IP address of this route is the final destination (140.252.13.65), but its link layer address must correspond to Router 140.252.13.35. It is important to understand the difference between G and H flags. The G logo distinguishes direct route and indirect routing, as described above. However, the H flag indicates that the destination address (the NetStat command output is the first line) is a complete host address. No H flag shows that the destination address is a network address (the host number portion is 0). When searching for a destination IP address, the host address item must be fully matched with the destination address, and the network address item only needs to match the network number and subnet number of the destination address. In addition, most versions of the NetStat command first print out all host routing tables, and then the network routing entry. Reference count Refcnt ("Reference Count") is given by the number of active processes that are using routing. Connected protocols such as TCPs are fixed when establishing a connection. If a Telnet connection is established between the host SVR4 and SLIP, we can see that the reference numerical value becomes 1. When another Telnet connection is established, its value will increase to 2, and go it. The next column ("USE") is displayed by the number of packets sent by the route. If we are the only packet of this route, then we run a ping program to send 5 packets, its value will become 5. The last list (Interface) is the name of the local interface. The second line output is a loopback interface (Section 2.7), and its name is always LO0. No G flag is set because the route is not a gateway. H Sign Description Destination Address (127.0.0.1) is a host address instead of a network address. Since the G flag is not set, this is a direct route, the gateway column is given an export IP address. The third line output is the default route. Each host has one or more default routes. This indicates that if a specific route is not found in the table, the packet is sent to the router 140.252.13.33 (Sun Host). This shows that the current host (SVR4) uses this routing entry to access other systems via the Internet through the router Sun (and its SLIP link). Establishing default routes is a strong concept. The route flag (UG) indicates that it is a gateway, which is what we expect. (The following is a translation of the original book P.1141) Here, we have a destination name of SUN as a router instead of a host because it is used as a default router to use, it is an IP forwarding function, but a host function. The Host Requirements RFC document specifically stated that the IP layer must support multiple default routes. However, many implementations do not support this. When there are multiple default routes, the general technology becomes a bird around them, for example, Solaris 2.2 is doing this.
The last line in the output is the Ethernet of the Ethernet. The H flag is not set, indicating that the destination address (140.252.13.32) is a network address, and its host address portion is set to 0. In fact, it is 5 (Figure 3.11). Since this is a direct route (G flag is not set), the IP address pointed out by the gateway column is an exit address. The last item output by the netstat command has another information, which is the subnet mask of the destination address (140.252.13.32). If you want to compare the destination address with 140.252.13.33, then you must first logic with the destination address mask (0xffffffe0, 3.7) before comparing. Since the kernel knows the interface corresponding to each routing entry, and each interface has a corresponding subnet mask, each routing entry has an implicit subnet mask. The complexity of the host routing table depends on the topology of the network where the host is located. 1. The simplest (and most interested) is that the host is not connected to any network. The TCP / IP protocol can still be used for such a host, but can only communicate with themselves! The routing table in this case only includes one ring back interface. 2. The next situation is that the host can only access the host on the local area on a local area. At this time, the routing table contains two items: one is a loopback interface, and the other is a local area network (such as Ethernet). 3. If the host can access other networks (such as the Internet) via a single router, then the next step is performed. Under normal circumstances, add a default entry to the router. 4. If you want to add other specific hosts or network routes, then the last step is performed. In our example, the routing of the host SLIP is to pass the router BSDI is an example. We use this routing table as some examples of the host SVR4 to select route to the steps of the above IP operations. 1. Assume the destination address is the host Sun, 140.252.13.33. First, the host address is matched. The two host address entries in the routing table do not match, and then the network address matches. This time the match is successful, finding the entry 140.252.13.32 (the network number and subnet number), so the EMD0 interface is used. This is a direct route, so the link layer address will be the address of the destination. 2. Assume the destination address is the host SLIP, 140.252.13.65. First search the host address in the routing table and find a match address. This is an indirect route, so the IP address of the destination is still 140.252.13.65, but the link layer address must be the link layer address of the gateway 140.252.13.65, the interface name is EMD0.3. This time we pass Internet to host AW .com (192.207.117.2) Send a datagram. First search the host address in the routing table, and the network address match is performed. Finally, the default entry is successfully found. The route is a indirect route, through the gateway 140.252.13.33 and uses the interface name EMD0.4. In our last example, we send a datam. There are four ways to complete this, such as host name, host IP address, ring back, or loopback IP address: FTP SVR4FTP 140.252.13.34
FTP localhostftp 127.0.0.1
In the first two cases, a second search of the routing table gets a matching network address 140.252.13.32, and transmits the IP packet to the Ethernet driver. As shown in Figure 2.4, the destination address in the IP packet is the native IP address, so the packet is given to the loopback driver, and then put the message into the IP output queue by the driver. In the latter case, since the first search finds the matching host address due to the designation of the name or IP address of the loopback interface, the packet is directly sent to the loopback driver, and then put the message by the driver Enter the IP output queue. The above four cases are sent to the loopback driver, but the two routing decisions employed are different. Initialization Routing Table We never said how these routing tables were created. When an interface is initialized (usually using the ifconfig command), a direct route is automatically created for the interface. For point-to-point links and loopback interfaces, the route is to reach the host (for example, setting H flag). For broadcast interfaces, such as Ethernet, route is to reach the network. If the route to the host or network is not directly connected, the routing table must be added. An ordinary approach explicitly runs the route command during the initialization file when the system is guided. On the host SVR4, we run the following two commands to add entries in the routing table:
Route Add Slip BSDI 1
The third parameter (Default and SLIP) represents the destination, the fourth parameter represents the gateway (router), and the last parameter represents the metric of the route. The route command is greater than 0 when the metric is greater than 0, otherwise the G flag is not set when the value is 0.
(The following is a translation of the original book P.1161) Unfortunately, there is almost no system to include the route command in the startup file. In the 4.4 BSD and BSD / 386 system, the startup file is / etc / netstart, in the SVR4 system, the startup file is /etc/inet/rc.inet, in Solaris 2.x, the startup file is / etc / rc2. D / S69inet, in SunOS 4.1.x, the startup file is /etc/rc.local, while AIX 3.2.2 uses file /etc/rc.net. Some systems allow a default router to be specified in a file, such as / etc / default, so add the default item in the routing table when the system is restarted. Other methods of initializing the routing table are to run the route daemon (Chapter 10) or with a newer router discovery protocol (9.6).
More complex routing tables On our subnet, host Sun is the default router for all hosts because it has a Coiled SLIP link to connect to the Internet (see Figure 2).
(See the original book P.117 1)
The first two are consistent with the first two of the host SVR4: the router BSDI reaches the specific host route of the SLIP, and the loopback route. The third line is new. This is a route directly reaches the host (no g signs, but set H flag), corresponds to the link of point-to-point points, the SLIP interface. If we compare it to the output of the IFConfig command:
Sun% ifconfig SL0SL0: Flags = 1051 <- Up, Pointopoint, Running> INET 140.252.1.29 -> 140.252.1.183 Netmask fffff00
We found that the destination address in the routing table is the other end of the point-to-point link (ie router Netb), the gateway address is the local export IP address (140.252.1.29). (We have already said before, NetStat is a gateway address that prints directly from the local interface is the IP address used by the local interface.) The default routing table is an indirect route to the network (set the G flag, but no H logo), This is exactly what we hope. The gateway address is the address of the router (140.252.1.183, the other end of the SLIP link), not the local IP address of the SLIP link (140.252.1.29). The reason is or because it is indirect route, not directly routed. We should also point out that the third and fourth lines of NetStat output (interface names SL0) are created by the SLIP software at startup and delete it when it is closed. Route all the examples of the destination is assumed to be routed The table's search can find the matching entry, even if the default item is matched. If there is no default item in the routing table, there is no match, what happens when it is? The result depends on whether the IP datagram is generated by the host (for example, we act as a router). If the duplicate is generated by the local host, then a error is sent to the application that the datagram is sent, or the "host is not a bad error" or "network is unreachable". If it is a forwarded datagram, then send a message that ICMP hosts that are unremnous with errors will be sent to the original sender. In the next section we will discuss this error.
9.3 ICMP host and network unparalleled error When the router receives an IP datagram, it is necessary to send an ICMP "host that cannot be reached" error messages. (The format of the ICMP host is not a message is shown in Figure 6.10). We can easily discover that in our network, the Coiled SLIP link on the router Sun is disconnected, and then tried to send a packet to any host for the default router via the SLIP link.
(Below is the original book P.1181 translation) The older version of the BSD produces a host that is unachable or the network is unreachable, depending on whether the destination is in a local subnet. 4.4 BSD only generates a host to not reach an error.
We can see in the previous section by running the netstat command on the router Sun, and when the SLIP link is turned on, add an entries that use the SLIP link in the routing table, while disconnecting the SLIP link. Then delete the table. This shows that when the SLIP link is disconnected, there is no default item in the routing table of the Sun. But we don't want to change the routing tables of other hosts on the network, but also delete their default routes. Instead, we count the ICMP hosts it produce the ICMP host to count the ICMP host. You can see this on the host SVR4, it can see this, it is on the other end of the SLIP link (the coament link has been disconnected):
(See the original book P.118 2)
The output of running the tcpdump command on the host BSDI is shown in Figure 9.2.
Figure 9.2 Response ping command ICMP host is not reaching a message
When the router Sun finds that when the router that can reach the host Gemini, it responds to an unreachable reconciliation request message. If we pick up the SLIP link to the Internet, then try ping a IP address that is not connected to the Internet, you should generate an error. But is interested in that we can see how far is you want to transfer in the Internet before returning an error message:
(See the original book P.118 3)
As can be seen from Figure 8.5, the packet has passed 6 routers prior to discovering that the IP address is invalid. The error is only detected when it reaches the boundary of the NSFNET backbone network. This shows that the reason why 6 routers can forward the packet because there is a default item in the routing table, and only when the packet arrives at the NSFNET backbone network, the router can know the information of each network connected to the Internet. This shows that many routers can only work within the local area. Reference [Ford, Rekhter, And Braun 1993] defines top-level routing domain, which maintained routing information of most Internet sites without using the default route. They pointed out that there are 5 such top-level selection fields on the Internet: NSFNET main network, Commercial Internet Exchange: Cix, NASA Science Internet: NSA Science Internet: NASA Science Internet: NSA Science Internet: NSA Science Internet: NSA Science Internet: Sprintlink, and European IP main network ( Ebone). 9.4 Forwarding or not forwarding We have already mentioned several times, generally assume that the host does not forward IP datagrams unless they are used as a router unless they are specially configured. How to make such a configuration? Most Berkeley derived systems have a kernel variable ipforwarding, or other similar names. (See Appendix E.) Some systems (such as BSD / 386 and SVR4) are only forwarded in the case where the variable value is not 0. Sunos 4.1.x allows the variable to be three different values: -1 means that it is always not forwarded and does not change its value; 0 indicates that the default condition is not forwarded, but it will be turned on when two or more interfaces are opened. The value is set to 1; 1 means always forwarded. Solaris 2.x change the three values to 0 (never forwarded), 1 (always forwarded) and 2 (forwarded when two or more interfaces). Older versions of 4.2BSD hosts can forward datots under default conditions, which brings many issues that have not been properly configured. This is why the kernel options are set to the default "never forwarded", unless the system administrator is special setting.
9.5 ICMP Change Routing error When IP datagram should be sent to another router, the router that receives the datagram will send an ICMP change route (redirect) error message to the IP datagram. This is conceptually very simple, just as we shown in Figure 9.3. Only when the host can select a router to send a packet, we may see the ICMP change routing message. (Memory we have seen in Figure 7.6.) 1. We assume that the host sends a IP datagram to R1. This route selection decision often occurs because R1 is the default route of the host. 2. R1 Receives the datagram and checks its routing table, and finding that R2 is the next stop for sending the datagram. When it sends the datagram to R2, R1 detects that the interface that it is sending is the same (ie, the host and the two routers are located). This gives the router to the router to provide a clue to the original sender. 3. R1 Send an ICMP change routing message to the host, telling it to send the datagram to R2 instead of R1.
Figure 9.3 ICMP change routing example
Changing the route is generally used to gradually establish a more complete routing table with rare routing information. The host startup time routing table can have only one default entry (in the example of Figure 9.3, R1 or R2), once the default routing error occurs, the default router will notify it to change the route, allow the host to the routing table . ICMP changes routing allows TCP / IP hosts that do not need intelligence characteristics when performing routing, and put all intelligent features in the router side. Obviously, in our example, R1 and R2 must know information about the more topologies of the connected network, but even the host on the LAN is started with a default route, by receiving the change routing packets to gradually study. . An example we can observe the operation of the ICMP change route on our network (see a picture of the seal). Although we only draw three hosts (AIX, Solaris and Gemini) and two routers (Gateway and Netb) in the topology map, the entire network has more than 150 hosts and 10 additional routers. Most hosts specify GATEWAY as the default router because it provides an inlet of the Internet. How does the host on the subnet 140.252.1 accesses the author's subnet (four hosts under the figure)? First, if there is only one host at one end of the SLIP link, then use the agent ARP (Section 4.6). This means that the host Sun (140.252.1.29) can be accessed by the host in the subnet (140.252.1) on top of the topology. The agent ARP software on NetB handles these things. However, when the network is on the other end of the SLIP link, it is necessary to relate to routing. One way is to let all hosts and routers know that the router NetB is a gateway of network 140.252.13. This can set static routes in the routing table of each host, or run a daemon on each host. Another simpler method (also actually adopted) is to achieve the use of ICMP to change routing packets. Let's run the PING program to host BSDI (140.252.13.35) on the host Solaris on the top of the network. Since the subnet number is different, the agent ARP cannot be used. Assume that no static route is installed, the first packet sent will use the default route to the router Gateway. Below is the routing table before we run the PING program: (see the original book P.121 1)
(The entry of 224.0.0.0 is the IP broadcast address. We will discuss this chapter 12.) If you specify the -v option for the PING program, we can see any ICMP packets received by the host. We need to specify this option to observe the change route packets sent.
(See the original book P.121 2)
Before we receive the first response from the PING program, the host first received an ICMP change routing message from the default router Gateway. If we view the routing table at this time, it will find a new route that has been inserted into the host BSDI. (The item is shown in black body.)
(See the original book P.121 3)
This is our first time I saw the D flag, indicating that the route is created by the ICMP routing message. G Sign Description This is an indirect route reaching Gateway (NetB). The H flag shows that this is a host route (as we expect), not a network route. Since this is a host route that is incremented by the host, it is only processed to the host BSDI. If we go to the host SVR4, we must generate another ICMP change routing message to create another host route. Similarly, accessing host SLIP also creates another host route. The three hosts (BSDI, SVR4, and SLIP) on the subnet can also be processed by a network route to the router Sun. However, ICMP changes routing packets create host routes, not network routing, because in this example, the router generating ICMP change routing packets does not know subnet information on the 140.252.13 network. More details ICMP changes the format of the routing message as shown in Figure 9.4.
Figure 9.4 ICMP change routing message
There are four different types of change routing messages, there are different code values, as shown in Figure 9.5.
Figure 9.5 ICMP varies in different code values for routing packets
ICMP changes the recipient of the routing message must view three IP addresses: (1) causes the IP address of the change to change the route (ie, the data of the ICMP change routing message is in the IP datagram); (2) Send a change routing message The IP address of the router (including the source address in IP datagram in change routing; (3) The router IP address (4-7 bytes in ICMP packet) should be used. About ICMP change routing packets have many rules First, changing the routing packets can only be generated by the router without being generated by the host. In addition, changing the routing message is used for the host instead of router. Assume the router and other routers to participate in a routing protocol, The protocol can eliminate the need to change the route. (This means that the routing table in Figure 9.1 should be eliminated or can be modified by the routing daemon, or can be changed, but cannot be modified by both. In the 4.4BSD system, when the host is used as the router, the following checks are performed, before generating the ICMP change routing message, these conditions must be met. 1. The outgoing interface must be equal to the entry. 2. Used to transmit data The routing of the newspaper cannot be created or modified by the ICMP to change the routing message, and cannot be the default route of the router. 3. Data report cannot be converted to the source station to forward. 4. The kernel must be configured to send change routing messages.
(The following is the original book P.1231) The kernel variable is named IP_sendredirects or other similar names. (See Appendix E.) Most current systems (such as BSD, SunOS 4.1.x, Solaris 2.x, and AIX 3.2.2) set this variable under default conditions to make the system can send change routing messages. Other systems, such as SVR4, turn off the feature.
In addition, after a 4.4BSD host receives ICMP to change the routing message, it is necessary to make some checks before modifying the routing table. This is to prevent malicious operation of the router or host, as well as malicious users, resulting in errors to modify the system routing table. 1. The new router must be connected directly to the network. 2. Change the routing message must come from the router that is currently selected to the destination. 3. Changing the routing message cannot make the host itself as a router. 4. The modified routing must be a indirect route.
About changing the routing is that the router should be sent only the change route to the host (code 1 or 3, as shown in Figure 9.5), rather than a change route to the network. The existence of subnets makes it difficult to accurately indicate when to send a change route to the network rather than changing the host. Only when the router sends an error type, some hosts proceeds to the change route of the network as a change route to the host.
9.6 ICMP Router Discovery Packet We have mentioned a method of initializing the routing table in front of this chapter, that is, specifying a static route in the configuration file. This method is often used to set the default route. Another new method is to use ICMP router notices and request packets. It is generally believed that the host should broadcast or multicast to send a router request message after booting. One or more routers respond to a router advertisement message. In addition, the router is periodically broadcast or multicast to deliver their router advertisement packets, allowing each of the hosts that are being listening to the corresponding updates their routing tables. RFC 1256 [DeERING 1991] determines the format of these two ICMP packets. The format of the ICMP router request message is shown in Figure 9.6. The format of the ICMP router advertisement message is shown in Figure 9.7. A plurality of addresses can be advertised in a message. The number of addresses refers to the number of addresses contained in the packet. The address size refers to the number of each router address 32-bit word, which is always 2. The life cycle refers to the time (second number of seconds) of the advertisement address.
Figure 9.6 ICMP Router Request Packet Format
Figure 9.7 ICMP Router General Packet Format
Next is a pair or more IP address and priority. The IP address must be an address that sends a router. The priority is a symbolic 32-bit integer indicating that the IP address is a priority of the default router address, which is compared to other routers on the subnet. The greater the value, the higher the priority. The priority is 0x80000000 Description The corresponding address cannot be used as the default router address, although it also includes in the announcement message. The default value of the priority is generally 0.
Router operation When the router is started, it regularly sends a notice message on all broadcast or multicast transmission interfaces. Accurately, these notice messages are not regularly sent, but randomly transmitted to reduce probability of conflicting with other routers on the subnet. Generally, each two notices are 450 and 600 seconds. A given advertisement message The default lifecycle is 30 minutes. Another timing using the life cycle field is when an interface on the router is turned off. In this case, the router can send the last advertisement message on the interface and set the lifecycle value to 0. In addition to regularly sending an active advertisement message, the router also listens to the request packet from the host and sends the router notice message to respond to these request messages. If there are multiple routers on the subnet, the system administrator sets priority to each router. For example, the primary default router has a higher priority than the backup router.
The host operator hosts typically send three routers request packets during boot, send once every three seconds. Once a valid advertisement message is received, the send request message is stopped. The host also monitors the request packet from the adjacent router. These advertisement messages can change the default router of the host. In addition, if no advertisement packets from the current default router is received, the default router will time out. As long as there is a general default router, the router will send a notice message every 10 minutes, and the life cycle of the message is 30 minutes. This shows that the host's default entry is not timeout, even if you miss one or two announce packets.
Implementing router discovery packets are typically created and processed by a user process (daemon). This way, there is another program that modifies the routing table in Figure 9.1, although it only increases or deletes the default table item. The daemon must be configured as a router or host.
(The following is the original book P.1251) These two ICMP packets are new, not all systems support them. In our network, only Solaris 2.x supports these two packets (IN.RDISC daemon). Although RFC is recommended for transmitting as much as IP multicast, router finding can also be implemented using broadcast messages.
9.7 Small junction IP routing is the most basic for systems running TCP / IP, whether it is a host or a router. The content of the routing entry is very simple, it includes: 5 Bit flag, destination IP address (host, network, or default), the IP address of the next stop router (indirect routing) or the IP address of the local interface (direct route), pointing Pointer for local interfaces. The host entry has a higher priority than the network entry, and the network entry has a higher priority than the default item. Each IP datagram generated by the system is to search for routing tables, which can be modified by routing daemons or ICMP change routing messages. The system is not forwarded by default unless special configuration is performed. Use the route command to enter the static route, you can use the new ICMP router to discover packets to initialize the default entry and perform dynamic modification. The host only has a simple routing table at startup, which can be dynamically modified by ICMP change routing packets from the default router. In this chapter, we focus on how a single system uses a routing table. In the next chapter, we will discuss how the routers exchange routing information. Exercise 9.1 Why do you think there are two types of ICMP change routing packets - network and host? 9.2 Is the routing table listed on the SVR4 host in 9.4, is it necessary to go to the host SLIP (140.252.13.65)? What changes will there be change this item from the routing table? 9.3 Consider having a cable connection 4.2BSD host and 4.3bsd host. It is assumed that the network number is a 140.1.4.2bsd host to bring the host number of the address recognition (140.1.0.0). In addition, the 4.2BSD host is trying to forward the received datagram under the default conditions, although only one interface is only. Please describe what happens when the 4.2BSD host receives a destination address of 140.1.255.255. 9.4 Continue the previous exercise, assuming that someone adds an item in a system ARP cache on the subnet 140.1, specifying the Ethernet address corresponding to the IP address 140.1.255.255 1 (Ethernet Broadcast address). Please describe the situation that occurs at this time. 9.5 Check the routing table on the system you use and explain each content.
9-6
