 వానినివ౎సినిగారాత్లాస్త్లుదిత్తాా. నింపోని నిసికి నిగాకాబా. మతోక౎త్రి నిరవంద్న యమ్పం చరి. We are going to see the definition of buffering. Buffering means the storage device. In the center and receiver we are using some buffering mechanisms. In distributed communication, messengers are transferred from one process to another. the receiver end. Following types of buffering mechanisms are used in distributed systems types. First in null buffer, second in unbounded buffer, third in single business buffer, fourth in finite bounded buffer. In message-based inter-process communication IPC, the last two cases are extremely used. No buffer. Between senders and receiver there will not be any buffer that is called the no buffer. Here it will take only one copy operation from senders address space to receiver address space. So no space to store the message temporarily. Following strategies may be used. The messages are remains in senders process address space and the execution of said is delayed until receiver executes the corresponding receive. The message is simply discarded and time out period for the acknowledgement is said to recent message having assumed that receiver has already executed receive. This advantage sender may have to try several times and may ultimately give up. So this is the figure of the null buffer where messages are copied from sender process to receiver process. Single message buffer. In synchronous communication mechanism are used single message buffer having the capacity store single message used to receiver the receiver's note. So in this this is the figure of single message buffer in which sending address space sends a message. This will be copied into the single message buffer. It will be capacity of copying only one message. Then the message will be copied from single message buffer to receiving process address space. So message transfer inverse to copy operations. Unbounded capacity buffer. In asynchronous communication a sender does not wait for receive acknowledgement. There may be several pending messages yet to be received by receiver. Hence unbounded capacity buffer is used to accommodate all messages of sender. Here the capacity of the buffer is unlimited. Infinite messages stored in the unbounded capacity buffer. Finite bounded buffer. Unbounded capacity buffer is practically impossible. So hence asynchronous communication use finite bounded buffers. It involves the strategy for handling buffer overflow which is dealt with one of the two ways. Unsuccessful communication and flow controlled communication. Unsuccessful communication means message transfer fails if no buffer space. The send return and error message to the sending process hence message passing is unreliable. Flow controlled communication. Here sender blocked until receiver accepts some messages thus creating space in buffer not only for asynchronous for all send primitives. This is the diagram of finite bounded buffer where sending process the another end is receiving process the messages copied from sending process to the message buffer where number of messages are stored and it may copied from message buffer to the receiving process buffer creation. So create buffer in system call is provided to the users. Receiver on executing this call creates buffer or mailbox of a size specified receiver. The receiver mailbox may be located in kernel or receiver's address space. This is the figure of the buffer creation where sending process and another is receiving process in between the there will be creation of buffer. In asynchronous communication message transfer with bounded buffer if it involves two copy operations better than no buffer and single buffer but overhead involved in buffer management. Multi datagram messages. MTU means message transfer in it of a network. A message whose size is greater than MTU is fragmented into multiples of MTU and each fragment is sent into packet or datagram that has control information addition to the message data. Hence multi datagram messages are sent. Reassembling of datagram on receiver side is the responsibility of the message passing system. Each packet is known as a datagram. Messages smaller than the MTU of the network can be sent into single packet or known as single datagram messages. Messages are larger than MTU of the network have to be fragmented into and sent into multiple packets. Such messages are known as multi datagram messages. Different packets of multi datagram messages bear a sequential relationship to one another. So we can pass the message from sender to receiver by using this different multi datagram messages. Pause the video and write the answer. In no buffering type of message transfer dash copy operations are required. Options are 1, 2, 3 or 4. So choose any one option among these four options. The answer is A1. In no buffering type of message transfer one copy operations are required. Stub generation. Stubs can be generated in two ways. Manually and automatically. Manually. RPC implements provides a set of translation functions from which a user can construct his or her own stub. Simple to implement and can handle every complex parameter types. Second one is automatically. It uses interface definition language that is used to define the interface between client and server. And ideal uses a list of procedures supported by the interface together with the type of arguments and results. Client and server independently perform compile time type checking to generate appropriate calling sequences. And interface definition also contains information that helps RPC reduce the amount of data transfer on the network. For example interface definition has information to indicate whether each arguments is input, output or both. Only input arguments are needed to be copied from client to server. Only output arguments are needed to copied from server to client. A server program that implements procedure is an interface, is said to export the interface. A client program that calls the procedure from an interface is said to import the interface when writing a distributed application. A programmer first writes an interface definition using interface definition language that is ideal. Then a client program and server programs are written. Ideal means interface definition language is processed using ideal compiler to generate client and server steps for each procedure called. These steps have appropriate marshalling and unmarshalling operations and a header file that supports the different types of interface definition. The header file also contains or the header file also included in the source files of both the client and server programs. These steps are compiled and linked with each with client and server programs. Next bit is group communication. So here the sender will be sending the messages to the receiver end. So many communication means one is sender and many are the receiver. Many to one means many are the sender, one is the receiver. Many to many means many are the senders and many are the receivers. Now we are going to see each one by one. Multi-cast. A very useful for several practical applications. For example a server manager sending a message to group of servers providing same type of service. Special case broadcast for example enquiry message broadcast for a specific type of service. Here best example of this is broadcasting. One messages can be sent to different receivers. So sender can send the message to the multiple receivers. Many to one. Here multiple sender and one receiver. The single receiver may be selective or non-selective. A selective receiver specifies the unique sender that is receiver will accept message of only that sender. A non-selective receiver specifies set of sender and if any one sender sends a message to that a message exchange takes place. Disadvantage. Non-deterministic. That is unable to know whose message will be accepted. Notation is needed to express and control non-determinism which is related to programming. Many to many. Here multiple are the sender and multiple are the receivers. Many to many and many to one implicit in this scheme. An important issue related to many to many is that of ordered message delivery. Ordered message delivery requires message passing sequencing. Next is group management. There are two types of group management. One is closed group management and another is open group management. So these are the two types of groups open and closed. Closed group only members of the group can send message to the group. For example a group of process working on a common problem. Open group. Outside member can send a message to the group. For example a group of replicated servers meet for distributed processing of client requests. These are the references. Thank you.