Operating System Interview Questions and Answers

Answer - 61 : - In an NTFS file system inside the transactions all the data structure updates are performed. Prior to the alteration of a data structure a transaction creates log record containing information on redo and undo functions. Once a transaction is completed commit record information is stored in the logs. An NTFS system recovers from a crash by accessing information from the created log records. The first step is to redo operations of committed transactions and undoing those transactions which could not be successfully committed. Although the NTFS file system after recovering from a crash might not reflect the same user data prior to a crash but it can guarantee the file data structures are undamaged. It restores the structure to a pre-crash and consistent state.

Answer - 62 : - The benefits of placing functionality in the device controller are: - System crasher due to the occurrence of a bug is greatly reduced. - By the utilization of dedicated hardware and algorithms that are hard coded the performance can be improved greatly. - Since the algorithms are hard coded the kernel gets simplified. The banes of placing functionality in the controller rather than the kernel are: - Once a bug occurs they are difficult to fix, a new firmware or revision may be required. - For performance improvement of algorithms hardware upgrades are required rather than a device driver update.

Answer - 63 : - The main advantage of having a system that supports multiple file structures is that the support for it is provided by the system itself no other individual application is required to provide the multiple structure support. Since the support is provided by the system itself the implementation is much more efficient as compared to application level. A demerit of such kind of implementation is that it can increase the overall size of the system. Also, since the support is provided by the system, for an application that requires a different file type may not be executable on such a system. A good alternative for this is that the OS does not define any support for file structures instead all files are considered to be a series of bytes. By doing so the support for file systems is simplified as the OS does not have to specify the different structures for the file systems. It allows the applications to define the file structures. This kind of implementation can be found in UNIX.

Answer - 64 : - The transaction process can be considered to be a series of read and write operations upon some data which is followed by a commit operation. By transaction atomicity it means that if a transaction is not completed successfully then the transaction must be aborted and any changes that the transactions did while execution must be roll backed. It means that a transaction must appear as a single operation that cannot be divided. This ensures that integrity of the data that is being updated is maintained. If the concept of atomicity in transaction is not used any transaction that is aborted midway may result in data to be inconsistent as there might be a possibility two transactions may be sharing the same data value.

Answer - 65 : - When the I/O is frequent and of very short durations polling is considered to be more efficient than an interrupt driven I/O. Although, a serial port individually can have fairly infrequent number of I/O and hence should ideally use interrupts the case of serial ports in a terminal concentrator is different. A terminal concentrator consists of multiple serial ports and this can lead to the creation of multiple short I/O instances this can create un-necessary load on the system in case of interrupts usage. Instead, if a polling loop is used it can greatly reduce the amount of load on the system by looping through without the requirement of I/O. Due to this reason interrupts are used for single ports as the frequency of I/O on such a port is less and can be managed effectively, whereas we use polling for multiple ports as the frequency of I/O increases and are of short durations which suits polling.

Answer - 66 : - 1. It is the ability to continue providing service proportional to level of hardware 2. Systems designed for graceful degradation are called fault tolerant. 3. If we have several processors connected together, then failure of one would not stop the system.4. 5. Then the entire system runs only 10% slower. 6. This leads to increased reliability of the system.

Answer - 67 : - 1. These systems are also called as the distributed systems. 2. It consist of collection of processors that do not share memory or clock. 3. The processors communicate through high speed buses or telephone lines. 4. It can be a centralized system where the server responds to client requests. 5. It can also be a peer to peer system.

Answer - 68 : - 1. It is called as symmetric multiprocessing which is multiprocessor system. 2. In it each processor runs an identical copy of the operating system. 3. These copies communicate with one another as needed. 4. These processor systems lead to increased throughput. 5. These systems are also called parallel systems or tightly coupled systems.

Answer - 69 : - 1. It is the service called as distributed lock manager. 2. In cluster systems to avoid file sharing the distributed systems must provide the access control and file locking. 3. This ensures that no conflicting operations occur in the system. 4. Here the distributed file systems are not general purpose therefore it requires locking.

Answer - 70 : - 1. Handheld devices are palm tops and cellular telephones with connectivity to a network. 2. These devices are of limited size which leads to limited applications. 3. They use a memory 512KB to 16MB as a result the operating system and applications must use the memory efficiently. 4. The speed of the processors is only a fraction of speed of the PC processors and for faster processors larger battery is required. 5. These devices use very small display screens so reading mails and browsing must be condensed to smaller displays.