Explain what is meant by the load sharing components: transfer policy, selection policy, location policy, and information policy for dynamic load sharing. Give some reasons why new benchmark suites are continuously released. How does the adaptive optimization mechanism improve the overall Java code performance? What are the 3 meanings conveyed by term 'Java Virtual Machine'? Explain the difference between the blocking and nonblocking send in a client-server model. What is the basic problem that the SubOS attempts to solve and what solution does it use? Compare and contrast the following event-handling mechanisms: select(), poll() and Real Time signals. How did the paper by Chandra in USENIX01 enhance the Real Time signals for better performance? In the paper on Time Sensitive Linux, what are the three types of latency for handling events? What technique was used to reduce timer latency? Explain in brief the three classes of solutions proposed to improve TCP performance over Wireless links. What are the main advantages and disadvantages of a preemptible kernel versus a non-preemptible kernel? Explain the basic architecture of a hyper-threading processor and its purported merits. When scheduling processes in a symmetric multiprocessor (SMP) what are the tradeoffs between considering idle processors and cached data? Under which conditions should each be given priority? Explain the differences between "at most once", "at least once", and "only once" policies when handling communication failures in RPC systems. What are some of the difficulties in enforcing each of these policies? What are some advantages of user level threads over kernel level threads? What are some advantages of kernel level threads over user level threads? Explain why device drivers in the Plan9 operating system do not need the ioctl() command. What are some advantages and disadvantages of this mechanism? Describe an obstacle associated with using an Ethernet network in a distributed real-time application. List three differences between NFS and AFS. Describe some of the basic problems of event synchronization in a distributed system and describe an algorithm proposed by Leslie Lamport's (in Time, Clocks, and the ordering of events) to overcome these problems. Describe basic Manual (e.g. C-Runtime) and Automatic Memory Management (i.e. garbage collection) and describe the relative advantages and disadvantages of each. Describe some reasons why an application may want to use threads instead of spawning a whole new process. What causes application level thread synchronization issues. Describe how the Windows 2000 Kernel provides for application level thread synchronization and describe three different synch. Objects it exposes. Explain some of the requirements for scalability of an Internet-scale message transport system. Compare NUMA and UMA architectures, and explain the advantages of each. Explain the "two armies" problem, and explain why it can't be fully solved. Data is distributed in web server farms in many ways. Name two ways that the data can be distributed for WEB CONTENT, along with an advantage and disadvantage of each technique. What is the difference between a write through cache, and a write back cache? What is priority inversion, and why can it be a problem in process scheduling? Explain the process states that a Linux system has? What are the five classes of failures in RPC systems? Explain the difference between FTP and NFS? Describe in brief what is a HTTP persistent connection. What benefit does does it give over the previous protocol implementation? Outline the processing required to perform an RPC from an application program running on one computer to a server running on a different computer. A grid implements two level scheduler - first is a super scheduler who is a centralized scheduler and the second is the local scheduler on each node. On receipt of jobs the super scheduler looks at the resources available on different nodes within the grid and the also finds which node can give quickest result for a particular job, and then schedule jobs on different nodes on a grid. Describe a problem you see in this type of scheduling? If you were designing this what would you keep in mind to solve this problem? Describe the general process of web load balancing. Name two issues in distributed shared memory and describe a proposed solution. Explain marshaling and its purpose. What is the Byzantine Agreement Problem? What is the significance of this? What is a operating system process, and operating system thread? What information is part of a process, as well as what information is part of a thread? In what way are these two parts of the operating system tied together? How is SMP scheduling carried out in Win2k and Linux? and explain how "Process affinity" plays a role in SMP scheduling Explain how Distributed Hash table or Clustered Hash table can be used as Data management layer in Internet Services environment to address issues like High-availability, Scalability, Data consistency and other characteristics of Internet Services environment? Compare two methods of maintaining cache coherency in a distributed system. Describe the value of leasing resources in a distributed environment. What is meant by the term of affinity in processor scheduling? What are the possible semantics for RPC in the face of failures? In the context of distributed shared memory, what is false sharing and why is it relevant? What is an election algorithm? Describe how one might be implemented. How does symbiotic job scheduling improve performance on multi-threaded systems? What are the tradeoffs between static and dynamic allocation in multiprocessor schedulers? How does kernel preemption affect real-time responsiveness? Classify different types of processes running in a system? Delineate the issues involved in Soft real time scheduling? If you were to design your own scheduler for your desktop, please provide an algorithm? Define fairness in scheduling? Discuss three different models for building network servers, and the potential advantages and disadvantages of each. Explain how a virtual cache alias can be created, and why it can cause problems. Name three characteristics of files as identified in the AFS work and how these characteristics influenced the AFS design. How are file permissions handled in the Ivy Peer-to-Peer File System? What is termination detection? Describe how it might be implemented. Of the Adaptive Load Sharing Policies, why should Threshold be implemented rather than the Shortest Policy? How do cache replacement policies in the Web compare to page replacement policies? What is the definition of a micro-benchmark? What are the advantages and disadvantages of micro-benchmark? How is pipe implemented in Linux IPC? Describe the features of Microkernel and Monolithic O.S. and their advantages and disadvantages. What O.S. fall into these 2 categories as you know? There are usually 2 major ways of handling requests in nowadays web servers: thread/process-based and event-based. Describe how they work, and why the thread/process-based mechanism is more expensive than the event-based one? Describe briefly five salient feature a distributed file system should have. Name three popular distributed file systems. Describe briefly three services a distributed file systems offers. What makes synchronizing a PDA with a PC different (harder) than synchronizing two PCs and what is done to get around these limitations? Explain access and location independence and how they relate to distributed filesystems.