On the Martin paper, focus on the following: - What are the characteristics of the SPECsfs benchmark? What types of real-life workloads does it model? - What are the characteristics of the SPECsfs performance curve? What do they mean? - How do the authors plan to answer the question "How much do improvements in network performance translate into improvements in application performance?" - What is the network model used in this paper? What do the parameters mean? - What is the analytical queueing model of an NFS server used to study performance? - What is the method used in this paper to study sensitivity to different network parameters? - What is the impact of each network parameter? Is there one that is dominant? - What are the general suggestions made by this paper's authors to improve system performance in the future? On the Direct Access File System (DAFS) paper, focus on the following: - What is the networking technology DAFS is based on? - What are the benefits of a user-level file system client? - How does a DAFS client differ from a typical NFS client? - How does the DAFS protocol differ from the NFS protocol? - How could a kernel-based NFS client be enhanced to reduce some of the overheads that DAFS is designed to reduce? - Explain the experimental results shown at the right graphs of Figures 3 and 4. - Explain the experimental results shown in Figure 7.