Milind M. Buddhikot

Design of a large scale multimedia storage server

Abstract Large scale multimedia storage servers will be an integral part of the emerging distributed multimedia computing infrastructure. However, given the modest rate of improvements in storage transfer rates, designing servers that meet the demands of multimedia applications is a challenging task that needs significant architectural innovation. Our research project, called Massively-parallel And Real-time Storage ( mars ) architecture, is aimed at the design and prototype implementation of a large scale multimedia storage server. It uses some of the well-known techniques in parallel I/O, such as data striping and Redundant Arrays of Inexpensive Disks ( raid ) and an innovative atm based interconnect inside the server to achieve a scalable architecture that transparently connects storage devices to an atm -based broadband network. The atm interconnect within the server uses a custom asic called ATM Port Interconnect Controller ( apic ) currently being developed as a part of an arpa sponsored gigabit local atm testbed. Our architecture relies on innovative data striping and real-time scheduling to allow a large number of guaranteed concurrent accesses, and uses separation of metadata from real data to achieve a direct flow of the media streams between the storage devices and the network. This paper presents our system architecture; one that is scalable in terms of the number of supported users and the throughput.

Efficient data layout, scheduling and playout control in MARS

Abstract. Large-scale on-demand multimedia servers that can provide independent and interactive access to a vast amount of multimedia information to a large number of concurrent clients will be required for a widespread deployment of exciting multimedia applications. Our project, called Massively-parallel And Real-time Storage (MARS) is aimed at prototype development of such a large-scale server. This paper primarily focuses on the distributed data layout and scheduling techniques developed as a part of this project. These techniques support a high degree of parallelism and concurrency, and efficiently implement various playout control operations, such as fast forward, rewind, pause, resume, frame advance and random access.

Enhancements to 4.4 BSD UNIX for efficient networked multimedia in project MARS

Cluster based architectures that employ inexpensive personal computers (PCs) interconnected by high speed commodity interconnect have been recognized as a cost effective way of building high performance scalable multimedia-on-demand (MOD) storage servers (W. Bolosky et al., 1996; M. Buddhikot et al., 1994). Typically, the PCs in these architectures run operating systems such as UNIX that have traditionally been optimized for interactive computing and lack fast disk-to-network data paths and support for guaranteed CPU and storage access. We report design, implementation and performance measurements of innovative enhancements to 4.4 BSD UNIX carried out to rectify these limitations in the context of our Massively-parallel And Real-time Storage (MARS) project (M. Buddhikot et al., 1994). We have proposed and implemented the following enhancements to a 4.4 BSD compliant public domain NetBSD UNIX operating system: (1) a new kernel buffer management system called Multimedia M-buf (mmbuf) which shortens the data path from a storage device to network interface; (2) fair queueing within the SCSI driver for equitable resource sharing between real time and non real time streams; and (3) integration of these new OS services with a CPU scheduling mechanism called Real Time Upcall (R. Gopal, 1996) and a software disk striping driver called Concatenated Disk (CCD). Our experimental results demonstrate that these enhancements provide throughput improvements and QOS guarantees on the data path from the disk to network.

Design of a Large Scale Multimedia Server

Large scale multimedia storage servers will be an integral part of the emerging distributed multimedia computing infrastructure. However, given the modest rate of improvements in storage transfer rates, designing servers that meet the demands of multimedia applications is a challenging task that needs significant architectural innovation. Our research project, called Massively-parallel And Real-time Storage (MARS) architecture, is aimed at the design and prototype implementation of a large scale multimedia storage server. It uses some of the wellknown techniques in parallel I/O, such as data striping and Redundant Arrays of Inexpensive Disks (RAID) and an innovative ATM based interconnect inside... Read complete abstract on page 2.

Simulation of an ATM-FDDI gateway

A detailed simulation model for the design of a high performance asynchronous transfer mode - fiber distributed data interface (ATM-FDDI) gateway based on a new connection-oriented internetwork abstraction called very high speed internetwork (VHSI) is presented. The hierarchical simulation model, constructed using a graphical simulation tool called BONES, serves the dual purpose of functional verification and performance evaluation. The conclusions of this simulation study are presented.

Load Balance Properties of Distributed Data Layouts for Clustered MOD Servers

Large scale storage servers that provide location transparent, interactive access to hundreds or thousands of concurrent, independent clients will be important components of hte furture information super-highway infrastructure. Two key requirements of such servers are as follows: support high parallelism and concurrency in data access to allow large number of access to the same or different data. Second, support independent interactive playout control operations such as fast-forward, rewind, slow-play, pause, resume, random access etc. with minimal latency. This paper assumes a distributed storage server architecture consisting of several high performance storage nodes interconnected by a high speed desk area... Read complete abstract on page 2.