A stream-based programming model for disklets (i.e., disk-resident code) is presented in [8, 12]. Their active disks are attached to the local bus of the host processor. The disklet programming model is similar to ours. They justify their design decisions by providing a detailed simulation of active disks. In contrast, our focus is on building an actual system which introduces practical restrictions.

In [12, 13], an evaluation of the disklet model is provided against two alternative architectures: shared memory multiprocessors (SMPs) and workstation clusters. For most of the applications tested, active disks and clusters significantly outperformed the SMP architecture.

The IDISKs (Intelligent disks) architecture proposed in [9] is based on replacing the nodes in a shared-nothing cluster server with intelligent disks that is disks capable of local processing. The main difference in the IDISKs architecture is that the disks are directly connected with each other via switches thus exhibiting much higher bandwidth disk-to-disk communication.

The architecture closest to SmAS is the active disks of [10]. The authors of [10] concentrate on developing a number of applications to validate the active disks approach. Their analytical and experimental results promise linear speed-ups in disk arrays of hundreds of active disks for certain data-intensive applications. Instead, the alternative of directly attaching a number of traditional SCSI disks to the local bus of a single server machine caused the server CPU or the interconnect bandwidth to saturate even when a small number of disks (less that ten) was attached.

Besides research on active disk, there is some recent interest in shipping application code to the data sources [15]. Application code is in the form of bytecodes. Using bytecodes is not a feasible approach in our case since the limited resources of the disk do not allow for a java execution machine.

[1] ANSI, "Information systems - Small Computer System Interface-2 (SCSI-2)", "ANSI X3.131-1994", 1994

[2] R. Winter and K. Auerbach, "The big time: the 1998 VLDB survey", "Database Programming and design", volume 11,8, August 1998

[3] Axis Communications, "CD-ROM Servers, White Paper", 1996

[4] J. Gray, "What Happens When Processors are Infinitely Fast and Storage is Free?", "5th Workshop on I/O in Parallel and Distributed Systems", November 1997

[5] D. Patterson and J. Hennessey, "Computer Architecture: A Quantitative Approach", "Morgan Kaufman", 1996

[6] A. S. Tanenbaum and A. S. Woodhull, "Operating Systems: Design and Implementation. 2nd ed.", "Prentice Hall", 1997

[7] George Lawton, "Storage Technology Takes Central State", "IEEE Computer", 32, 11, November 1999

[8] A. Acharya and M. Uysal and J. Saltz, "Active disks: programming model, algorithms and evaluation", "ASPLOS '98, 8th Conference on Architectural Support for Programming Languages and Operationg Systems", "San Jose, California", pages 212-217, October 1998

[9] K. Keeton and D. A. Patterson and J. M. Hellerstein, "A case for Intelligent Disks (IDISKs)", "SIGMOD Record", 27, 3, pages 42-52, July 1998

[10] E. Riedel and G. Gibson and C. Faloutsos, "Active storage for large-scale data mining and multimedia", "VLDB '98, 24th Int'l Conference on Very Large Data Bases", "New York, USA", pages 62-73, August 1998

[11] D. J. DeWitt and P. Hawthorn, "A Performance Evaluation of Database Machine Architectures", "VLDB '81", September 1981

[12] M. Uysal and A. Acharya and J. Saltz, "An evaluation of architectural alternatives for rapidly growing datasets: active disks, clusters, SMPs", "Technical Report TRCS98-27", "Dept. of Computer Science, University of California, Santa Barbara", October 1998

[13] M. Uysal and A. Acharya and J. Saltz, "Evaluation of Active Disks for Decision Support Databases", "HPCA", 2000

[14] G. Gibson and D. Nagle and K. Amiri and F. Chang and E. Feinberg and H. Gobioff and C. Lee and B. Ozceri and E. Riedel and D. Rochberg and J. Zelenka, "File server scaling with network-attached secure disks", "Sigmetrics '97, Int'l ACM Conference on Measurement and Modeling of Computer Systems", "Seattle, Washington", June 1997

[15] M. Rodriguez and N. Roussopoulos, "Automatic Deployment of Application-Specific Metadata and Code in MOCHA", "7th Conference on Extending Database Technology (EDBT)", March 2000