sunflash-Distributed to mailing list sun/NC/north-carolina sunflash-Send requests, problems to owner-sunflash@suntri.east.sun.com ---------------------------------------------------------------------------- The Florida SunFlash Aggregate Computing: Seminar Notice & Technical Newsletter SunFLASH Vol 59 #23 November 1993 ---------------------------------------------------------------------------- Abstracts 59.23.A Educational Seminar from Aggregate Computing "Distributed Computing in the 90s The Future of Computing" is a free, half-day technical seminar for software engineers and project managers who are interested in improving the performance of client/server applications through parallel and distributed processing Locations: Santa Clara, CA New York, NY Boston, MA (54 lines) 59.23.B Distributed Directions - The Distributed Computing Newsletter Welcome to Distributed Directions, the distributed computing newsletter from Aggregate Computing. Whether you are a developer, manager, or executive, Distributed Directions is designed to keep you informed about the latest distributed computing technologies. If you manufacture software or if software applications are criitical to your company's mission, you should know what's happening in the world of distributed computing. Subject: DISTRIBUTED COMPUTING: THE SECOND WAVE Subject: SPECIAL REPORT: CATCH THE WAVE Subject: CHANGING THE WAY WE USE NETWORKS Subject: WRITING SECOND GENERATION DISTRIBUTED APPLICATIONS WITH NETSHARE Subject: FREE DISTRIBUTED COMPUTING SEMINARS Subject: SERIES: DEVELOPMENT ISSUES FOR DISTRIBUTED APPLICATIONS (33 lines) -------------------------------------------------------------------------------- Subject: Educational Seminar from Aggregate Computing From: rajesh@aggregate.com "Distributed Computing in the 90s The Future of Computing" A free, half-day technical seminar for software engineers and project managers who are interested in improving the performance of client/server applications through parallel and distributed processing. Presentations and discussions will focus on current trends in distributed computing; enabling technologies that facilitiate the building of "network-aware" applications; case studies and open forum discussion. Presenters: o John Rymer, Vice President and Senior Analyst, Patricia Seybold Group o Joan Wrabetz, President, Aggregate Computing, Inc. o Dean Mason, Vice President of Engineering, Aggregate Computing, Inc. Dates: November 30, 1993 December 2, 1993 January 26, 1994 Location: Westin Hotel Holiday Inn Westin Copley Hotel 5101 Great America Pkwy 49th & Broadway 10 Huntington Ave Santa Clara, CA New York, NY Boston, MA 408/986-0700 212/977-4000 617/262-9600 Times: 8:30 - 11:30am 8:30 - 11:30am 8:30 - 11:30am or or or 1:30 - 4:30pm 1:30 - 4:30pm 1:30 - 4:30pm To register for this seminar, complete the registration form below and send it to seminar@aggregate.com, or call us at 612/546-5579 (fax: 612/546-9485). Space is limited, so register today. Name: Title: Company: Address: City/State/Zip: e-mail address: telephone: -------------------------------------------------------------------------------- Subject: Distributed Directions - The Distributed Computing Newsletter From: lisa@aggregate.com Published Quarterly by Aggregate Computing, Inc. Volume 1, Number 4 November 1993 Volume 1, Numbers 1, 2, and 3 are also available from Aggregate Computing. Just send your request to newsletter@aggregate.com. ******************************************************************************** Copyright 1993 Aggregate Computing, Inc. All rights reserved. Aggregate is a registered trademark and NetMake, NetShare, and NetShare SDK are trademarks of Aggregate Computing, Inc. All other trademarks are the property of their respective holders. You are free to forward unmodified sections of this newsletter via e-mail if you include this copyright notice. ******************************************************************************** Welcome to Distributed Directions, the distributed computing newsletter from Aggregate Computing. Whether you are a developer, manager, or executive, Distributed Directions is designed to keep you informed about the latest distributed computing technologies. If you manufacture software or if software applications are criitical to your company's mission, you should know what's happening in the world of distributed computing. ******************************************************************************** We welcome your opinions, suggestions, and ideas. Send your comments to: Aggregate Computing, Inc. 300 South Highway 169, Suite 400 Minneapolis, MN 55426 Phone: 612-546-5579 Fax: 612-546-9485 E-mail: newsletter@aggregate.com ******************************************************************************** In This Issue... Distributed Computing: The Second Wave Special Report: Catch the Wave - Changing the Way We Use Networks - Writing Second Generation Distributed Applications with NetShare Series: Development Issues for Distributed Applications - Component Placement ******************************************************************************** Subject: DISTRIBUTED COMPUTING: THE SECOND WAVE Even though distributed computing has been in use for many years, it still hasn't become a mainstream method of computing. As software developers and software users continue to migrate towards distributed computing, they realize the need for new and improved applications and technology for what could be called "the second wave of distributed computing." What is the key to developing this second wave of distributed computing? John Rymer, senior analyst and vice president at Patricia Seybold Group and editor of the Distributed Computing Monitor, has conducted extensive research and analysis in this area. He suggests the answer is to add an additional set of underlying distributed computing services that complement today's client-server technologies and support the flexible deployment of resources across networks without complex system-level coding. This set of services should provide functionality to make distributed computing a production technology. The "pioneer" distributed computing products are in a state of virtual disuse. Because the first client-server projects proved costly and didn't provide the expected benefits, they were not widely implemented or understood. In general, commercial distributed computing products are incomplete, and few applications deliver the flexibility and "reuse" promised by distributed computing. Distributed computing still requires too much programming and low-level expertise that developers don't have and don't want. They want to focus on their own jobs-building vertical applications. With the second wave of distributed computing, "the network is the computer." In other words, you can use your local machine to access resources across the network as if all of the activity is happening locally. To provide this "network-transparent computing," distributed computing platforms must be made of heterogeneous parts that can easily plug & play, allowing you to build distributed applications without knowledge of the underlying network structures. These platforms should also behave in a coherent, coordinated fashion, based on your changing needs, no matter which pieces you use. Currently, this isn't the case. The distributed file systems, client-server databases, remote procedure calls, and message-oriented middleware available today don't provide complete distributed computing solutions or even work together. You need new architectural scenarios, plug & play components, and easy-to-use applications. How will you meet these demands? One solution is to use a distributed engine as a basis for all distributed applications, so no low-level programming is necessary. This "distributed computing brain" would handle the remote execution, resource management, and communications for each application and could be used with any combination of underlying distributed computing tools. A distributed engine is one way to meet the needs of the second wave of distributed computing. ******************************************************************************** Subject: SPECIAL REPORT: CATCH THE WAVE In the following two articles, Joan Wrabetz, president and founder of Aggregate Computing, Inc., discusses how resource management and network transparent computing meet the needs of the second wave of distributed computing. ******************************************************************************** Subject: CHANGING THE WAY WE USE NETWORKS The networks of the 90s are comprised of an array of disparate computing resources, from mainframes to desktop machines. A second generation of distributed computing applications is being developed that attempts to meet users' needs by running distributed applications across this loose, cooperating mix of computing resources. To get at and use this network of resources, applications need resource management. Resource management makes the diverse computing resources on a network look like a single, cohesive computing resource to an application. It also effectively matches application needs to those computing resources. To build this second wave of distributed applications, the essential problem that must be solved on networks is matching the supply of diverse computing resources to application demand on a network-wide basis. Matching isn't Easy Matching computing resources to application needs is not easy because both the supply of computing resources and the applications that demand those resources are diverse. Networked computers come in a variety of types and sizes. They range in capacity and are suited to different types of problems. For example, a superserver emphasizes high performance I/O bandwidth, while a graphics machine emphasizes high resolution graphics and graphics accelerator hardware as well as special purpose graphics software. Even two machines with the same hardware often become specialized by the way they are configured and the type of software installed. To make matters worse, both the hardware and software configured on a machine are likely to change over its lifetime. Networks of machines are a constantly changing continuum of capability. Thus, it's important for the matching of the supply of computing resources to application demand to occur dynamically and automatically. Autonomy is Critical Making the computing resources on a network look like a single, coherent computing resource is difficult because all the resources to be managed are autonomous. They can't be expected to behave consistently, and their use is limited by the rights and access privileges allowed by the site and its users. Thus, the available computing resources to be managed are also dynamic and limited by the constraints of autonomy. All of these factors must be taken into account when providing access to network-wide resources. Aggregate Computing has developed a distributed processing engine called NetShare that performs resource management, effectively making a network of computers accessible to an application as a single computing resource. NetShare determines what capacity is available on each computer based on dynamic usage, total machine size, and user limits on usage. The individual, available capacities are then managed as a single, cohesive unit by NetShare. NetShare lets users and applications access and use this virtual computing resource through interfaces that make the capacity appear to exist on the local machine. Meeting Application Needs But managing the network-wide supply of computing resources is only half the problem. Applications' needs must be met in the process. Because capacity is really physically separated, it's important to select resources that can best meet application needs. Some resources aren't appropriate for an application. For example, you wouldn't do part of a compile of a Sun-4 application on an IBM workstation. Nor would you call an application on a machine that isn't licensed to run it. Since the network is heterogeneous, resources must be classified and checked for appropriateness prior to use by a specific application. NetShare addresses this need by being policy-oriented. The NetShare engine matches the application policy that describes an application's resource needs with available and appropriate resources. This is done within the constraints of site and user policy for usage that defines what resources can be used and what access and security limitations apply. These policies are set by each application to meet its individual needs. Thus, NetShare manages resources by matching application needs to computing capacity in spite of heterogeneity, autonomy, and application diversity. ******************************************************************************** Subject: WRITING SECOND GENERATION DISTRIBUTED APPLICATIONS WITH NETSHARE The first generation of distributed tools are just too hard to use. This leads to poor architectures and poor programming. It also results in the development of programs prone to failure because the network is inherently more unreliable than a single machine. The second generation of distributed computing tools recognizes and addresses the needs of distributed applications for interfaces that mask applications from network interfaces, operating system dependencies, and lower level distributed computing tools. NetShare provides these capabilities. Because applications access the network through NetShare, they need not be programmed to any network interface, related operating system interfaces, or installed distributed computing services, such as the distributed file system, security service, name service, or transport interface. NetShare acts as a layer between these interfaces and the distributed application. NetShare manages the work of interacting with the network, so you write very little code. NetShare interfaces are high-level and can be used without supplying any network related information. As a result, applications written to NetShare are not only easy to write, but they are highly portable. Since NetShare allows the underlying distributed computing and network services to be configured during installation, applications are easily transferred to environments that have different services or levels of service (security, for example) installed. Because NetShare dynamically matches your application's needs to network resources, your application is completely independent of your network topology and configuration. Thus, it is unaffected by changes to the configuration and requires no network or system administration. Since network access is encapsulated in the NetShare engine and since NetShare responds dynamically to changes in network configuration, distributed applications written using NetShare do not require any network administration or configuration. In fact, all the applications written to NetShare share a single engine, resulting in consistency and guaranteed interoperability of all NetShare applications. NetShare is an important second generation distributed computing tool that divorces applications from the network, the operating system, and machine types. It transcends machine/network boundaries to allow network-wide access to computing resources. In other words, applications are network transparent. And with NetShare, these applications can be built with only minor changes to existing code and without complicated network interfaces. ******************************************************************************** Subject: FREE DISTRIBUTED COMPUTING SEMINARS Aggregate Computing presents a free seminar designed to educate industry professionals about distributed computing and how to implement it. Distributed Computing in the 90s: The Future of Computing * Tuesday, November 30, 1993 Westin Hotel, Santa Clara, CA * Thursday, December 2, 1993 Holiday Inn Crowne Plaza, New York, NY * Wednesday, January 26, 1994 The Westin Hotel, Boston, MA This half-day seminar features John Rymer, vice president, senior analyst, and editor of the Distributed Computing Monitor, Patricia Seybold Group. John discusses the distributed computing market. Joan Wrabetz, founder and president of Aggregate Computing, reveals how your network can become a "virtual machine" and how Aggregate's technology easily enables dramatic application performance improvements. You'll hear customer case studies as well as a panel discussion featuring distributed computing experts. Choose to attend the morning (8:30 - 11:30) or afternoon (1:30 - 4:30) session. Space is limited, so call today to register for this free educational seminar. 1-800-966-1666 ******************************************************************************** Subject: SERIES: DEVELOPMENT ISSUES FOR DISTRIBUTED APPLICATIONS by D. Dean Mason This article is the fourth in a series that addresses development issues for distributed computing applications and discusses component placement. The following information is reprinted with the permission of UNIX Review. Component Placement Once a distribution model and its underlying communication model have been chosen, the next issue developers face is component placement. Where should the pieces of the distributed application run? Proper component placement is entirely application dependent, and it is crucial that both the applications's developers and ultimately its administrators have a thorough understanding of the service's execution characteristics. Server components that are computation-bound, such as number and simulators, will benefit the most from a fast machine that isn't typically bogged down with I/O. Conversely, I/O-bound server components, such as file services, will generally reap the most benefit by being on the machine to which their principal I/O device is attached. Although the proper choices are often obvious, proper component placement is critical since it can have a direct impact upon the performance, availability, reliability, and security of the distributed application. Next Issue: Naming ABOUT THE AUTHOR... D. Dean Mason is Aggregate's Vice President of Software Development and a recognized expert in distributed application development. For a complete reprint of this article, "Distributed Programming," call 1-800-966-1666 or send e-mail to newsletter@aggregate.com. ********************************************************************** For information about SunFlash send mail to info-sunflash@Sun.COM. Subscription requests should be sent to sunflash-request@Sun.COM. Archives are on draco.nova.edu, ftp.uu.net, sunsite.unc.edu, src.doc.ic.ac.uk and ftp.adelaide.edu.au For Gopher and WAIS access: sunsite.unc.edu. (Login as 'gopher' for a simple gopher client, 'swais' for a simple WAIS client (over 500 databases). All prices, availability, and other statements relating to Sun or third party products are valid in the U.S. only. Please contact your local Sales Representative for details of pricing and product availability in your region. Descriptions of, or references to products or publications within SunFlash does not imply an endorsement of that product or publication by Sun Microsystems. Send brief articles (e.g. third party announcements) and include contact information (non-800#, fax #, email, etc) to: John McLaughlin, SunFlash editor, flash@Sun.COM. +1 305 351 4909