Jump to content

RSCS

From Alfawiki
Adapted from the English Wikipedia article RSCS, available under CC BY-SA 4.0Updated 9 August 2025
Remote Spooling Communications Subsystem (RSCS)
Company typeSoftware component
IndustryComputer networking, Operating systems
Founded1973
FounderEdson Hendricks, Tim C. Hartmann
Headquarters
Poughkeepsie
,
United States
ProductsFile and message routing software for VM/370
ParentIBM

Remote Spooling Communications Subsystem (commonly known as RSCS) is a component—referred to as a "virtual machine" within VM terminology—of IBM’s VM/370 operating system. It functions by receiving files sent from both local and remote users or systems, then routing these files to their intended recipients. In addition, RSCS facilitates the exchange of commands and messages among various users and systems.

Before the advent of the Internet, RSCS supported what was then the largest network globally, significantly shaping the evolution of internet technology and fostering acceptance of networking across independently operated organizations. Created by Edson Hendricks and Tim C. Hartmann, the system was internally known at IBM as VNET. The protocols compatible with RSCS extended beyond VM environments, connecting IBM platforms such as MVS and even non-IBM machines.

The development and influence of RSCS within IBM and its broader user base have been chronicled through interviews and contemporary reports by Melinda Varian. Foundational technical objectives and breakthroughs are elaborated upon in key publications authored by Creasy, Hendricks, and Hartmann. Academically, this same software underpinned networks like BITNET on a global scale.

Background

RSCS emerged from IBM’s established requirement for efficient file exchange. Hendricks initially created CPREMOTE around mid-1969, marking the first implementation of a "service virtual machine" and validating the concept’s practicality.

In 1971, Norman L. Rasmussen, then Manager of IBM’s Cambridge Scientific Center (CSC), commissioned Hendricks to enable communication between CSC and other IBM Scientific Centers. Leveraging insights gained from CPREMOTE, Hendricks redesigned the system. Collaborating with Hartmann from IBM’s Technology Data Center in Poughkeepsie, New York, they developed RSCS, which began internal deployment in 1973.

The initial external release of RSCS in 1975 did not yet include comprehensive networking capabilities but contained dormant subroutines for store-and-forward functionality used internally. This feature was later incorporated through the VNET PRPQ, initially supporting file transfer and subsequently extending to messages and command transmissions.

Following these improvements, “the network began to grow like crazy.” While early connectivity depended on manual dial-up lines, VNET quickly transitioned to using leased lines for communication.

By SHARE XLVI in February 1976, Hendricks and Hartmann reported that VNET connected 50 systems across the continent. By SHARE 52 in March 1979, the network had expanded to 239 systems spanning 38 U.S. cities and 10 countries worldwide. VNET surpassed 1,000 nodes by 1983, reached 3,000 nodes in 1989, and by the 1990s connected over 4,000 nodes—approximately two-thirds running VM operating systems. For comparison, ARPANET had grown to 213 hosts by 1981; both networks experienced rapid expansion.

In 1986, IBM’s Think magazine estimated that VNET contributed to annual savings of $150 million for the company through enhanced productivity.

Other RSCS Protocol Compatible Networks

Due to its pivotal role in networking development, RSCS line drivers became known collectively as the "RSCS Protocols," integrating protocols from diverse sources. The CPREMOTE protocol is believed to have been the first symmetrical ("balanced") protocol. To incorporate MVS systems, Hartmann reverse-engineered the HASP Network Job Interface protocol, which accelerated network expansion. Subsequently, he added support for JES2 Network Job Entry as an RSCS/VNET line driver.

BITNET, a collaborative university network established in 1981 by Ira Fuchs (CUNY) and Greydon Freeman (Yale), was founded on VNET principles. Its inaugural connection linked CUNY with Yale. BITNET (using RSCS protocols) was ported to non-IBM platforms and widely deployed on VAX/VMS, alongside DECnet; notably, the VAX/VMS NJE stack was known as Jnet. At its peak around 1991, BITNET encompassed nearly 500 institutions and approximately 3,000 nodes, predominantly educational entities across North America (including NetNorth in Canada), Europe (EARN), India (TIFR), and parts of the Persian Gulf region (GulfNet). The network also saw significant adoption in South America with roughly 200 actively used nodes during the late 1980s and early 1990s.

Over time, BITNET integrated into the Internet ecosystem. Later versions of RSCS along with Jnet and UNIX NJE stacks supported TCPNJE line drivers. As most BITNET sites had Internet connectivity, previous leased line and dial-up BITNET connections were tunneled over Internet links. Additionally, operating NJE over SNA remained common practice.

Technical Challenges

R. J. Creasy characterized RSCS as a core component of the VM/370 Time-Sharing System's operating environment, which included the Control Program (CP), Conversational Monitor System (CMS), and RSCS itself. Collectively, these elements provided extensive access to IBM System/370 computational resources. The primary role of RSCS was to oversee data exchange between interconnected systems.

From a technical standpoint, RSCS operated as a point-to-point "store and forward" network, more akin to UUCP than ARPANET. It did not depend on dedicated Interface Message Processor hardware or constant connections; instead, messages and files were transmitted in full from one server to another until they reached their intended endpoint. In the event of a connection failure, RSCS preserved the data and attempted retransmission once the remote host became reachable again.

VNET Compared to ARPANET

  • VNET represented the first large-scale network without permanent connections, enabling computer access through dial-up lines, which lowered expenses compared to ARPANET’s original dedicated 50kbps lines (subsequently upgraded to 230kbps). At that period, most leased lines had a maximum speed of 9600 baud.
  • The routing and path selection approach used by VNET was streamlined and later influenced Internet protocols.
  • Unlike ARPANET, which depended on a centralized control facility managed by Bolt, Beranek, and Newman in Cambridge, MA, VNET exhibited genuine "distributed control."
  • While VNET served primarily as a store-and-forward file transfer system, ARPANET functioned as a versatile network supporting both interactive and batch applications.

References

See also