Review of Implementing RPCs

From: Honghai Liu (liu789_at_hotmail.com)
Date: Wed Jan 28 2004 - 16:48:19 PST

  • Next message: Richard Jackson: "Review: Andrew D. Birrell and Bruce Jay Nelson. Implementing Remote Procedure Calls."

    Reviewer: Honghai Liu

     

    The paper describes the implementation of Remote Procedure Calls in Cedar programming environment in Ethernet LAN in early 1980s.

     

    It is interesting to see this paper since it is widely believed it RPC was invented by Sun and first deployed in Unix based on TCP/IP, however, it was not the case.

     

    RPC was designed and implemented to achieve several goals: clean, efficiency and generality. And it actually turned out to be that way. Most importantly, the idea of generality is novel: it does not depend on any language, hardware or even network protocol (very similar to Java:)

     

    It is also interesting to see how the authors can implement a few things we don't have to worry about today. They had to design the protocol (instead of using TCP/IP), location & naming service (instead of using DNS). This again shows the power of generality and flexibility of the RPC.

     

    RPC becomes a very popular tool in distributed operating systems such as Unix and Windows NT, however, it does not include the objection interface. The authors could have included the notion of objects in the design, given that Smalltalk- the first object oriented language, was available at Xerox at the time of the writing. I guess perhaps it was because one of the goals is simplicity, and the idea of object orientated was not thought to be a simple one then.


  • Next message: Richard Jackson: "Review: Andrew D. Birrell and Bruce Jay Nelson. Implementing Remote Procedure Calls."

    This archive was generated by hypermail 2.1.6 : Wed Jan 28 2004 - 16:48:25 PST