DataMuseum.dk

Presents historical artifacts from the history of:

DKUUG/EUUG Conference tapes

This is an automatic "excavation" of a thematic subset of
artifacts from Datamuseum.dk's BitArchive.

See our Wiki for more about DKUUG/EUUG Conference tapes

Excavated with: AutoArchaeologist - Free & Open Source Software.


top - metrics - download
Index: T c

⟦8cea658ce⟧ TextFile

    Length: 6311 (0x18a7)
    Types: TextFile
    Names: »comments.tex«

Derivation

└─⟦3d0c2be1b⟧ Bits:30001254 ISODE-5.0 Tape
    └─⟦eba4602b1⟧ »./isode-5.0.tar.Z« 
        └─⟦d3ac74d73⟧ 
            └─⟦this⟧ »isode-5.0/doc/comments/comments.tex« 
└─⟦2d1937cfd⟧ Bits:30007241 EUUGD22: P.P 5.0
    └─⟦35176feda⟧ »EurOpenD22/isode/isode-6.tar.Z« 
        └─⟦de7628f85⟧ 
            └─⟦this⟧ »isode-6.0/doc/comments/comments.tex« 

TextFile

\documentstyle [small,blackandwhite] {NRslides}

\author {S.E. Kille \\
Department of Computer Science \\
University College London}

\date {October 1988}
\raggedright
\title {ISODE --- Further Comments}

\begin {document}

\maketitlepage

\begin {bwslide}
\ctitle {Overview}
\begin {itemize}
\item To complement Marshall Rose's talk
\item To give a European perspective
\item To give some more details on lower layers and ASEs (other talks focus
on the applications)
\end {itemize}
\end {bwslide}

\begin {bwslide}
\ctitle {Coding Approaches}
\begin {itemize}
\item Procedural interfaces corresponding to service definitions
\begin {itemize}
\item Easy to understand
\item Flexible
\item General
\item Sometimes wrong (e.g. RTS)
\end {itemize}

\item Full implementations of the protocols and services

\item All ASEs and lower layers can be used asynchronously
\begin {itemize}
\item Must be suitable support from Transport or Network interface (e.g.,
Sockets)
\item Needs to have sufficient kernel buffering for OPDU writes
\item ISODE deals with OPDU read buffering
\end {itemize}

\item Flexible layering 
\end {itemize}
\end {bwslide}



\begin {bwslide}
\ctitle {What about the X.25}
\begin {itemize}

\item X.25/CONS is essential for real OSI 
\begin {itemize}
\item Statement of religion
\item TCP/IP or TP4/CLNS are not suitable/desirable for WANs
\end {itemize}

\item Current implementations
\begin {itemize}
\item SUNOS + SUNLINK X.25
\item Vax + BSD UNIX + UBC Code + DMF 32
\item Vax + Ultrix + CAMTEC Dexpand with X.25 or Pink Book (latter is not a
product).    
\item Straightforward to add in others
\begin {itemize}
\item Easiest if socket-like interface
\item Device-like interface possible
\end {itemize}

\end {itemize}
\end {itemize}
\end {bwslide}


\begin {bwslide}
\ctitle {What about the X.25 (2)}

(or why TCP/IP is used on UNIX LANs)


\begin {itemize}

\item Problems with UNIX X.25
\begin {itemize}
\item Not sufficient coverage
\item No standard interface
\item Poor performance
\item Not robust
\item Some interfaces will not allow for asynchronous working 
\item Problem with the implementations, not X.25
\end {itemize}

\item What is needed
\begin {itemize}
\item Standard interface (POSIX transport may give this)
\item Pink Book
\item Realistic Performance
\end {itemize}

\item ISODE supports X.25, but X.25 does not support ISODE

\end {itemize}
\end {bwslide}

\begin {bwslide}
\ctitle {Transport}
\begin {itemize}
\item TP0 is a part of ISODE
\item TP0++ over TCP in ISODE
\begin {itemize}
\item RFC 1006
\item Best choice in LAN environment (currently)
\end {itemize}

\item Straightforward to interface to other Transport {\em Services}

Variety of application disciplines:
\begin {itemize}
\item Use of Transport Daemon (tsapd) 
\item Application Servers
\item Callback
\item Multithreading
\end {itemize}

\item TP0 extensively tested
\end {itemize}
\end {bwslide}

\begin {bwslide}
\ctitle {Session}

\begin {itemize}
\item Full Session Protocol and Service implemented
\item Substantial interworking tests
\item Version 2 (88) in ISODE 4.0 
\end {itemize}

\end {bwslide}


\begin {bwslide}
\ctitle {Presentation}
\begin {itemize}
\item Most of Protocol and Service implemented
\item IS
\item Simple and Full encoding
\item X.410 mode
\item Some interworking (FTAM + VTP)
\end {itemize}

\end {bwslide}



\begin {bwslide}
\ctitle {ASN.1}

\begin {itemize}
\item Not bound to presentation layer
\item Special Tools (discussed elsewhere)
\item Abstract representation as C structures as intermediate form 
\item Used extensively
\item Tools have handled a wide range of specifications
\end {itemize}
\end {bwslide}

\begin {bwslide}
\ctitle {ACSE}
\begin {itemize}
\item Full implementation (lots of parameters)
\item The interface used by the application
\item AE Titles still a problem
\begin {itemize}
\item OBJECT IDENTIFIER used by stub directory
\item Distinguished Name will be used in QUIPU context
\end {itemize}

\end {itemize}
\end {bwslide}



\begin {bwslide}
\ctitle {RTS}
\begin {itemize}
\item 1988 and 1984 (X.410)
\item Some interworking tests of X.410 done
\item Not spooled (i.e. RTS application has to do the work)
\item New interface being done
\end {itemize}
\end {bwslide}


\begin {bwslide}
\ctitle {ROS}


\begin {itemize}
\item Full implementation
\item  ROSY/POSY facilitate use
\item  Various versions 
\begin {itemize}
\item X.410 
\item ECMA ROS (BCS Session)
\item 1988 ROS (RTS optional)
\end {itemize}

\item Common interface, except for initialisation

\item ECMA ROS used by THORN

\item 1988 ROS (no RTS) used by other ROS applications
\end {itemize}
\end {bwslide}



\begin {bwslide}
\ctitle {Performance}
\begin {itemize}
\item Full implementation leads to large processes:
\begin {itemize}
\item Minimum 300kbyte
\item DSA 500 kbyte
\item FTAM 800 kbyte
\end {itemize}

\item Substantial effort to avoid byte copying
\begin {itemize}
\item qbufs
\item PPPPPP
\end {itemize}

\item Low bulk transfer overheads
\begin {itemize}
\item  ROS is 12\% slower than raw TCP
\item  FTAM is 8\% slower than Arpanet FTP
\item  More CPU intensive
\item  X.25 too slow to be relevant
\end {itemize}

\item Initialisation not so fast
\begin {itemize}
\item Typically a few seconds
\item Paging in process is a big overhead
\end {itemize}

\end {itemize}
\end {bwslide}


\begin {bwslide}
\ctitle {How good is it really?}
\begin {itemize}
\item Does not claim to be production quality
\item Better than much ``production software'' I have seen
\item Code Quality
\begin {itemize}
\item Few comments
\item Well structured
\item Long variable names
\item Documented (700 pages)
\item Careful version control
\item Thorough testing of releases
\end {itemize}

\item Ideal for experimentation and early services

\item Possible drawbacks for products
\begin {itemize}
\item Lower layers and ASEs not tuned for specific applications

\item ASN.1 Tools
\begin {itemize}
\item Good overall functionality, but
\item Too much code output
\item Memory based
\item Need SOPY
\end {itemize}

\end {itemize}


\end {itemize}

\end {bwslide}

\begin {bwslide}
\ctitle {Other Applications}
\begin {itemize}
\item NRS Lookup Protocol
\item IDIST (ROS based RDIST)
\item Management Protocols
\item ODA/ODIF
\end {itemize}

\end {bwslide}


\end {document}