John William Lumley

User-space protocols deliver high performance to applications on a low-cost Gb/s LAN

Two important questions in high-speed networking are firstly, how to provide Gbit/s networking at low cost and secondly, how to provide a flexible low-level network interface so that applications can control their data from the instant it arrives. We describe some work that addresses both of these questions. The Jetstream Gbit/s LAN is an experimental, low-cost network interface that provides the services required by delay-sensitive traffic as well as meeting the performance needs of current applications. Jetstream is a combination of traditional shared-medium LAN technology and more recent ATM cell- and switch-based technology. Jetstream frames contain a channel identifier so that the network driver can immediately associate an incoming frame with its application. We have developed such a driver that enables applications to control how their data should be managed without the need to first move the data into the applications address space. Consequently, applications can elect to read just a part of a frame and then instruct the driver to move the remainder directly to its destination. Individual channels can elect to receive frames that have failed their CRC, while applications can specify frame-drop policies on a per-channel basis. Measured results show that both kernel- and user-space protocols can achieve very good throughput: applications using both TCP and our own reliable byte-stream protocol have demonstrated throughputs in excess of 200 Mbit/s. The benefits of running protocols in user-space are well known- the drawback has often been a severe penalty in the performance achieved. In this paper we show that it is possible to have the best of both worlds.

Afterburner (network-independent card for protocols)

Many current implementations of protocols such as the Transmission Control Protocol/Internet Protocol (TCP/IP) are inefficient because data are often accessed more frequently than necessary. Three techniques that reduce the need for memory bandwidth are proposed. The techniques are copy-on-write, page remapping, and single-copy. Afterburner, a network-independent card that provides the services that are necessary for a single-copy protocol stack, is described. The card has 1 MByte of local buffers and provides a simple interface to a variety of network link adapters, including HIPPI and asynchronous transfer mode (ATM). Afterburner can support transfers to and from the link adapter card at rates up to 1 Gbit/s. An implementation of TCP/IP that uses the features provided by Afterburner to reduce the movement of data to a single copy is discussed. Measurements of the end-to-end performance of Afterburner and the single-copy implementation of TCP/IP are presented.<<ETX>>

From XML inclusions to XML transclusions

Modularized documents, composed of fragments from multiple sources, provide users high maintainability and reuse. In the world of XML, powerful and widely-supported solutions exist to create such documents. Surprisingly enough, a lot of interesting features - especially those envisioned by the pioneers of the hypermedia community - are still missing for XML inclusions. The goal of this paper is to investigate these issues and identify possible improvements in this area. Our main inspiration is the Xanadu project and the concept of transclusions. This paper proposes an enhanced model to describe and interact with XML inclusions. We identify multiple classes of inclusions and multiple views of multi-source documents. Particular attention is also given to the idea of making inclusions transparent to both users and applications. An engine producing composite documents, with rich information about inclusions, and a viewer for modularized XML documents are presented as well.

Higher-level layout through topological abstraction

Existing layout languages provide support for geometric properties allowing - and in a sense forcing - users to give a complete geometric description of the desired output: if the characteristics of the output medium change, the layout of the whole document has to be reworked completely, as the properties set by the user are no longer appropriate for the modified context. In this paper we propose a different paradigm which allows users to produce layouts by describing their topological and abstract properties, rather than geometric ones. We first define and detail topological properties as abstract relationships between the document components, independent from the output characteristics, and then describe an XML-based layout language based on these concepts, called TALL. A running engine able to transform topological layouts into actual PDF files, based on XSLT and the DDF framework, is presented as well.

AAL5 at a Gigabit for a Kilobuck

We present a novel LAN that has been designed to meet three criteria: it should be low cost, support standard protocols, and provide high performance. The LAN interconnects many workstations in a ring topology with a link rate of 1 Gbit/s. The packet format is the same as that of a B-ISDN cell except that our packets can have arbitrary length. The network also provides hardware support for ATM Adaptation Layer 5 as weIl as TCP/IP.We have implemented the ring network and we describe a network interface card which provides hardware support for critical functions such as calculating checksums. This card is used in conjunction with a second card that supports a single-copy implementation of the TCP/IP protocols. The application-to-application throughput has been measured at rates in excess of 200 Mbit/s between two workstations.

Configurable editing of XML-based variable-data documents

Variable data documents can be considered as functions of their bindings to values, and this function could be arbitrarily complex to build strongly-customised but high-value documents. We outline an approach for editing such documents from example instances, which is highly configurable in terms of controlling exactly what is editable and how, capable of being used with a wide variety of XML-based document formats and processing pipelines, if certain reasonable properties are supported and can generate appropriate editors automatically, including web-service deployment.

Evaluating invariances in document layout functions

With the development of variable-data-driven digital presses where each document printed is potentially unique there is a need for pre-press optimization to identify material that is invariant from document to document. In this way rasterisation can be confined solely to those areas which change between successive documents thereby alleviating a potential performance bottleneck.Given a template document specified in terms of layout functions, where actual data is bound at the last possible moment before printing, we look at deriving and exploiting the invariant properties of layout functions from their formal specifications. We propose future work on generic extraction of invariance from such properties for certain classes of layout functions.

Resolving layout interdependency with presentational variables

In the construction of variable data documents, the layout of component parts to build a composite section with heterogeneous layout functions can be implemented by a tree-evaluating layout processor. This handles many cases with well-scoped structure very smoothly but becomes complex when layout relationships between components cut across a strict tree. We present an approach for XML-described layouts based on a post-rendering set of single-assignment variables, analagous to XSLT, that can make this much easier, does not compromise layout extensibility and can be a target for automated interdependency analysis and generation. This is the approach used in the layout processor associated with the Document Description Framework (DDF).

Functional, extensible, svg-based variable documents

Architectures for documents that vary in response to binding to data, or user interaction, are usually based on limited layout semantics, such as text flows, and simple data variability, such as replacing reserved constructs. By using a generalised XML graphical representation (SVG), decorated with an extensible set of layout intent declarations, and with embedded fragments of XSLT decorated with program retention directives, it is possible to produce self-contained documents that are both highly flexible and extensible and can adapt their presentation to multiple stages of data binding, as well as user interaction. The essentials of the architecture are presented with examples and details of the necessary implementation and support tools, most of which are written in declarative, functional XSLT. Recent developments in XSLT technologies make it possible to consider such documents operating within unmodified browsers - techniques are discussed.

Speculative document evaluation

Optimisation of real world Variable Data printing (VDP) documents is a difficult problem because the interdependencies between layout functions may drastically reduce the number of invariant blocks that can be factored out for pre-rasterisation. This paper examines how speculative evaluation at an early stage in a document-preparation pipeline, provides a generic and effective method of optimising VDP documents that contain such interdependencies. Speculative evaluation will be at its most effective in speeding up print runs if sets of layout invariances can either be discovered automatically, or designed into the document at an early stage. In either case the expertise of the layout designer needs to be supplemented by expertise in exploiting potential invariances and also in predicting the effects of speculative evaluation on the caches used at various stages in the print production pipeline.