W3C | Architecture Domain | Web Architecture
Quality of Service
on the tension between reliability and scalability
taken from old research notebook
another source: reliabe links
see also: metadata architecture
- Resource Discovery
- Finding relavent resources. Traversing a link can be considered
a special case of resource discovery. Other examples:
- "Where's that paper on Distributed Indexing I found yesterday?"
- "What discussion forums, archives, etc. are relavent to
real-estate in Oklahoma?"
- "What's new on the web today?"
- "Show me everything Thomas Jefferson wrote."
- In a fault-tolerant, distributed system, reliability is a measure
of how many times a given computation succeeds out of the number
of times it is attempted. So to define reliability of resource
discovery in general and specifcally linking, we must define what
it means to succeed or fail.
The acceptable level of reliability will vary between
applications, and even between users. A reliable system is one
in which any party may achive arbitrarily high reliability by
investing sufficient resources.
The defintion of "success" will vary as well. Successful link
traversal generally means finding a resource with perfect
precision and recall, and retrieving an authentic representation
of the resource in a timely fashion.
- In a database query or index search, precision is a measure of how
many of the results are relavent. Recall is a measure of how many
of the relavent items are in the results. Related failures include:
- insufficient precision: finding too much "junk." Including
all items in a result set gives perfect recall at the cost of
- insufficient recall: failing to find the relavent resources.
An empty result set has infinite precision, but zero recall.
- Availability and Latency
- Availability means that a resource can be retrieved: the server is up
and reachable via the network. Latency is a measure of the amount
of time between the initation of a request and the completion of
the request. Related Failures:
- latency too high: server machine load too high, request waits
too long in a queue, insufficient bandwidth.
- resource not available: connection refused, machine crashed,
- Authenticity means getting the right data. Failurs include:
- Version skew. If a resource is replicated to increase
availability or decrease bandwidth, it is important that the
various replicas are in sync (or close -- some applications may
be willing to accept out of date information some small
percentage of the time.)
- Access Control
- Access to resources may be restricted to authorized parties.
Copyright, Ownership, Charging, Payment systems
- Attributes and Meta-Information
- indexing information often used during searching. I finally found
a definition of "meta-information" that I like:
Metadata has two main functions:
WAIS through the Web - About Metadata
- to provide a means to discover that the data set exists and
how it might be obtained or accessed; and
- to document the content, quality, and features of a data set
and so give an indication of its fitness for use.
The question arises: whence comes the list of attributes? What's the
expressive capability, structure, and meaning of attribute values?
We can look to directory systems like DCE CDS, Prospero and X.500,
or to frame-based knowledge bases like Algernon and Lincks.