dotgnu-general
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [DotGNU]The Worldwide Computer


From: Bill Lance
Subject: Re: [DotGNU]The Worldwide Computer
Date: Wed, 13 Feb 2002 05:58:00 -0800 (PST)

--- Rhys Weatherley <address@hidden> wrote:
> 
> Re-reading the article, I don't think they went
> quite that far.
> They argued that some data (e.g. government and
> accounting
> information) cannot be distributed all through-out
> the network,
> because of the privacy implications.
> 
> Some central management of this kind of data is
> necessary.
> I don't think they implied that all data should be
> managed
> this way.  DotGNU has a similar goal: protect the
> user's
> privacy by limiting what goes where.
> 


Perhaps we should distinguish between a 'central
control server' and a 'sole source server'.  A
participants personal and business data may indeed be
available only from a single server in the control of
that data owner, a 'sole-source server'. Other
business's data will be stored on thier server as
well.  And dotGNU and the VRS intends to support and
allow this. It is one signle point of control for the
entire system that is the issue.  Quoted from the
article:

"The system should be coordinated by servers operated
by the ISOS provider, which could be a
government-funded organization or a consortium of
companies that are major resource sellers and buyers.
(One can imagine competing ISOS providers, but we will
keep things simple and assume a unique provider.)
Centralization runs against the egalitarian approach
popular in some peer-to-peer systems, but central
servers are needed to ensure privacy of sensitive
data, such as accounting data and other information
about the resource hosts. Centralization might seem to
require a control system that will become excessively
large and unwieldy as the number of ISOS-connected
hosts increases, and it appears to introduce a
bottleneck that will choke the system anytime it is
unavailable. These fears are unfounded: a reasonable
number of servers can easily store information about
every Internet-connected host and communicate with
them regularly. Napster, for example, handled almost
60 million clients using a central server. Redundancy
can be built into the server complex, and most ISOS
online services can continue operating even with the
servers
 temporarily unavailable.

 The ISOS server complex would maintain databases of
resource descriptions, usage policies and task
descriptions. The resource descriptions include, for
example, the host's operating system, processor type
and speed, total and free disk space, memory space,
performance statistics of its network connections, and
statistical descriptions of when it is powered on and
connected to the network. Usage policies spell out the
rules an owner has dictated for using her resources.
Task descriptions include the resources assigned to an
online service and the queued jobs of a
data-processing task."

The above mixes the two ideas of 'central server and
'sole-source server'.  The 'sole-source server' stores
someones sensitive data, presumably under especially
tight access controls.  The 'central server' controls
the operation of the network itself, storing and using
information about the nodes.  We are finding that to
be a neceessary function.  And we are defining the
special role of the Cluster Administrator to meet
these needs.  The administrator, in effect, has root
privilages to the running Cluster.

The above recognizes the need for 'sole-source
servers' to protect confidential data, which I agree
with. They then use that as a justification for a
'central control server', which I do not agree with. 
They then assure us that fears of a bottleneck are
unfounded, which I probably agree with.  But then
never mention the fears of the power to monitor,
control and restrict activities on the ISOS.  No
government, no consortium and no single company or
individual can EVER be granted that power if the rest
of us are to remain free in any meaningful sense of
the word.

 

> The ISOS is nice in theory.  I'll believe it when I
> see it.
> 


The artcle does raise one interesting point in the
very begining.  Most of the really useful things that
an ISOS can do are already being done, but in user
space.  As the Microsoft assimilation strategy has
taught the poor abused user, many thing should STAY in userspace.

__________________________________________________
Do You Yahoo!?
Send FREE Valentine eCards with Yahoo! Greetings!
http://greetings.yahoo.com


reply via email to

[Prev in Thread] Current Thread [Next in Thread]