Skip to Content.
Sympa Menu

bluesky - RE: the mercury file system

bluesky AT lists.ibiblio.org

Subject: Global-Scale Distributed Storage Systems

List archive

Chronological Thread  
  • From: "Josh" <josh AT mercuryfs.net>
  • To: "'Global-Scale Distributed Storage Systems'" <bluesky AT franklin.oit.unc.edu>
  • Cc: "Charles \(E-mail\)" <charles AT mercuryfs.net>, "Steven Torres \(E-mail\)" <steven.torres AT oracle.com>, "Lana Akamine \(E-mail\)" <lana AT valinux.com>
  • Subject: RE: the mercury file system
  • Date: Mon, 30 Jul 2001 19:25:26 -0700

Yes, but you gotta see the engineering details. This really is a very basic
design, as far as global file systems go. I actually won a bet by keeping it
within a 2 inch binder (its still incomplete). I agree, everybody else out
there has problems with this type of project. I have studied the political
challenges more than the engineering challenges, to be honest. I figured
that even if I get the design to work, I would never be allowed to
distribute it. The only way MFS will be successful is if the world trusts
it, so it has to have strong encryption.

I've put together a basic plan for the revisions, its attached. Basically, I
intend to get it up and running as a read only file sharing program, then
read write, then move it into the kernel. I've gotta have reliability as #1.
That's why I'm going for funding. Even if we did code it, we don't have the
support infrastructure that the initial versions will require, before
Microsoft and novell take over and write their own clients. If it gets to
that point.

By starting off as a database, I can borrow all their clustering, fault
tolerance, mirroring, load balancing, etc etc.
I intend to keep it as a dbase project until versions 2 or 3.



-----Original Message-----
From: Bram Cohen [mailto:bram AT gawth.com]
Sent: Monday, July 30, 2001 7:26 PM
To: Global-Scale Distributed Storage Systems
Cc: Steven Torres (E-mail); Lana Akamine (E-mail); Charles (E-mail)
Subject: Re: the mercury file system

On Mon, 30 Jul 2001, Josh wrote:

> True! However, I hope to entice the home users because of the other
> features, such as security and async mirroring (what will make The
Internet
> as Primary Disk scenario a reality).

While it's natural to try and make your project do as much as it possibly
can, the failure rate of software projects is such that I for one try and
make mine do just one thing for one person. A successful project which
does very little is, after all, more useful than a failed one which tried
to do a lot.

-Bram Cohen

"Markets can remain irrational longer than you can remain solvent"
-- John Maynard Keynes


---
You are currently subscribed to bluesky as: josh AT mercuryfs.net
For list information visit http://www.transarc.ibm.com/~ota/bluesky/
0.1
Single OS, file sharing utility (outside kernel), read only, no
encryption, no mfs domains, no friendship protocol, no default
replication, no http/html.

0.2
focus on S2 and S3 design, begin designing load test apps

0.3
focus on S1 design but stil read only, my group dbase, processes
(1601)

0.4
dns load balancing, mfs domains, begin building S0 network

0.5
manual sync groups, lay groundwork for auto sync groups, default
record replication, begin building UNI ID

0.6
encryption

0.7
multiple S1s (multi location), lay groundwork for read/write mode

0.8
timestamping, NTP, virtual servers, processes (1604, 1607)

0.9
read/write mode, processes (1602, 1603, 1604, 1605), virtual servers

1.0
auto sync groups, begin linux kernel driver development

1.1
Virtual CDs

1.2
async mirroring

1.5
SE linux kernel mode version, begin NT driver development,





Archive powered by MHonArc 2.6.24.

Top of Page