Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!tut.cis.ohio-state.edu!rutgers!aramis.rutgers.edu!klaatu.rutgers.edu!josh
From: josh@klaatu.rutgers.edu (J Storrs Hall)
Newsgroups: comp.arch
Subject: Re: Cryo-refrigerators
Keywords: breakthroughs refrigerators
Message-ID: <Mar.13.15.58.38.1989.8399@klaatu.rutgers.edu>
Date: 13 Mar 89 20:58:44 GMT
References: <4387@pt.cs.cmu.edu> <13288@steinmetz.ge.com> <4409@pt.cs.cmu.edu> <11020@pasteur.Berkeley.EDU>
Organization: Rutgers Univ., New Brunswick, N.J.
Lines: 30

muir@postgres writes:
    >expected - like optic fiber.  (The single-mode optic fibers that are in
    >the ground right now, have an intrinsic bandwidth of 35,000 gigahertz,
    >in the IR window alone. This time, it isn't a typo.)

    Correct me if I'm wrong, but does this mean that you could store ~100GB
    in a 3000 mile fiber loop?

    Is an optical loop a viable storage medium?  Some benifits: n access points,
    < 1 second access to ALL the data....  Problems: reliability.

Well, if you have something that will switch at 35,000 GHz, let me
know.  Optic fiber can be a viable storage medium if we come up with
the appropriate (read photonic) switching capability.  

A 3000-mile loop is a bit much, I would think.  You don't want that
anyway; it would give you 16 millisecond latency.  3000 miles of
50-micron fiber occupies about 250 cubic meters, the volume of a small
house. 

The only reason to talk about fiber is that we have fiber drivers now;
but if we had the photonic switching technology necessary to use it as
a storage medium efficiently, it would be easier to use free-space
storage (bounce the beam around between mirrors).  At a reasonable
limit for visible light you can store a bit in about a 5-micron cube
of empty space; this gives you 100 terabytes in an (empty!) 20-inch box.

Need those switches, though.

--JoSH