Aucun titre de diapositive

1
THE UK DATA STORAGE NETWORK (DSNet-UK) C David
Wright School of Engineering and Computer
Science University of Exeter Exeter EX4 4QF, UK
2
The UK Data Storage Network
Outline of Talk Data Storage in the UK Network
Aims and Objectives Network Targets and
Scope Network Membership Data Storage Families
and Market Trends Technological Limits? Scanning
Probe Based Storage - a new paradigm for small,
low power, high density storage for un-tethered
applications ?
3
The UK Data Storage Scene
According to UK Directory of Information Storage
Manufacturing and RD (2nd edition - DTI
Publication - see http//www.mackintoshconsultants
.co.uk)
54 UK Companies 15 UK Universities
Involved in some aspect of data storage drive
manufacture components and sub-assemblies test
and manufacturing equipment consumables and
services
HP Bristol - digital tape systems - 1000
employees
Seagate Northern Ireland - HDD read/write heads -
1500 employees
Infortrend - Surrey - RAID controllers - 5
employees
SomerData -Wells- PC-based real time data
storage - 3 employees
APH Industries - Buxton - HDD process chemicals -
60 employees
4
Aims of DSNet-UK
To promote and grow an inclusive
industrial-university network to determine
appropriate goals, aspirations and development
strategies for the UK's data storage research,
technology and manufacturing base.
Routes for the successful implementation of such
strategies will then be explored and put in
place. Likely to include
development of UK Data Storage 'Road Map' general
and topical meetings and workshops invited
addresses by world-leading (industrial) storage
experts staff exchanges joint funding bids
Steering Committee Chairman - Eddie Townsend
(Xyratex) Co-ordinator - David Wright (University
of Exeter) Members - Andrew Pauza (Plasmon), Eric
Mayes (NanoMagnetics), Barry Middleton
(University of Manchester), David Jenkins
(University of Plymouth) DTI Monitor - Nigel
Mackintosh
5
Network targets and scope
A UK Roadmap/strategy document by the end of
month 9 (Dec 2004) 1 general and 2 topical
meetings per year 1 invited world-leading
industrial speaker per year 30 person/days
exchanges per year joint funding applications
totalling 1 million per year Aim to double the
number of industrial members by months
18 Technological scope and priorities of the
network will of course be heavily influenced by
the outcome of the UK Roadmap exercise.
However, likely focus is magnetic recording
(disk and tape) optical data storage
(particularly phase-change) scanning probe
based storage MRAM and PCRAM memories
6
Start-up membership
Industrial NanoMagnetics - Bristol (Eric
Mayes) Philips - Southampton (Simon Bramwell
John Kinghorn) Plasmon - Cambridge (Andrew
Pauza) Xyratex - Havant (Eddie Townsend)
Academic University of Aston (Prof John Sullivan
- tribology) University of Central Lancashire
(Prof Phil Bissell - noise in magnetic
media) University of Exeter (Prof C David Wright
- storage materials and systems) University of
Glasgow (Prof John Chapman - electron
microscopy) University of Manchester (Prof Barry
Middleton - storage materials
systems) University of Plymouth (Prof Des Mapps -
storage materials and systems) University of
Sheffield (Prof Mike Gibbs - magnetic materials
and SPM)
7
Industrial members - Xyratex
Sites in UK, USA, Singapore, China Malaysia
Fibre channel RAID - 690MBytes/s with up to
35TBytes storage
8
Industrial members - Plasmon
Sites in UK and USA,
12 inch TrueWorm technology 5.25 inch MO jukebox
technology Ultra Density Optical technology (UDO)
UDO Roadmap
9
Industrial members - Philips Systems Laboratory
Sites Worldwide Systems Lab in Southampton
focuses on IC design for future optical disk
formats (Multi-layer DVD, blu-ray, MAMMOS,
near-field systems, portable formats etc)
Optical card
Portable blue
10
Industrial members - NanoMagnetics
Site in Bristol UK
Biologicallly inspired particulate media for
high-density magnetic storage regular 8nm
diameter with 4nm ferromagnetic core Focusing on
low-cost, high-density flexible storage e.g.
miniature floppy for removable applications with
DVD-like capacity 100 million digital video tapes
shipped in 2002 !
11
Mass storage families and markets
Mass storage families Magnetic recording hard
disks magnetic tapes (analogue digital),
floppy disks etc Optical recording CD, DVD,
Blu-Ray magneto-optic (Sony Minidisc) etc Solid
state storage Compact flash card, memory stick,
USB drive etc Emerging technologies - MRAM,
PCRAM SPM-based storage (MEMS-based storage)
12
Future market trends
13
A time for change ?
A changing environment ? Personal computer has
been dominant electronic platform in past (office
tasks, e-mail, web, games, computing, data
logging etc) -relatively power
hungry Un-tethered (mobile) devices will be the
dominant platform in the future (laptops, PDAs,
digital cameras, mobile phones, personal music
and video players etc etc) - need
(ultra) low power and (ultra) small form factors
Technological limits ? Magnetic recording -
superparamagnetic limit - no clear cut way to
true nanoscale storage? Optical recording -
optical diffraction limit Solid-state storage -
scaling problems
somewhere in the not too distant future we are
going to have to change technologies to keep
going forward
Mark Kryder, Senior Vice President, Seagate
Research
un-tethered devices will usher in a new
component set, consisting of non-volatile PLDs,
non-volatile memory, and MEMS-based storage
Gilder Technology Report, March 2003
14
Some un-tethered platforms
Philips HDD060 audio player 1.5GByte HDD-based
storage 10 hr battery life, 150 euro
PDA/pocket PC
Samsung camera phone SGH-D410 2inch VGA display,
10MByte storage POP3/IMAP4 e-mail compliant games
software still pictures plus 30 seconds MPEG video
15
Roadmap for storage density
Atoms ?
10 Pbit/sq.in
30 CGR
1 Pbit/sq.in
Molecules ?
100 CGR
100 Tbit/sq.in.
HDD industry aiming for 1Tbit/sq.in. storage
density by 2008 to 2012
20 25
16
Can we follow the roadmap ?
1 Tbit/sq.in 1 bit - 25 x 25nm
by 2010 ?? 100 Tbit/sq.in. 1 bit - 2.5 x
2.5nm by 2020 ?? 1 Pbit/sq.in 1
bit - 0.8 x 0.8nm by 2025 ?? diameter
cobalt atom - 0.25 nm
If we continue to use surface storage, only
available tools known today that can manipulate
on these scales are based on scanning probe
microscopy AFM, EFM, MFM, MRM, STM, etc
Ultra-high density storage that is not
lithographically dominated
Alternatively - we need to consider volumetric
storage
17
IBM Millipede
Current status 1 Tbit/sq.in. demonstrated 10nJ
per bit to write
64 x 64 tips erasability demonstated
18
Storage Evolution (from IBM)
Drive
Micro drive
Nano drive
19
Practical Nanoscale Storage with PC material ?
InProM Project GeSbTe alloy electro-thermal
recording electrical readout
20 nm dot 100 nm pitch
gt 300 Gbit/sq.in
500 nm
Contact recording - probes suitable for 2-D
array low power - 0.1nJ per bit to record 10nm
bits already achieved - 3nm stable even at high
temperatures 50-100 Tbits/sq.in. possible?
1 Tbit/sq.in
Image courtesy of Serge Gidon, Yves Samson,
Olivier Bichet, CEA-LETI, Grenoble
20
Various Probes Storage Techniques
Mechanism
Media
Physical mode
Pressure
21
Possible system performance
1st generation 5x5mm device, 2.5 GByte capacity,
4 Mbit/s data rate 40x40nm bits, 400 Gbit/sq.in.,
tip pitch 100?m, 32x32 tip array, 4kbit/s per tip
2nd generation 1x1cm device, 20 GByte capacity,
50 Mbit/s data rate 25x25nm bits, 1Tbit/sq.in.,
tip pitch 100?m, 64x64 tip array, 6kbit/s per tip
3rd generation 1x1cm device, 80 GByte capacity,
200 Mbit/s data rate 12x12nm bits, 4Tbit/sq.in.,
tip pitch 50?m, 128x128 tip array, 12kbit/s per
tip
Millipede V1.0 3x3mm device, 1 GByte capacity, 1
Mbit/s data rate 40x40nm bits, 375 Gbit/sq.in.,
tip pitch 100?m, 32x32 tip array, 1kbit/s per tip
22
Advent of the true single-chip computer ?

• A true single-chip computer would consist of
• CPU
• fast-volatile core memory
• mass storage
• communications (i/o)
• The mass-storage element is missing from todays
  single-chip computers
• If we could find a way to include it, we could
  open the way for true embedded intelligence
• (truly intelligent behaviour needs lots of
  software and lots of processing - needs lots of
  memory)
• Every appliance might become smart and
  communicative - true ambient intelligence

To implement this with MEMS-based storage is the
dream of several researchers world-wide (Richard
Carley, CMU) e.g. CPU gt 500MIPS, RAM gt 64MB,
Mass memory gt 1GB, i/o gt 100MB/s by 2010 ? Also
require low cost (order magnitude lower than
flash), low power
23
A way forward - research areas ?
System architecture and integration development
of probe array memories compatible with
conventional IC processing development of
alternatives to traditional cantilever
architectures multiplexing, interconnections,
coding, signal processing issues cost reductions
(simple architectures, fewest contacts per tip,
passive rather than active actuation, reduction
in number of processing steps) actuation
(scanning, tracking, tip approach - comb drives,
linear electrostatic drives, piezo, other?)

• Media development
• write-once or rewritable media or both?
• phase-change, magnetic, polymer, other?
• low power write (and read) operation
• high density capability (sub 10nm bits)
• cyclability and archivability
• high SNR capability
• integration with IC processing

Nanoscale science thermal and electrical
properties on the nanoscale (ballistic
conduction, quantum effects) material processes
on the nanoscale (crystallisation/amorphisation,
magnetic switching, melting/freezing,
polymerisation)
24
Memory of the future ?
Magnetic hard disk, with perpendicular/HAMR to
1Tbit/sq.in by 2010 - 2015 ? Too power hungry for
next-generation dominant un-tethered platforms
? Can we ever reach 50 - 100 Tbit/sq.in. with
traditional granular media approach ?
Phase-change media have thermodynamic stability
for storage in the 50 - 100 Tbit/sq.in. range
(chemical stability ?) - How do we write and
read bits 3 nm in size ? SPM-based techniques
may hold the answer - EU is strong in this field !
Will disk-based 2-D storage always be King ?
Users dont care about what memory - just
performance and cost
What would we do with 1 Parabit/sq.in. density
? 10kB image, 5 frames/second, 75 years ? 1
Parabit video of entire life on 1 inch square !