les robertson cernit0999 1

1
Technology Trackingprocessors, memory, storage,
architecures

• Summary of conclusions
• LCB Workshop, Marseilles, 1 October 1999
• Team members Ingo Augustin, Jean-Philippe
  Baud, Fabien Collin, Charles Curran,
  Fabrizio Gagliardi, Frédéric Hemmer,Sverre
  Jarp, Gordon Lee, Alessandro Marchioro,
  Bernd Panzer Steindel, Les Robertson, Ben
  Segal, Rainer Többicke, Pierre Vande Vyvre

2
processors, memories, basic systems

• Previous report (1996) was
• pessimistic about frequency
• optimistic about parallelism (instructions/cycle)
• about right for performance and cost
• The Semiconductor Industry Association 1997
  roadmap predicts a new generation every 2 years
  with 0.1? at 2 GHz appearing around LHC startup
• Previous forecasts have been too
  conservativeBUT we start to come close to
  fundamental limits like the thickness of the gate
  oxide and power dissipationfor which novel
  solutions are required

3
Semiconductor Industry Association (SIA) 1997
technology forecast
Solution being pursued No known solution
4
Processor Cost Caveats

• Estimated cost in 2005
• 0.75-1.60 per CERN-Unit
• CAUTION!
• Investments, costs and prices
• fabulous fab costs
• production volume?
• competition?
• Achieved parallelism
• the best processors only reach 1.8 ipc on integer
  codes
• will Merced make a significant advance?
• compilers may be as important as semiconductors
• Home office PCs are the driving force

5
Secondary Storage
75 of hard disks go into PCs

• hard disk market is big healthy
• 30B with 5-6 large manufacturers
• areal density should continue to evolve at 60
  per year
• from 4 Gbits/inch2 to about 60 Gbits/inch2 in
  2005
• 20-25 Gbits/inch2 demonstrated
• and prices should continue to decrease at 35
  per year
• 2-4/GB in 2005
• Technology limits - room for optimism
• super-paramagnetic limit currently around 30-40
  Gbits/inch2
• competitive RD in materials and heads
• alternative technologies using advanced
  magneto-optics
• Data rate per unit of installed capacity - could
  be a concern
• data rate increases only with the linear density
• if the mass market does not demand significantly
  smaller disks we shall have to purchase more
  capacity than we really need

6
Optics Exotics

• Paradoxically
• now that magnetic recording has caught up with
  optical recording densities (5 Gbits/inch2 with
  a red laser)
• the entertainment business has produced a product
  which could play an important role in low end
  computer storage
• DVD-RAM - 9.4GB in 2000
• access time, data transfer rate - no match for
  hard disk
• but may well displace magnetic tape in the home
  office
• Holographic storage
• development stalled waitingfor a new material
• active consortium (military interest)
• Nothing else beyond the stage of technology
  demonstration

STOP PRESS The Keele Ultra High Density
Memory 2.3 TB on a credit card for 50
7
Tertiary Storage

• Major doubts about the long term
• the market driver is backup and archive --
  write once, read never
• the commodity market will have strong
  competition from DVD-- low-end tape devices may
  disappear
• BUT - several recent/announced products for the
  data centre (half inch, linear)
• STK 9840, DLT 8000, LTO
• with strong roadmaps for the next 5-6 years
• 2005 20-50 MB/sec gt100GB per cartridge
• Costs
• reliable drives unlikely to go below 10-20K
• robotics - gt20 per slot (no improvement on
  todays best price)

1998 Estimated tape revenue by device class
0.5"19 mm Helical
5B market
6
0.25" Linear
9
8 mm Helical
12
0.5" Linear
50
4 mm Helical
23
8
Tape media costs

• At LHC media costs will dominate
• cannot assume that there will be successors to
  current product families in the next 5 years
• so we may only see factor of 4 improvement in
  recording density over todays half inch linear
• giving about 0.50 per GB for raw tape
• Worst case scenario?
• If hard disk is 2/GB and DVD-RAM 0.20many
  applications will abandon tape for the higher
  functionality of random access storage
• Big problem for the magnetic tape business - and
  HEP
• We should be very careful about planning for tape
  as a medium for active data storage

9
Storage Management Systems

• Lack of standardisation gt no exchange of data
• Limited high end market
• missing functionality
• product plans / survival unclear
• Relatively straightforward HEP requirements
• good scalability, predictable performance
• Could be a case where a HEP solution would be
  better and affordable (but surely not multiple
  HEP solutions!)
• Storage Area Networks, Global Shared File Systems
• very interesting developments
• could mature by 2005, and should be followed
• but watch the impact on storage costs

10
Architecture, Interconnects Clusters

• Exotic architectures and technologies will have
  no impact on general purpose computing in the
  next 5 years
• quantum, optical .. multi-threaded (TERA), chaos,
  ..
• games consoles, integrated TV/Internet
  devices will help to squeeze prices, push
  performance of regular PCs
• Expect the distinction between large SMPs, NUMA
  shared memory and SPP systems to become less
  clear
• BUT large integrated systems are likely to remain
  relatively expensive and be unnecessary for HEP
• Should see healthy developments in scalable
  clusters with standardised high performance
  commodity interconnects
• management tools a concern
• there are many promising development projects,
  but they may emphasise high availability/performan
  ce rather than high throughput

11
Summary
full report http//nicewww.cern.ch/les/pasta/wel
come.html

• processors OK
• disks OK - but watch the I/O-rate/capacity ratio
• Concern about the future of magnetic tapes
• it would be wise to consider using disk for
  active data
• relegating tape to archive and data exchange
• The inherent HEP parallelism makes it realistic
  to build large computing fabrics from inexpensive
  components
• But the management - farms and storage - may
  require HEP developments
• Maybe open source may allow this to be done as
  part of a wider collaboration