Since the June 1998 Von Ruden Review, progress has been made on many fronts. Here is a short summary of the effort during the last 6 months.

• Requirements understood - many discussions with D0 folks, and SAM group in particular
• Technical Design Document - expanded greatly, includes examples and command syntax.
• Encp
  • User client program - Distributed via Fermilab's UPD mechanism
  • Completely static, binary task with no dependencies
  • Same syntax as normal Unix cp command, including input file list and output directory specification
  • Special D0 SAM output completed.
  • Option added that lists what transfers are active and pending for a given node.
  • Optimal transversal over tape.
• Pnfs
  • The pnfs servers have been installed and thoroughly tested on 3 Linux machines.
  • A user query oriented script has been written.
  • The servers can be installed via Fermilab's UPD mechanism.
  • A directory tree based on volumes and file families has been added to aid the users in doing optimal tape transversals.
• Error Control, Robustness and Retries - extensively studied and developed in all sections of Enstore.
  • Trace debugging tool (based on VME work) written for python - solved many problems
  • Timer task written and available for scheduling
• Enstore Servers
  • Underlying communication protocol very reliable.
  • Config Server - is now the sole location of all Enstore configuration information allowing easy maintenance.
  • Volume and File Clerks - more robust, disaster recovery procedures almost in place.
  • Library Manager - Priority, aging, optimal tape transversal
  • Mover - optimal and overlapped IO, flexible user wrappering mechanism in place, CRC studies complete (adler32)
  • Log Server - little work done, useful as is.
  • Media Changer - converted to C module, CDF STK code incorporated, EMASS module almost written EMASS training attended by 3 of Enstore group.
  • Inquisitor and Alarms - new since last review, central location for user to see what is happening with system, many useful web pages, plots, patrol investigations begun.
• Volume Import and Export - plan for remote sites understood (not done).
• Test System - 10 Linux node cluster constructed, installed into FNAL and RIP networks, Tests working, SAM integration begun.
  
  (also available in Postscript)

  Preliminary Rates Measurements

  Tape Drive	Writing	Reading	Network	"Memory->tape"
  	(MB/S)	(MB/S)	(Mbits/S)	(MB/S)
  AIT	2.7	2.7	94	2.7
  Mammoth	2.8	2.7	94	2.8
  DLT	4.9	4.8	94	4.9
  STK	8.8*	7.5*	83	9.8

  * == STK tests not repeated (yet) after coding changes that improved performance

  Typical CPU usage, per transfer, for the AITs, Mammoths, and DLTS is 5-10% on a client node 10-20% on the mover nodes.

For further details, please consult the Technical Design Document at  Enstore Technical Design Document (http://www-hppc.fnal.gov/enstore/design.html)

Questions can be sent to the Enstore mailing list at enstore@fnal.gov.