Current driftscan network

1
Computing Network Required For Real-Time
Transfer/Analysis of Palomar Quest Data

• Current driftscan network
• Current pointstare network
• Real-time data transfer
• Real-time analysis only
• D. Rabinowitz, 2005 Dec 7

2
Palomar Quest Computers
camera
q7
q16
disks
q17
disks
asok
pointy
telescope
T1
disks
berlioz
camdata
disks
Microwave
Yale
Berkeley
Caltech
3
Microwave transfer rate to Yale 20
Mbits/sec (one compressed image in 70 s using
bbftp and tar files)
Driftscan Data Flow
camera
q7
q16
disks
q17
Gigabit hub
disks
pointy
telescope
100 Mbit hub
berlioz
Microwave
Yale
4
PointStare Data Flow
expose
transfer
40s
camera
q7
readout
30s
q16
repoint
Gigabit hub
asok
pointy
telescope
100 Mbit hub
disks
berlioz
camdata
disks
Microwave
median transfer rate to Caltech one compressed
image in 190 s (using ftp and no tar files)
Berkeley
Caltech
5

• key issues
• q16 must be fast enough to read/write without
  interrupting continuous scan
• microwave transfer sometimes too slow for realtime

Real-time Transfer
camera
q7
q16
disks
q17
Gigabit hub
disks
pointy
telescope
acquire
acquire
Microwave
gzip/tar
gzip/tar
Caltech
transfer
transfer
SDSU
Berkeley
Yale
6

• key issues
• q16 must be fast enough to read/write without
  interrupting scan
• Transfer detections only in real-time over T1
  line
• Gzip/tar raw during day, transfer next night

Real-Time Analysis
camera
q7
Caltech
detections
q16
disks
blade Server
Gigabit hub
q17, q18, q19, q20,
pointy
telescope
100 Mbit hub
T1
disks
acquire
acquire
gzip/tar
Microwave
transfer
transfer
data
Yale
analyze
analyze
7
Questions to be considered

• Is microwave link fast enough for real-time
  transfer ?
• (2) Is q16 fast enough to write/read without
  interruption
• (3) How will data be packaged for transfer to
  Yale ?
• (4) How will access to SDSU be managed ?
• (5) Who will manage SDSU node
• (6) Why not process on the mountain ?