IU's TeraGridaccessible HPSS archival storage system

1
IU's TeraGrid-accessible HPSS archival storage
system

• Overview about HPSS
• Feeds and speeds, and particulars of IU's
  distributed HPSS environment
• Moving data to and from IU's HPSS via the
  TeraGrid and gridftp
• Hands on exercises with gridftp and HPSS

2
HPSS Background

• IU installed HPSS system in 1998
• HPSS is distributed between Indianapolis and
  Bloomington campuses.
• Over 1.8 petabytes of storage available and can
  scale to 4.2 petabytes
• In aggregate, HPSS can handle as much as 2GB/s of
  data flowing in or out.

3
HPSS Design
IUB Subsystem
IUPUI Subsystem
HPSS Core Servers
Research Network
Research Network
FC SAN
FC SAN
4
HPSS Performance

• 100MB/sec single thread over gigabit
• IU HPSS capable of gt1GB/sec in parallel over
  multiple 10 gigabit
• 150TB of DDN disk for cache
• 40 IBM Jaguar drives - 100MB/sec transfers

5
Storing Data

• Minimum file size should be one MB
• Prefer data to be at 10MB or larger
• Small files should be aggregated
• Maximum file size is 10TB
• By default, data is stored on tape at each campus

6
Class of Service

• Each file has class of service (COS)
• COS defines
• maximum file size
• number of copies
• stripe width
• With GridFTP, COS is chosen automatically by file
  size
• User can override by appending ,,ltCOSgt to the
  filename

7
COS Listing

• COS 1 - default max file size 10GB
• COS 2 - max file size 40GB
• COS 3 - max file size 10TB
• COS 4 - max file size 10TB
• 16 way to disk
• 4 way to tape
• Single copy COS available

8
Quotas

• Default is 1TB
• Can request up to 10TB
• Double these numbers if single-copy COS used

9
Accessing MDSS

• Data goes to disk cache first
• Data copied to tape ASAP
• Copy sent to other campus by default
• Data purged from disk cache as space needed
• Data staged back to disk cache as needed

10
Accessing MDSS

• Fastest Methods
• hsi (hsi.mdss.iu.edu)
• gridftp (gridftp.mdss.iu.edu)
• pftp_client (ftp.mdss.iu.edu)
• kerberized ftp (ftp.mdss.iu.edu)

11
Accessing MDSS

• Convenient Methods
• sftp (sftp.mdss.iu.edu)
• https (www.mdss.iu.edu)
• Samba (smb.mdss.iu.eu)
• Hpssfs

12
hsi example

• intrigue seiffert hsi
• Principal seiffert
  
• seiffertPassword
  
• Username seiffert UID 1529 CC 1529 Copies 1
  hsi.3.3 Wed Jan 17 113107 EST 2007
• /hpss/s/e/seiffert-gtput myfile.tar
• put myfile.tar' '/hpss/s/e/seiffert/myfile.tar
  \ ( 4051006 bytes, 4198.5 KBS (cos1))
• /hpss/s/e/seiffert-gtget myfile.tar
• get '/Users/seiffert/myfile.tar' \
  '/hpss/s/e/seiffert/myfile.tar' (2007/01/14
  120936\ 4051006 bytes, 8275.6 KBS )
• /hpss/s/e/seiffert-gt

13
sftp example

• intrigue seiffert sftp sftp.mdss.iu.edu
• Connecting to sftp.mdss.iu.edu...
• seiffert_at_sftp.mdss.iu.edu's password
• sftpgt put welcome.wmv
• Uploading welcome.wmv to /ssh\-chroot/cos1/hpss/s/
  e/seiffert/welcome.wmv
• welcome.wmv 100 3956KB 3.9MB/s 0001
• sftpgt get welcome.wmv
• Fetching /ssh-chroot/cos1/hpss/s/e/seiffert/welcom
  e.wmv to welcome.wmv
• /ssh-chroot/cos1/hpss/s/e/seiffert/welcome.wmv
  \ 100 3956KB 3.9MB/s 0000
  

14
IU HPSS from TeraGrid

• Accessible by either
• gridftp.archive.iu.teragrid.org
• gridftp.hpss.iu.teragrid.org
• Need to add TeraGrid certificate to IU Big Red
  System
• uberftp and globus-url-copy tested and supported
• tgcp is not tested or supported

15
globus-url-copy Example

• Simple example to and from
• globus-url-copy -rp gsiftp//my.local.site/myfile.
  tar \ gsiftp//gridftp.archive.iu
  .teragrid.org/myfile.tar
• globus-url-copy -rp \
• gsiftp//gridftp.archive.iu.teragrid.org/myfile.t
  ar \ gsiftp//my.local.site/myfile.tar
• Parallel transfer to and from
• globus-url-copy -rp -stripe -bs 8388608 \
  gsiftp//my.local.site/bigfile.tar \
  gsiftp//gridftp.archive.iu.teragrid.
  org2814/bigfile.tar,,4
• globus-url-copy -rp -stripe \
  gsiftp//gridftp.archive.iu.teragrid.org2814/bigf
  ile.tar \ gsiftp//my.local.site/

16
uberftp example

• Simple Example
• uberftpgt put myfile.tar myfile.tar
• uberftpgt get myfile.tar
• Parallel Example
• uberftpgt open -P 2814 gridftp.archive.iu.teragrid
  .org
• uberftpgt lopen my.local.site
• uberftpgt parallel 4
• uberftpgt blksize 8388608
• uberftpgt put bigfile.tar bigfile.tar,,4