Re: tapes

[Bernhard Lauss <lauss@berkeley.edu>]

Dear Peter,


Our present status on LTO2 tapes should be
Berkeely Tapes:

30  from run 7
45  from run 8
10  from Illinois  ?


Run 8 was 4 weeks data + 10 days of calibration ~ 45 tapes

adding for FADC 20%
           WFD  40%

Total 72 Tapes:

Consequently 90 Tapes should be sufficent for the fall run.
An in case we run ut of tapes, we still have the archive
for a few runs at the end.

Copying via tape transfer to NCSA
1 week shipping PSI -> UIUC
1 TByte / 2 days copying -> 6 TByte = 14 days
1 week shipping UIUC -> PSI
Total of 6 weeks for data recycling.


I suggest that we should have
3 full tapeloads for the robot = 90 Tapes


1) I think we can use the 30  run7 tapes right away, as nobody
   ever is going to look at them.

2) What is the progress of copying run8 data to NCSA,
   when will it be finished ?
   We should make sure that we have those 45 tapes by end of October

3) Are there 10 tapes from Illinois ?

So in total you must have  85 LTO-2 Tapes  for MuCAp sitting
around at UIUC.


P l e a s e     c o u n t    a n d    c o n f i r m .


Berkeley will buy the additional 5 missing
tapes to complete 3 full robot loads.
Presently $ 37,80 per Tape at
http://store.tapedata.com/maxelllto2.html


                                Best regards

                                        Bernhard


*******************************************************************
Bernhard Lauss                 E-Mail: lauss@socrates.berkeley.edu
Physics Department
366 LeConte Hall
University of California @ Berkeley
Berkeley, CA 94720                             Tel: (510)-642 4057
United States                                  Fax: (510)-642 9811
*******************************************************************


On Wed, 10 Aug 2005, Peter Kammel wrote:

> Dear friends,
>
> Let us think a bit, how we will solved the storage problem
> in the fall.
>
> At the moment, I have started copying the skimmed data on the
> USB2 disks to the NCSA archive. With two streams I get a total
> transfer rate of 8+ MB/s. This allows the transfer of 1TB
> in two days. As I have received no complaints by now, I will
> try that rather than moving the data disks to NCSA.
> In the future that speed should improve once the connection
> is fully on Gbit.
>
> After Steve and Fred succeed with the tape backup of the skimmed
> data, we will start to tranfer the 5 TB run data to to the
> NCSA archive. After that is done and a production completed
> successfully, I believe we can recycle our LTO2 tapes for
> the upcoming run!
>
> Now the questions are
>
> - which tape robot to use for primary storage at PSI?
>    Should we try to twist the arms of our MuLan friends, so that
>    the leave their robot at PSI during MuCap running?
>    Should we send the Berkeley robot to PSI? If we do that, I
>    suggest leaving it there, because any transport is
>    dangerous and expensive. This implies using NCSA storage
>    in the future.
>    Or should be purchase a LTO3 drive for the Berkeley robot.
>    How expensive is that.
>
> - During the run we then would have ??TB of tapes and 1TB of
>    USB2 disk plus 1TB? of local disk. We could consider sending
>    LTOs to Illinois and loading them to NCSA in quasi real time.
>    The other option is network copying, but it is not clear
>    if that's fast enough.
>
> Bernhard, please update your spreadsheet according to these options
> and advice us. Of course, it would be ideal if we can avoid the
> tape expenses and use our money one critical experimental items,
> like hydrogen filters etc.
>
> Best regards
>
> Peter
>
>
>
> Peter Kammel  /  pkammel@uiuc.edu
> Department of Physics, Loomis Laboratory
> University of Illinois at Urbana-Champaign
> 1110 West Green Street, Urbana, IL 61801
> Tel (217) 333-5424 / Fax (217) 333-1215
>
>