[147] in Athena_Backup_System
No subject found in mail header
daemon@ATHENA.MIT.EDU (Jonathon Weiss)
Fri Dec 15 15:40:13 1995
To: vrt@MIT.EDU
Cc: athena-backup@MIT.EDU
In-Reply-To: Your message of "Thu, 14 Dec 1995 17:39:48 GMT."
<199512141739.RAA19152@horton.MIT.EDU>
Date: Fri, 15 Dec 1995 15:40:00 EST
From: Jonathon Weiss <jweiss@MIT.EDU>
>> 2) Down time before restoration of service *
you need an actual amount of time here
I think it should be from when we find out it's down, but it should
make some attempt to tell us when that is plausible. (eg., it
shouldn't be required to tell us the master server is down because f a
power hit (we have other tools for that, but if it can't read the DB
when it starts up, it should probably scream pretty loudly)
>> 3) The system must restore tapes that are 6 months old.
As we discussed, it needs to be able to read older tapes, but testing
that may be impractical.
>> 6) A complete backup cycle should cover all of the data
...that is supposed to be backed up (as was pointed out this seems
more like a def'n than a req.)
>> 7) Is there a req for the backup cycle time
I believe the req. is 2 weeks althogh we'd like to be able to cycle
more quickly.
>> 8) req for performance
I'm not sure I know what you mean, but I think that a req on the cycle
time is sufficient.
>> 9) How big a universe should it deal with
Teh Athena and dev cells currently contain ~200G of disk space, of
which ~140G is used. Given that we want the system to work for some
time, I would say that 1,000G is not an unreasonable number. However,
it shoudl be noted that additional tape drives may be purchased if the
cycle time is too large with that much data.
>> 10) max amount we would have to restore at one time
It should probably be able to handle 10G, in up to several thousand
volunes without thinking too hard.
>> 11) The system shall have the following documents: Programmer ref, Op guide
>> Admin guide, etc.
man pages
>> 13) The system will deal with media errors on write.
I think at one point teh concept of a permenant record of all errors
on a tape was mentioned.
>> 17) The system will reject incorrect tapes
(ie wrong or no label)
>> 18) The system shall have available the number of tape errors on a given tape
>> available
put this next to #13
>> 19) The system shall have an ascii interface
do you mean curses? command line?
>> 20) The system shall be able to backup every volume in all the cells we control
Is this a meaningful req?
>> 21) The system shall be able to describe the system
huh?
>> Design critiria
>>
>> 1) The system Shall allow for different tape types
>> 2) The system shall allow for different file servers (NFS vs AFS)
inital version will be AFS only.
