Subject: RAITH SYSTEM Practices: Items I wish all users would consider to increase through put on the system:
From: James Conway <jwc@snf.stanford.edu>
Date: Fri, 19 Nov 2004 15:14:16 -0800
Fri, 19 Nov 2004 15:14:16 -0800
Thread: RAITH SYSTEM Practices:

Some items I wish all users would consider to increase through put on 
the system:

ALL users are encouraged to be more conciencious about hand offs between 
users with the goal to improve through put and access to the system for 
all users.  I am examining Coral user history  to reservations made,  
honored, and canceled carefully and there has been significant loss of 
opportunity by those whose writes complete early but the system was 
unattended through the early morning hours, and of all users whom 
working on chips have not 'shared the ride' with other users needing the 
system.

E.G.,  6 users working on chips share the ride save 2.5 - 3 hours just 
by reducing the load lock transfer time overhead and you will get better 
results on your write because you are already thermally stabilized on 
the column!


Users canceling or not showing up for their session must post their 
opening to the RAITH list at the earliest opportunity and with a 
explanation as to why they are canceling. Yes -- sometime everyone will 
need to cancel a session, but the level of cancellations recently and 
not informing others of the cancellation is totally unacceptable and 
will not be tolerated further.

E.G., Last night in fact the write completed at 22:37 hours, the next 
user did not enable until ~ 7 AM.  That would have been a week of access 
for anyone waiting -- AND USERS WERE WAITING FOR ACCESS TO MEET URGENT 
DEADLINES. 

Remember the 16 minute rule on reservations and if no one is writing and 
there is not a reservation pending or in place the system is available 
for any user. 

Long write times, such as overnight writes, by users should have a firm 
estimate of when the write completes and preferably a user whom will 
check the system at this time, be ready to hand off the system to the 
next users in the CORAL queue, or then to any available user ready to 
write, if that next reserved user is not ready or available in that order.

Please everyone:
Let us all work together as a community on handing off the system more 
smoothly, reducing idle time, and load lock overhead.  The beam time you 
save could be yours! 

Thank you for your support! Your comments are invited.

James Conway


Thread: RAITH SYSTEM Practices:

Some items I wish all users would consider to increase through put on the system:

ALL users are encouraged to be more conciencious about hand offs between users with the goal to improve through put and access to the system for all users.  I am examining Coral user history  to reservations made,  honored, and canceled carefully and there has been significant loss of opportunity by those whose writes complete early but the system was unattended through the early morning hours, and of all users whom working on chips have not 'shared the ride' with other users needing the system.

E.G.,  6 users working on chips share the ride save 2.5 - 3 hours just by reducing the load lock transfer time overhead and you will get better results on your write because you are already thermally stabilized on the column!


Users canceling or not showing up for their session must post their opening to the RAITH list at the earliest opportunity and with a explanation as to why they are canceling. Yes -- sometime everyone will need to cancel a session, but the level of cancellations recently and not informing others of the cancellation is totally unacceptable and will not be tolerated further.

E.G., Last night in fact the write completed at 22:37 hours, the next user did not enable until ~ 7 AM.  That would have been a week of access for anyone waiting -- AND USERS WERE WAITING FOR ACCESS TO MEET URGENT DEADLINES. 

Remember the 16 minute rule on reservations and if no one is writing and there is not a reservation pending or in place the system is available for any user. 

Long write times, such as overnight writes, by users should have a firm estimate of when the write completes and preferably a user whom will check the system at this time, be ready to hand off the system to the next users in the CORAL queue, or then to any available user ready to write, if that next reserved user is not ready or available in that order.

Please everyone:
Let us all work together as a community on handing off the system more smoothly, reducing idle time, and load lock overhead.  The beam time you save could be yours! 

Thank you for your support! Your comments are invited.

James Conway