[EMAIL PROTECTED] writes:
> ARLA? Puleeeze!!! PTUI!!!!! I need a file system that can stay up
> for more than 10 minutes,
We are running ARLA on several Linux-workstations for researchers,
they replace Tru64/Alpha machines with Transarc/AFS, mashines uptime
is very well. And the users are pleased.
> and that has decent, as opposed to
> ridiculous, performance.
instressting, in execly which cases ARLA does not perform well ?
> You people should relax and lighten up on IBM/Transarc.
I see no light cause of the following reasons:
1) 3 Months ago we reported bugs in AFS 3.5 fileservers which make the
fileservers 'hang', they are very very important, but IBM does not
even care.
2) IBM tells us out sitelicense has no relevance for new plattforms
like Linux and NT cause it stands in our sitelicense that we only
have support for 'current plattforms' , which means we have to pay
for envery client each, which is totally impossible.
3) Transarc has not implemented a new 'feature' since over 5 years
ago. There are intressting ideas for ex. replication of RW-volumes.
Take a look at http://www.nada.kth.se/~noora/exjobb/index.html , it
is very related to the ARLA-projekt.
4) IBM care more about integration with WebSphere instead of fix bugs
or making real development on the product. I belive that 99% of the
old customers does not even care about 'bla bla bla
WWW-integration' .
> They're not as stupid as they seem.
why should that matter anyway ?
> WHen Transarc releases AFS as open source later this
> summer, I hope to see you all go crawling back to them in gratitude.
I will not care about their Opensource-campain cause of the following
reasons:
1) The source is readable for most customers anyway.
2) I do not belive it will be easier than today to get in patches to
their codebase just cause they are 'opensource'.
3) The client has bad design, you don't want your hole big AFS-client
in your kernel anyway. Compare it to ARLA if you want.
4) I do not belive that IBM will stamp an accepable license on AFS,
which means:
a) I will not be able to read the code if i still want to write
ARLA-code.
b) The opensource-community will not or will not be able to commit
code to the project.
/Jimmy
