[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Position paper

From: Tom Bachmann
Subject: Re: Position paper
Date: Wed, 10 Jan 2007 18:50:37 +0100
User-agent: Thunderbird (X11/20061231)

Hash: SHA1

Neal H. Walfield schrieb:
>> However, it is not clear to me how the policies creatable in this way
>> can be ordered (e.g., can a pool of the first type exemplfied created as
>> a subpool of the second example?).
> Does my first example (with Alice and Bob) clarify this point?

To some extent. I'll try to give an example of the case I consider unclear:
Suppose Bob and an unspecified number of other users all have allocated
from the user pool (a proportional share pool). Now Bob wants to run an
audio player. For this player, real-time guarantees about the scheduling
are preferable, e.g. "run at least 2 ms every 20ms".
Here it is not clear if Bob can derive such a pool from his main pool.

>> I don't think cpu time pools are to be passed to servers. Although this
>> would increase accounting, it would as well horrify the complexity of
>> the server and require special kernel support, as has been discussed on
>> the list (or on coyotos-dev?).
> I think they should.  Especially if we are to try and provide quality
> of service guarantees and reduce cross talk.  [1] provides one possible model 
> for this.
>  [1] U. Steinberg, J. Wolter, H. Härtig: Fast Component Interaction
>      for Real-Time Systems
>      http://os.inf.tu-dresden.de/papers_ps/steinberg_ecrts2005.pdf

As has been discussed on the list, the most complex problem is callers
revoking donated time pools at arbitrary times. I don't think the paper
pays appropriate account on this (at least not when considered as
solving the general problem, which is not actually it's intent, iiuc.
Also, the needed in-kernel mechanism has to traverse a potentially
unbound number of process relationships [from time to time], which I
think could be problematic, theoretically).

>> What other policies exist? How are they ordered?
> What do you mean by ordered?

The same thing which I described in terms of a music player above, but
this time about memory.

>> [...] how are malicious applications
>> stopped from slowing down the system by dictating a very bad page-out
>> policy?
> Applications are charged for the memory and CPU time this requires.
> Thus, they should only hurt themselves.

This indeed looks like an elegant solution, if the time-donation can be
realized properly.
- --
- -ness-
Version: GnuPG v1.4.6 (GNU/Linux)


reply via email to

[Prev in Thread] Current Thread [Next in Thread]