tech-kern archive

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

Re: What is the best layer/device for a write-back cache based in nvram?



On Sat, Sep 10, 2016 at 4:18 PM, Thor Lancelot Simon <tls%panix.com@localhost> wrote:
> On Fri, Sep 09, 2016 at 11:09:49PM +0200, Jose Luis Rodriguez Garcia wrote:
>> B- It can use the buffer cache for avoid read/write cycles, and do
>> only writes if the data to be read is in memory.
>
> I'd stay more focused on avoiding _copies_.  Any sane implementation
> must satisfy reads from the cache layer if the data are cached.  I
> would avoid confusing this layer with the page or buffer caches.

What are these _copies_ ? I don't understand it.

>> E- It can be useful to use elevator algorithms for do the writes from
>> buffer cache to raid.
>
> I definitely don't agree.  If you do this, you'll end up with request
> sorting at at least three layers!  The only reason to sort requests
> would be if you were going to coalesce them, and I would again urge
> that you do that later.

I was thinking in do synchronous writes, from the cache. But it seems
better to do
asynchronous writes instead, and let Raidframe/other layer to do the sort. It is
 inefficient to do the sort several times.

Two questions:
1- Do Raidframe/ld/other drivers allow asynchronous writes?
2- What drivers/layers do the actual sort/elevator algorithm?

Yes I was thinking in coalesce them. I don't think that it is
incompatible with that
other layer performs the sorting


Home | Main Index | Thread Index | Old Index