Re: [Xen-devel] [PATCH 1/3] qemu-xen-traditional: use O_DIRECT to open disk images for IDE

[Joseph Glanville <joseph.glanville@xxxxxxxxxxxxxx>]

On 28 March 2012 04:22, Ian Jackson <Ian.Jackson@xxxxxxxxxxxxx> wrote:
> Stefano Stabellini writes ("RE: [Xen-devel] [PATCH 1/3] qemu-xen-traditional: 
> use O_DIRECT to open disk images for IDE"):
>> On Tue, 27 Mar 2012, Zhang, Yang Z wrote:
>> > Doesn't cache mode have better performance than NOCACHE?
>>
>> Actually you are correct. I think that this patch should be dropped from
>> the series. Of course we need O_DIRECT for QDISK otherwise we do loose
>> correctness but considering that IDE should only be used during
>> installation it can stay as it is.
>
> I don't think this assumption about IDE is correct.  To say that "IDE
> should only be used during installation" is not an excuse for
> providing an IDE controller which violates the usual correctness
> rules.
>
> Ian.
>
> _______________________________________________
> Xen-devel mailing list
> Xen-devel@xxxxxxxxxxxxx
> http://lists.xen.org/xen-devel

I would highly recommend using direct IO where virtualized disks are
concerned, regardless of correctness.
At the very least based on the principle of "least surprise" noone
likes to find out their data was corrupted due to caches they didn't
know about and didn't configure.
Over the years this has been well documented in the Xen archives and
old wiki as unwanted and best avoided behavior.

This is the driving force behind persuing qdisk over blkback+loop
until loop attains O_DIRECT support upstream.

The other option is to document that this is the case and recommend
users always use the phy:/ handler with proper block devices.
Though IMHO this is not desired and it would be preffered to indicate
that use of file:/ will result in performance degradation.

Joseph.

-- 
Founder | Director | VP Research
Orion Virtualisation Solutions | www.orionvm.com.au | Phone: 1300 56
99 52 | Mobile: 0428 754 846

_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxx
http://lists.xen.org/xen-devel