[pve-devel] Default cache mode for VM hard drives

[Cesar Peschiera]

I guess that i should talk more clear:

The recommendation of use VMs with "noop" as i/o scheduler (noop = no
optimized), is for a storage that have his own filesystem that is in
optimized mode. This is due to that the VM don't know that in the upper
layers exist another layer of storage with a filesystem that are working in
a optimized mode.

Working with this mode, we  get these advantages:

1) As the processor of the VM not need solve the pending tasks of reads and
writes of block of disks, then we get lower consumption of processor.

2) There will be no contradictions between the i/o schedulers nor "LOSS OF
DATA", because the i/o scheduler of the VM not know that in the upper layers
exists another I/O scheduler that is performing his work of optimization
correctly, and this last one, must also handle other requests of writes and
reads of block of disk that the VM does not know, and finally as the VM with
his own i/o scheduler is demanding a time limit for reads and writes of
blocks of disk, that if it are not satisfied, will occur a series of
unfortunate events:

A) Lose of information in writes of blocks of disk .... :-(
B) Lose of information in reads of blocks of disk .... :-(

In both cases, is very likely that the VM to crash.

Best regards
Cesar

----- Original Message ----- 
From: "Alexandre DERUMIER" <aderumier at odiso.com>
To: "Cesar Peschiera" <brain at click.com.py>
Cc: "datanom.net" <mir at datanom.net>; <pve-devel at pve.proxmox.com>
Sent: Friday, November 21, 2014 4:25 PM
Subject: Re: [pve-devel] Default cache mode for VM hard drives


>>Also, another recommendation of IBM (redbook) to gain speed of disk access
>>in the VMs consist in use LVM as device block and not files of images of
>>disk

Well, this really depend of storage implementation of iscsi or nfs.

I have netapp storage with nfs, and I can reach same speed (iops or
bandwith) than iscsi with zfs ;)


----- Mail original ----- 

De: "Cesar Peschiera" <brain at click.com.py>
À: "datanom.net" <mir at datanom.net>, pve-devel at pve.proxmox.com
Envoyé: Vendredi 21 Novembre 2014 17:34:28
Objet: Re: [pve-devel] Default cache mode for VM hard drives

No only for DRBD, for any VM that dependent of a storage with a filesystem.

And the recommendation is of IBM (redbook)

Also, another recommendation of IBM (redbook) to gain speed of disk access
in the VMs consist in use LVM as device block and not files of images of
disk (for have minus layers intermediate between the VM and the data of
disk, for example a file system that have files as "raw" , "qcow2" , "vmdk"
, or some other format of file of image of disk)

----- Original Message ----- 
From: "datanom.net" <mir at datanom.net>
To: <pve-devel at pve.proxmox.com>
Sent: Thursday, November 20, 2014 6:58 AM
Subject: Re: [pve-devel] Default cache mode for VM hard drives


> On 2014-11-20 10:29, Cesar Peschiera wrote:
>>
>> I forgot to ask you if you have configured the I/O Scheduler as
>> "deadline" in PVE, and "noop" for your Linux VMs, this combination is
>> very important.
>>
> Is this a general recommendation or only for DRBD?
>
> -- 
> Hilsen/Regards
> Michael Rasmussen
>
> Get my public GnuPG keys:
> michael <at> rasmussen <dot> cc
> http://pgp.mit.edu:11371/pks/lookup?op=get&search=0xD3C9A00E
> mir <at> datanom <dot> net
> http://pgp.mit.edu:11371/pks/lookup?op=get&search=0xE501F51C
> mir <at> miras <dot> org
> http://pgp.mit.edu:11371/pks/lookup?op=get&search=0xE3E80917
> -------------------------------------------------------------- 
>
> ---- 
>
> This mail was virus scanned and spam checked before delivery.
> This mail is also DKIM signed. See header dkim-signature.
>
> _______________________________________________
> pve-devel mailing list
> pve-devel at pve.proxmox.com
> http://pve.proxmox.com/cgi-bin/mailman/listinfo/pve-devel
>

_______________________________________________
pve-devel mailing list
pve-devel at pve.proxmox.com
http://pve.proxmox.com/cgi-bin/mailman/listinfo/pve-devel