Hello Nico

When you see the slow performance, what do you see which you define as
slow?  What exactly are you measuring?

When looking at the basics (CPU, IO, MEM) where do you expect the bottle
neck to be?

look at
prstat -va
iostat -xnczC 5
mpstat 5
vmstat 5

If the application do run better in a global zone, you are hitting a bug.
Zone based applications have the same level of access to the Kernel (system
calls), Memory and CPU, as do non-zone based applications (with the
exception of system calls in the case of a branded zone)

Network may be a different story, if you are using non-shared IP (exclusive
IP mode) for the zone, but before you look any further at networking,
establish whether you have a bottle neck on CPU, memory or IO.

If you do think Network is the bottle neck, try a shared-IP zone.  And
eliminate other things on the network as being the cause of the problems.

Your post on the OpenSolaris forums suggest that you are running Some
build/version of Open Solaris, though you did not specify this.  Get updated
to the latest version, and in the case of Solaris, get your patches up to
date as there are fixes for various network errors, including packet loss.
In particular look at the driver patches and NIC firmware.




On Fri, Apr 10, 2009 at 4:38 PM, Nico Sabbi <[email protected]> wrote:

> On Friday 10 April 2009 16:19:28 Mike DeMarco wrote:
> > You should only see a very little performace hit when using zones.
> > Maybe 1 or 2 %. Are you using exclusive NIC to move the TCP stack
> > down into the zone? Have you tuned the zone for mysql?
> > What disk do you have the zone OS on?
> > Do you have enough memory and swap?
> >
>
> update, there's some connectivity problem: we have packet losses in
> the range 4-20%. The server is using  4 aggregated gb ethernet cards.
> I was thinking to detach each of the 4 cables in turn to identify the
> possibly faulty connection. Is it a good idea? Is there any command
> I can use to identify the broken links?
>
> [r...@nodo1: ~ ]$ dladm show-link
> bge0            type: non-vlan  mtu: 1500       device: bge0
> bge1            type: non-vlan  mtu: 1500       device: bge1
> e1000g0         type: non-vlan  mtu: 1500       device: e1000g0
> e1000g1         type: non-vlan  mtu: 1500       device: e1000g1
> e1000g2         type: non-vlan  mtu: 1500       device: e1000g2
> e1000g3         type: non-vlan  mtu: 1500       device: e1000g3
> aggr1           type: non-vlan  mtu: 1500       aggregation: key 1
>
> [r...@nodo1: ~ ]$ dladm show-aggr
> key: 1 (0x0001) policy: L4      address: 0:e:c:7e:1f:a4 (auto)
>           device       address                 speed           duplex
> link    state
>           e1000g0      0:e:c:7e:1f:a4    1000  Mbps    full    up
> attached
>           e1000g1      0:e:c:7e:1f:a5    1000  Mbps    full    up
> attached
>           e1000g2      0:e:c:7e:1f:a6    1000  Mbps    full    up
> attached
>           bge1         0:9:3d:12:67:dd   1000  Mbps    full    up
> attached
>
>
> _______________________________________________
> opensolaris-discuss mailing list
> [email protected]
>



-- 
Any sufficiently advanced technology is indistinguishable from magic.
   Arthur C. Clarke

My blog: http://initialprogramload.blogspot.com
_______________________________________________
opensolaris-discuss mailing list
[email protected]

Reply via email to