Hi, I'm considering to deploy my paging space in a SAN product, say,
HDS AMS1000. I know it's a safer decision to use rootvg but I think it
worths a discussion.

It's a olap application so the paging is major of big data blocks.

Pls share your opinions and pls share the experience using a paging on
SAN if you have...

Thank you.

[ Post a follow-up to this message ]

On 20 Mar, 13:11, "wyh...@gmail.com" <wyh...@gmail.com> wrote:
> Hi, I'm considering to deploy my paging space in a SAN product, say,
> HDS AMS1000. I know it's a safer decision to use rootvg but I think it
> worths a discussion.
>
> It's a olap application so the paging is major of big data blocks.
>
> Pls share your opinions and pls share the experience using a paging on
> SAN if you have...
>
> Thank you.

We use SAN for our paging (but keep hd6 on rootvg of course which are
non-san disks).
I cant see why it would be safer to have it all on rootvg really?
We dont have any problems with it being on the SAN.
Of course you have to make sure you have good SAN performance, but
then you'd want that anyway.
Regards,
Scott

[ Post a follow-up to this message ]

On Mar 20, 9:11 am, "wyh...@gmail.com" <wyh...@gmail.com> wrote:
> Hi, I'm considering to deploy my paging space in a SAN product, say,
> HDS AMS1000. I know it's a safer decision to use rootvg but I think it
> worths a discussion.
>
> It's a olap application so the paging is major of big data blocks.
>
> Pls share your opinions and pls share the experience using a paging on
> SAN if you have...
>
> Thank you.

Personally, I don't like paging, or booting from SAN, as it adds
additional
layers of complexity, and additional equipment that MUST be working
error
free, in order to have an operational system.
(HBA's, link cables, switches, fabrics, trunking, array firmware....
etc)

In this particular case, I would worry more about WHY I'm paging, than
where to place the paging space on disk. If you're paging, application
performance is going to suffer in either case.

[ Post a follow-up to this message ]

On Mar 20, 10:19 pm, "scott" <scott_doyl...@johnlewis.co.uk> wrote:
> On 20 Mar, 13:11, "wyh...@gmail.com" <wyh...@gmail.com> wrote:
> 
> 
oh, I mean it's safe for a decision maker for paging on rootvg is very
common...

I think what I want to do it what you have done.[vbcol=seagreen]
> 
> 
>
> We use SAN for our paging (but keep hd6 on rootvg of course which are
> non-san disks).
> I cant see why it would be safer to have it all on rootvg really?
> We dont have any problems with it being on the SAN.
> Of course you have to make sure you have good SAN performance, but
> then you'd want that anyway.
> Regards,
> Scott

[ Post a follow-up to this message ]

Thank you. I totally agree with you that we should do everything to
avoid paging.
but in a customer environment, where their system is not tuned  very
well and there maybe ad-hoc queries on Tbs data...
I'm considering for the worst conditions.

Here, HDS AMS1000 is involved.  Customer forwarded the note HDS
provides: they said HDLM is not supporting paging space which I highly
suspected and they suggest only a paging on rootvg solution.

>
> Personally, I don't like paging, or booting from SAN, as it adds
> additional
> layers of complexity, and additional equipment that MUST be working
> error
> free, in order to have an operational system.
> (HBA's, link cables, switches, fabrics, trunking, array firmware....
> etc)
>
> In this particular case, I would worry more about WHY I'm paging, than
> where to place the paging space on disk. If you're paging, application
> performance is going to suffer in either case.

[ Post a follow-up to this message ]

On Mar 20, 11:20 pm, aixd...@yahoo.com wrote:
> On Mar 20, 9:11 am, "wyh...@gmail.com" <wyh...@gmail.com> wrote:
> 
> 
> 
in a situation where not well tuned ad-hoc queries on Tbs data, or
something similar you cannot control...
I know either is a disaster when we trap into a large paging situation
but I just want the one but the worst.
[vbcol=seagreen]
> 
>
> Personally, I don't like paging, or booting from SAN, as it adds
> additional
> layers of complexity, and additional equipment that MUST be working
> error
> free, in order to have an operational system.
> (HBA's, link cables, switches, fabrics, trunking, array firmware....
> etc)
>
> In this particular case, I would worry more about WHY I'm paging, than
> where to place the paging space on disk. If you're paging, application
> performance is going to suffer in either case.

[ Post a follow-up to this message ]

scott wrote:
> We use SAN for our paging (but keep hd6 on rootvg of course which are
> non-san disks).
> I cant see why it would be safer to have it all on rootvg really?

Well, simply because you are relying on an external path that may
include a fiber HBA, cables, a switch and an external storage unit. More
potential points of failure.

I'm not violently disagreeing, since if your applications all reside on
all external disk my whole argument would be pointless. Furthermore, if
you do it right you will see a performance benefit over internal SCSI disk.

Even so, I'm not convinced about the external route and see no reason
why my internal SCSI disks can't do the job. I'd rather ensure that my
system has enough memory (to avoid much paging), tune that and stick to
a conservative approach that maximises uptime.

cheers
Niel

[ Post a follow-up to this message ]

scott wrote:
> We use SAN for our paging (but keep hd6 on rootvg of course which are
> non-san disks).

Do you think that somehow this will protect you from an outage if one of
your other paging spaces became unavailable?

Niel

[ Post a follow-up to this message ]

On Tue, 20 Mar 2007 06:11:31 -0700, wyhang@gmail.com wrote:

> Hi, I'm considering to deploy my paging space in a SAN product, say,
> HDS AMS1000. I know it's a safer decision to use rootvg but I think it
> worths a discussion.
>
> It's a olap application so the paging is major of big data blocks.
>
> Pls share your opinions and pls share the experience using a paging on
> SAN if you have...
>
> Thank you.

I hate to be the voice of doom and gloom, but I have seen infrequent
scenarios where a hiccup of some sort in the SAN while AIX was servicing a
page fault to/from a SAN-based paging space started the O/S on a slow
spiral to oblivion or crashed it.  I've seen that with McData and Cisco
switches in the middle but always with EMC disk in the back end, so it may
have more to do with the EMC PowerPath software driver stack than it does
anything else...

-Chris

[ Post a follow-up to this message ]

On Mar 20, 12:21 pm, "Hardy" <wyh...@gmail.com> wrote:
> On Mar 20, 11:20 pm, aixd...@yahoo.com wrote:> On Mar 20, 9:11 am, "wyh...
@gmail.com" <wyh...@gmail.com> wrote:
> 
> 
> 
>
> in a situation where not well tuned ad-hoc queries on Tbs data, or
> something similar you cannot control...
> I know either is a disaster when we trap into a large paging situation
> but I just want the one but the worst.
>
>
>
>
> 
> 
> 
>
> - Show quoted text -

Hi,
I am using SAN for paging on one of my larger AIX DB servers.  I do
have hd6 of course, but all my additional paging space is on SAN.  I
am running 10 Oracle apps databases off this particular server.  I
have been doing this for almost a year with no issues.   The SAN is
EMC symetrics, not power path.  I will agree with a previous poster
EMC power path software has caused me issues in the past, using EMC
Clarion on AIX.

[ Post a follow-up to this message ]