speed up onspaces utility
Posted in 2006
Yunyao (Frank) Qu found onspaces slow creating a mirrored 896000KB dbspace on striped raw disks, IDS 10 UC5 on AIX 5.3. Art S. Kagel answered get faster disks, since onspaces must initialize much of the chunk and 10.00 writes a larger tblspace tblspace. Neil Truby doubted raw initialization should be slow. Art asked if a minute or two counts as slow. Unresolved.
Auto-generated by Claude from the posts below — may be imperfect; read the full thread.
Topics: Server Administration, Platform-Specific Issues, Cloud, Docker & Containers
We use raws stripped disk. IDS10 UC5, AIX 5.3
onspaces utility runing very slow, how can we speed it up?
example,
onspaces -c -d dbphylog -p /usr/informix/dev-links/int1/DBPHYLOG_INT1 -o
4 -s 896000 -m /usr/informix/dev-links/int1/MDBPHYLOG_INT1 4
Thanks,Frank
Yunyao (Frank) Qu wrote:
>
> We use raws stripped disk. IDS10 UC5, AIX 5.3
>
> onspaces utility runing very slow, how can we speed it up?
>
> example,
> onspaces -c -d dbphylog -p /usr/informix/dev-links/int1/DBPHYLOG_INT1 -o
> 4 -s 896000 -m /usr/informix/dev-links/int1/MDBPHYLOG_INT1 4
Pretty much the only useful answer is: "Get faster disks." To create a
dbspace onspaces has to initialize a large portion of the disk chunk being
presented. That just takes time dependent on how fast your disk farm is.
Art S. Kagel
Yunyao (Frank) Qu wrote:
>
> We use raws stripped disk. IDS10 UC5, AIX 5.3
>
> onspaces utility runing very slow, how can we speed it up?
>
> example,
> onspaces -c -d dbphylog -p /usr/informix/dev-links/int1/DBPHYLOG_INT1 -o
> 4 -s 896000 -m /usr/informix/dev-links/int1/MDBPHYLOG_INT1 4
Pretty much the only useful answer is: "Get faster disks." To create a
dbspace onspaces has to initialize a large portion of the disk chunk being
presented. That just takes time dependent on how fast your disk farm is.
Art S. Kagel
"Art S. Kagel" <kagel@bloomberg.net> wrote in message
news:4514133B.2090901@bloomberg.net...
> Yunyao (Frank) Qu wrote:
>>
>> We use raws stripped disk. IDS10 UC5, AIX 5.3
>>
>> onspaces utility runing very slow, how can we speed it up?
>>
>> example,
>> onspaces -c -d dbphylog -p /usr/informix/dev-links/int1/DBPHYLOG_INT1 -o
>> 4 -s 896000 -m /usr/informix/dev-links/int1/MDBPHYLOG_INT1 4>
> Pretty much the only useful answer is: "Get faster disks." To create a
> dbspace onspaces has to initialize a large portion of the disk chunk being
> presented. That just takes time dependent on how fast your disk farm is.
Do you think so? Initialisation of raw disk should be more-or-less
instantaneous, because IDS doesn't have to write all the binary zeros that
it would to a cooked chunk.
But even with a cooked chunk you'd not expect 900m to take more than a
minute or two ...
Neil Truby wrote:
> "Art S. Kagel" <kagel@bloomberg.net> wrote in message
> news:4514133B.2090901@bloomberg.net...
>
>>Yunyao (Frank) Qu wrote:
>>
>>>We use raws stripped disk. IDS10 UC5, AIX 5.3
>>>
>>>onspaces utility runing very slow, how can we speed it up?
>>>
>>>example,
>>>onspaces -c -d dbphylog -p /usr/informix/dev-links/int1/DBPHYLOG_INT1 -o
>>>4 -s 896000 -m /usr/informix/dev-links/int1/MDBPHYLOG_INT1 4>>
>>Pretty much the only useful answer is: "Get faster disks." To create a
>>dbspace onspaces has to initialize a large portion of the disk chunk being
>>presented. That just takes time dependent on how fast your disk farm is.
>
>
> Do you think so? Initialisation of raw disk should be more-or-less
> instantaneous, because IDS doesn't have to write all the binary zeros that
> it would to a cooked chunk.
>
> But even with a cooked chunk you'd not expect 900m to take more than a
> minute or two ...
Frank doesn't say "it's taking hours". He said it's 'slow'. Is a minute or
two 'slow'?
Besides for 10.00 there's a larger default tablespace tablespace to write
out in the first chunk of a new dbspace.
Art S. Kagel