Translating with DrWatson… this can take a few seconds the first time.
This is a genuine, complex translation. DrWatson protects commands, error codes, and log output while naturally translating the surrounding text. It’s translated once and saved.
A user reported that dbexport of a 180 GB database suddenly became much slower after a weekend bulk insert of invoices (only ~57 MB of new data), and asked whether missing/stale UPDATE STATISTICS was to blame (high-mode stats weren't possible due to limited temp space). Replies noted dbexport reads tables sequentially, so statistics should be largely irrelevant; one user claimed stats helped on small databases, another suggested skipping big tables with --no-data-tables. Fernando Nunes advised checking the session's wait state (I/O, CPU, netnorm/client wait), the client machine and destination filesystem, and raising FET_BUF_SIZE (e.g. 256000) to enlarge the client/server buffer. No confirmed cause or resolution is recorded.
Auto-generated by DrWatson from the posts below — may be imperfect; read the full thread.
Dear colleagues:
Since Monday I have noticed that the dbexport of my database takes much longer
than usual. Last Saturday a process of massive insertion in the same has been
executed, through the generation of invoices.
Can the update statistics influence the slowness of the dbexport ?. I can not
do an update statistics high because the temporary chunks I have do not reach.
How else can the export process be accelerated?
Of course, I appreciate your attention.
dbexport supposedly unloads the tables by reading sequential. Update
statistics should be irrelevant in this case. What about the amount of data
produced by that massive inserts?
Hi Tanja!
The total data inserted last Saturday is not much, they are 57 MB.
The base has 180 GB.
What amazes me is that as of Saturday the slowness began to be noticed.
↪ replying to GUSTAVO ECHENIQUE
Jack Parker — — source: IIUG Forums & Mailing Lists
dbexport? Do you mean dbimport?
j.
> On Apr 25, 2018, at 8:55 AM, GUSTAVO ECHENIQUE
<gustavo.echenique@cemdo.com.ar> wrote:
>
> Hi Tanja!
>
> The total data inserted last Saturday is not much, they are 57 MB.
>
> The base has 180 GB.
>
> What amazes me is that as of Saturday the slowness began to be noticed.
>
>
>
*******************************************************************************
> Forum Note: Use "Reply" to post a response in the discussion forum.
>
Jack Parker — — source: IIUG Forums & Mailing Lists
You confused me with the insert.
j.
> On Apr 25, 2018, at 9:20 AM, GUSTAVO ECHENIQUE
<gustavo.echenique@cemdo.com.ar> wrote:
>
> Hi Jack! No, I mean dbexport, for data export.
>
>
>
*******************************************************************************
> Forum Note: Use "Reply" to post a response in the discussion forum.
>
↪ replying to GUSTAVO ECHENIQUE
Ivan Zavis — — source: IIUG Forums & Mailing Lists
Update statistics have positive influence on speed of dbexport, definitely.
I have small databases, and I do update statics medium (on database =
all tables), and after that, dbexport is faster.
Maybe is matter update statistics on systables ...
Another acceleration is not export all (big) tables every day, if some
tables is not very important, with parameter:
dbexport database -no-data-tables=some_arhive_table, some_log_table (if
you have newest informix version ...)
On 25.04.2018 13:33, GUSTAVO ECHENIQUE wrote:
> Dear colleagues:
>
> Since Monday I have noticed that the dbexport of my database takes much
longer
> than usual. Last Saturday a process of massive insertion in the same has been
> executed, through the generation of invoices.
>
> Can the update statistics influence the slowness of the dbexport ?. I can not
> do an update statistics high because the temporary chunks I have do not
reach.
>
> How else can the export process be accelerated?
>
> Of course, I appreciate your attention.
>
>
>
*******************************************************************************
> Forum Note: Use "Reply" to post a response in the discussion forum.
>
I honestly can't imagine a good reason why UPDATE STATS would fasten a
dbexport (Obnoxio would probably disagree with me...).
The only possible explanation for that, would be that if your databases are
really small and you do the export immediately after the stats, some of the
data will be in memory...
As for the original question, I would try to see what's the status of the
database session doing the export... waiting for I/O? Waiting on client
(netnorm?) and also the general machine status.
The bottleneck could be on the engine side (all the usual ones like I/O,
CPU etc.) or on the client side (dbexport command) if there is not enough
CPU on the machine, or if the destination filesystem is slow.
The session stat will give us leads on what's wrong. Apart from that, if
you have a version about 11.70 (can't remember the fixpack) you can define
FET_BUF_SIZE=256000 (for example) to increment the default size for the
communication buffer between client and server. The default is I believe
4096... you may need to try different values to find the optimum one for
your system. This will give you a bit more performance when there is no
evident bottleneck like CPU or I/O...
Regards.
On Wed, Apr 25, 2018 at 2:34 PM, Ivan Zavis <ivan.zavis@mi-system.co.rs>
wrote:
> Update statistics have positive influence on speed of dbexport,> definitely.
>
> I have small databases, and I do update statics medium (on database =
> all tables), and after that, dbexport is faster.
> Maybe is matter update statistics on systables ...
>
> Another acceleration is not export all (big) tables every day, if some
> tables is not very important, with parameter:
> dbexport database -no-data-tables=some_arhive_table, some_log_table (if
> you have newest informix version ...)
>
> On 25.04.2018 13:33, GUSTAVO ECHENIQUE wrote:
> > Dear colleagues:
> >
> > Since Monday I have noticed that the dbexport of my database takes much
> longer
> > than usual. Last Saturday a process of massive insertion in the same has
> been
> > executed, through the generation of invoices.
> >
> > Can the update statistics influence the slowness of the dbexport ?. I
> can
> not
> > do an update statistics high because the temporary chunks I have do not
> reach.
> >
> > How else can the export process be accelerated?
> >
> > Of course, I appreciate your attention.
> >
> >
> >
>
*******************************************************************************
>
> > Forum Note: Use "Reply" to post a response in the discussion forum.
> >
>
>
>
*******************************************************************************
>
> Forum Note: Use "Reply" to post a response in the discussion forum.
>
>
--
Fernando Nunes
Portugal
http://informix-technology.blogspot.com
My email works... but I don't check it frequently...
We use strictly necessary cookies to make this site work. With your
consent we’d also use optional cookies for analytics and marketing. You can accept all,
reject all, or choose. Read our Cookie Policy.