Mailing List Archive

Is the url in your signature the website? It looks a beautiful UI design.
And you can try this method:
https://metacpan.org/pod/ModPerl::Registry

Thanks.

On Sun, Feb 7, 2021, at 8:59 AM, Steven Haigh wrote:
> Hi all,
>
> So for many years I've been slack and writing perl scripts to do various things - but never needed more than the normal apache +ExecCGI and Template Toolkit.
>
> One of my sites has become a bit more popular, so I'd like to spend a bit of time on performance. Currently, I'm seeing ~300-400ms of what I believe to be execution time of the script loading, running, and then blatting its output to STDOUT and the browser can go do its thing.
>
> I believe most of the delay would be to do with loading perl, its modules etc etc
>
> I know that the current trend would be to re-write the entire site in a more modern, daemon based solution - and I started down the Mojolicious path - but the amount of re-writing to save 1/3rd of a second seems to be excessive
>
> Would I be correct in thinking that mod_perl would help in this case?
>
> I did try a basic test, but I have a 'use functions' in all my scripts that loads a .pm with some global vars and a lot of common subs - and for whatever reason (can't find anything on Google as to why), none of the subs are recognised in the main script when loaded via ModPerl::PerlRun.
>
> So throwing it out to the list - am I on the right track? wasting my time? or just a simple mistake?
>
> --
> Steven Haigh ???? netwiz@crc.id.au ???? https://www.crc.id.au

As welsey said – try Registry, that was the standard way of using mod_perl to cache perl in the server – but your problem might be due to the note in PerlRun…

https://perl.apache.org/docs/2.0/api/ModPerl/PerlRun.html#Description
META: document that for now we don't chdir() into the script's dir, because it affects the whole process under threads. ModPerl::PerlRunPrefork<https://perl.apache.org/docs/2.0/api/ModPerl/PerlRunPrefork.html> should be used by those who run only under prefork MPM.
{tbh most people don’t use mod perl under threads anyway as there isn’t really a gain from using them}

It suggests you use ModPerl/PerlRunPrefork – as this does an additional step to cd to the script directory – which might be your issue….

From: Steven Haigh <netwiz@crc.id.au>
Sent: 07 February 2021 01:00
To: modperl@perl.apache.org
Subject: Moving ExecCGI to mod_perl - performance and custom 'modules' [EXT]

Hi all,

So for many years I've been slack and writing perl scripts to do various things - but never needed more than the normal apache +ExecCGI and Template Toolkit.

One of my sites has become a bit more popular, so I'd like to spend a bit of time on performance. Currently, I'm seeing ~300-400ms of what I believe to be execution time of the script loading, running, and then blatting its output to STDOUT and the browser can go do its thing.

I believe most of the delay would be to do with loading perl, its modules etc etc

I know that the current trend would be to re-write the entire site in a more modern, daemon based solution - and I started down the Mojolicious path - but the amount of re-writing to save 1/3rd of a second seems to be excessive

Would I be correct in thinking that mod_perl would help in this case?

I did try a basic test, but I have a 'use functions' in all my scripts that loads a .pm with some global vars and a lot of common subs - and for whatever reason (can't find anything on Google as to why), none of the subs are recognised in the main script when loaded via ModPerl::PerlRun.

So throwing it out to the list - am I on the right track? wasting my time? or just a simple mistake?

--
Steven Haigh ???? netwiz@crc.id.au<mailto:netwiz@crc.id.au> ???? https://www.crc.id.au [crc.id.au]<https://urldefense.proofpoint.com/v2/url?u=https-3A__www.crc.id.au_&d=DwMFaQ&c=D7ByGjS34AllFgecYw0iC6Zq7qlm8uclZFI0SqQnqBo&r=oH2yp0ge1ecj4oDX0XM7vQ&m=bosoTbkecbnrPukObNK-5Duc1p3JTllIM7_FHhBYKW4&s=vQDi0ezyEZDOz86GVraerPdT76UjN2in3UdPh8fglRM&e=>



--
The Wellcome Sanger Institute is operated by Genome Research
Limited, a charity registered in England with number 1021457 and a
company registered in England with number 2742969, whose registered
office is 215 Euston Road, London, NW1 2BE.

Interestingly, I did get things working with ModPerl::PerlRegistry.

What I couldn't find *anywhere* is that the data I was loading in
Template Toolkit was included in the file in the __DATA__ area - which
causes mod_perl to fall over!

The only way I managed to find this was the following error in the
*system* /var/log/httpd/error_log (didn't show up in the vhost
error_log!):
readline() on unopened filehandle DATA at
/usr/lib64/perl5/vendor_perl/Template/Provider.pm line 638.

Took me a LONG time to find a vague post that reading in lines from
<DATA> kills mod_perl. Not sure why - but I stripped all the templates
out and put them in a file instead and re-wrote that bit of code, and
things started working.

I had to fix a few lib path issues, but after getting my head around
that, most things seem to work as before - however I don't notice much
of an improvement in execution times, I do see this improvement using
'ab -n 100 -c32':

Apache + ExecCGI: Requests per second: 13.50 [#/sec] (mean)
Apache + mod_perl: Requests per second: 59.81 [#/sec] (mean)

This is obviously a good thing.

I haven't gotten into the preload or DBI sharing yet - as that'll end
up needing a bit of a rewrite of code to take advantage of. I'd be open
to suggestions here from those who have done it in the past to save me
going down some dead ends :D

--
Steven Haigh

???? netwiz@crc.id.au <mailto:netwiz@crc.id.au>
???? https://www.crc.id.au <https://www.crc.id.au/>

On Sun, Feb 7, 2021 at 12:49, James Smith <js5@sanger.ac.uk> wrote:
> As welsey said – try Registry, that was the standard way of using
> mod_perl to cache perl in the server – but your problem might be
> due to the note in PerlRun…
>
> <https://perl.apache.org/docs/2.0/api/ModPerl/PerlRun.html#Description>
> META: document that for now we don't chdir() into the script's dir,
> because it affects the whole process under threads.
> ModPerl::PerlRunPrefork
> <https://perl.apache.org/docs/2.0/api/ModPerl/PerlRunPrefork.html>
> should be used by those who run only under prefork MPM.
> {tbh most people don’t use mod perl under threads anyway as there
> isn’t really a gain from using them}
>
> It suggests you use ModPerl/PerlRunPrefork – as this does an
> additional step to cd to the script directory – which might be your
> issue….
>
>
>
> *From:*Steven Haigh <netwiz@crc.id.au>
> *Sent:* 07 February 2021 01:00
> *To:* modperl@perl.apache.org
> *Subject:* Moving ExecCGI to mod_perl - performance and custom
> 'modules' [EXT]
>
>
>
> Hi all,
>
>
>
> So for many years I've been slack and writing perl scripts to do
> various things - but never needed more than the normal apache
> +ExecCGI and Template Toolkit.
>
>
>
> One of my sites has become a bit more popular, so I'd like to spend a
> bit of time on performance. Currently, I'm seeing ~300-400ms of what
> I believe to be execution time of the script loading, running, and
> then blatting its output to STDOUT and the browser can go do its
> thing.
>
>
>
> I believe most of the delay would be to do with loading perl, its
> modules etc etc
>
>
>
> I know that the current trend would be to re-write the entire site in
> a more modern, daemon based solution - and I started down the
> Mojolicious path - but the amount of re-writing to save 1/3rd of a
> second seems to be excessive
>
>
>
> Would I be correct in thinking that mod_perl would help in this case?
>
>
>
> I did try a basic test, but I have a 'use functions' in all my
> scripts that loads a .pm with some global vars and a lot of common
> subs - and for whatever reason (can't find anything on Google as to
> why), none of the subs are recognised in the main script when loaded
> via ModPerl::PerlRun.
>
>
>
> So throwing it out to the list - am I on the right track? wasting my
> time? or just a simple mistake?
>
>
>
> --
>
> Steven Haigh ????netwiz@crc.id.au
> <mailto:netwiz@crc.id.au>????https://www.crc.id.au [crc.id.au]
> <https://urldefense.proofpoint.com/v2/url?u=https-3A__www.crc.id.au_&d=DwMFaQ&c=D7ByGjS34AllFgecYw0iC6Zq7qlm8uclZFI0SqQnqBo&r=oH2yp0ge1ecj4oDX0XM7vQ&m=bosoTbkecbnrPukObNK-5Duc1p3JTllIM7_FHhBYKW4&s=vQDi0ezyEZDOz86GVraerPdT76UjN2in3UdPh8fglRM&e=>
>
> -- The Wellcome Sanger Institute is operated by Genome Research
> Limited, a charity registered in England with number 1021457 and a
> company registered in England with number 2742969, whose registered
> office is 215 Euston Road, London, NW1 2BE.

In fact, I just realised that 'ab' test is rather restrictive.... So
here's a bit more of an extended test:

# ab -k -n 1000 -c 32

Apache + ExecCGI:
Requests per second: 14.26 [#/sec] (mean)
Time per request: 2244.181 [ms] (mean)
Time per request: 70.131 [ms] (mean, across all concurrent
requests)

Apache + mod_perl (ModPerl::PerlRegistry):
Requests per second: 132.14 [#/sec] (mean)
Time per request: 242.175 [ms] (mean)
Time per request: 7.568 [ms] (mean, across all concurrent
requests)

Interestingly, without Keepalives, the story is much the same:

# ab -n 1000 -c 32

Apache + ExecCGI:
Requests per second: 14.15 [#/sec] (mean)
Time per request: 2260.875 [ms] (mean)
Time per request: 70.652 [ms] (mean, across all concurrent
requests)

Apache + mod_perl (ModPerl::PerlRegistry):
Requests per second: 154.48 [#/sec] (mean)
Time per request: 207.140 [ms] (mean)
Time per request: 6.473 [ms] (mean, across all concurrent
requests)

Running some benchmarks across various parts of my site made me realise
I also had some RewriteRules in the apache config that still had
H=cgi-script - changed those to H=perl-script and saw similar
improvements:

ExecCGI - Requests per second: 11.84 [#/sec] (mean)
mod_perl - Requests per second: 130.97 [#/sec] (mean)

That's quite some gains for a days work.

--
Steven Haigh

???? netwiz@crc.id.au <mailto:netwiz@crc.id.au>
???? https://www.crc.id.au <https://www.crc.id.au/>

On Sun, Feb 7, 2021 at 23:58, Steven Haigh <netwiz@crc.id.au> wrote:
> Interestingly, I did get things working with ModPerl::PerlRegistry.
>
> What I couldn't find *anywhere* is that the data I was loading in
> Template Toolkit was included in the file in the __DATA__ area -
> which causes mod_perl to fall over!
>
> The only way I managed to find this was the following error in the
> *system* /var/log/httpd/error_log (didn't show up in the vhost
> error_log!):
> readline() on unopened filehandle DATA at
> /usr/lib64/perl5/vendor_perl/Template/Provider.pm line 638.
>
> Took me a LONG time to find a vague post that reading in lines from
> <DATA> kills mod_perl. Not sure why - but I stripped all the
> templates out and put them in a file instead and re-wrote that bit of
> code, and things started working.
>
> I had to fix a few lib path issues, but after getting my head around
> that, most things seem to work as before - however I don't notice
> much of an improvement in execution times, I do see this improvement
> using 'ab -n 100 -c32':
>
> Apache + ExecCGI: Requests per second: 13.50 [#/sec] (mean)
> Apache + mod_perl: Requests per second: 59.81 [#/sec] (mean)
>
> This is obviously a good thing.
>
> I haven't gotten into the preload or DBI sharing yet - as that'll end
> up needing a bit of a rewrite of code to take advantage of. I'd be
> open to suggestions here from those who have done it in the past to
> save me going down some dead ends :D
>
> --
> Steven Haigh
>
> ???? netwiz@crc.id.au <mailto:netwiz@crc.id.au>
> ???? https://www.crc.id.au <https://www.crc.id.au/>
>
> On Sun, Feb 7, 2021 at 12:49, James Smith <js5@sanger.acuk> wrote:
>> As welsey said – try Registry, that was the standard way of using
>> mod_perl to cache perl in the server – but your problem might be
>> due to the note in PerlRun…
>>
>> <https://perl.apache.org/docs/2.0/api/ModPerl/PerlRun.html#Description>
>> META: document that for now we don't chdir() into the script's dir,
>> because it affects the whole process under threads.
>> ModPerl::PerlRunPrefork
>> <https://perl.apache.org/docs/20/api/ModPerl/PerlRunPrefork.html>
>> should be used by those who run only under prefork MPM.
>> {tbh most people don’t use mod perl under threads anyway as there
>> isn’t really a gain from using them}
>>
>> It suggests you use ModPerl/PerlRunPrefork – as this does an
>> additional step to cd to the script directory – which might be
>> your issue….
>>
>>
>>
>> *From:*Steven Haigh <netwiz@crc.id.au>
>> *Sent:* 07 February 2021 01:00
>> *To:* modperl@perl.apache.org
>> *Subject:* Moving ExecCGI to mod_perl - performance and custom
>> 'modules' [EXT]
>>
>>
>>
>> Hi all,
>>
>>
>>
>> So for many years I've been slack and writing perl scripts to do
>> various things - but never needed more than the normal apache
>> +ExecCGI and Template Toolkit.
>>
>>
>>
>> One of my sites has become a bit more popular, so I'd like to spend
>> a bit of time on performance. Currently, I'm seeing ~300-400ms of
>> what I believe to be execution time of the script loading, running,
>> and then blatting its output to STDOUT and the browser can go do its
>> thing.
>>
>>
>>
>> I believe most of the delay would be to do with loading perl, its
>> modules etc etc
>>
>>
>>
>> I know that the current trend would be to re-write the entire site
>> in a more modern, daemon based solution - and I started down the
>> Mojolicious path - but the amount of re-writing to save 1/3rd of a
>> second seems to be excessive
>>
>>
>>
>> Would I be correct in thinking that mod_perl would help in this case?
>>
>>
>>
>> I did try a basic test, but I have a 'use functions' in all my
>> scripts that loads a .pm with some global vars and a lot of common
>> subs - and for whatever reason (can't find anything on Google as to
>> why), none of the subs are recognised in the main script when loaded
>> via ModPerl::PerlRun.
>>
>>
>>
>> So throwing it out to the list - am I on the right track? wasting my
>> time? or just a simple mistake?
>>
>>
>>
>> --
>>
>> Steven Haigh ????netwiz@crc.id.au
>> <mailto:netwiz@crc.id.au>????https://www.crc.id.au [crc.id.au]
>> <https://urldefense.proofpoint.com/v2/url?u=https-3A__www.crc.id.au_&d=DwMFaQ&c=D7ByGjS34AllFgecYw0iC6Zq7qlm8uclZFI0SqQnqBo&r=oH2yp0ge1ecj4oDX0XM7vQ&m=bosoTbkecbnrPukObNK-5Duc1p3JTllIM7_FHhBYKW4&s=vQDi0ezyEZDOz86GVraerPdT76UjN2in3UdPh8fglRM&e=>
>>
>> -- The Wellcome Sanger Institute is operated by Genome Research
>> Limited, a charity registered in England with number 1021457 and a
>> company registered in England with number 2742969, whose registered
>> office is 215 Euston Road, London, NW1 2BE.

If you can take time to rewrite all codes with modPerl handlers, that will improve performance a lot.

On Sun, Feb 7, 2021, at 9:14 PM, Steven Haigh wrote:
> In fact, I just realised that 'ab' test is rather restrictive.... So here's a bit more of an extended test:
>
> # ab -k -n 1000 -c 32
>
> Apache + ExecCGI:
> Requests per second: 14.26 [#/sec] (mean)
> Time per request: 2244.181 [ms] (mean)
> Time per request: 70.131 [ms] (mean, across all concurrent requests)
>
> Apache + mod_perl (ModPerl::PerlRegistry):
> Requests per second: 132.14 [#/sec] (mean)
> Time per request: 242.175 [ms] (mean)
> Time per request: 7.568 [ms] (mean, across all concurrent requests)
>
> Interestingly, without Keepalives, the story is much the same:
>
> # ab -n 1000 -c 32
>
> Apache + ExecCGI:
> Requests per second: 14.15 [#/sec] (mean)
> Time per request: 2260.875 [ms] (mean)
> Time per request: 70.652 [ms] (mean, across all concurrent requests)
>
> Apache + mod_perl (ModPerl::PerlRegistry):
> Requests per second: 154.48 [#/sec] (mean)
> Time per request: 207.140 [ms] (mean)
> Time per request: 6.473 [ms] (mean, across all concurrent requests)
>
> Running some benchmarks across various parts of my site made me realise I also had some RewriteRules in the apache config that still had H=cgi-script - changed those to H=perl-script and saw similar improvements:
>
> ExecCGI - Requests per second: 11.84 [#/sec] (mean)
> mod_perl - Requests per second: 130.97 [#/sec] (mean)
>
> That's quite some gains for a days work.
>
> --
> Steven Haigh ???? netwiz@crc.id.au ???? https://www.crc.id.au
>
> On Sun, Feb 7, 2021 at 23:58, Steven Haigh <netwiz@crc.id.au> wrote:
>> Interestingly, I did get things working with ModPerl::PerlRegistry.
>>
>> What I couldn't find *anywhere* is that the data I was loading in Template Toolkit was included in the file in the __DATA__ area - which causes mod_perl to fall over!
>>
>> The only way I managed to find this was the following error in the *system* /var/log/httpd/error_log (didn't show up in the vhost error_log!):
>> readline() on unopened filehandle DATA at /usr/lib64/perl5/vendor_perl/Template/Provider.pm line 638.
>>
>> Took me a LONG time to find a vague post that reading in lines from <DATA> kills mod_perl. Not sure why - but I stripped all the templates out and put them in a file instead and re-wrote that bit of code, and things started working.
>>
>> I had to fix a few lib path issues, but after getting my head around that, most things seem to work as before - however I don't notice much of an improvement in execution times, I do see this improvement using 'ab -n 100 -c32':
>>
>> Apache + ExecCGI: Requests per second: 13.50 [#/sec] (mean)
>> Apache + mod_perl: Requests per second: 59.81 [#/sec] (mean)
>>
>> This is obviously a good thing.
>>
>> I haven't gotten into the preload or DBI sharing yet - as that'll end up needing a bit of a rewrite of code to take advantage of. I'd be open to suggestions here from those who have done it in the past to save me going down some dead ends :D
>>
>> --
>> Steven Haigh ???? netwiz@crc.id.au ???? https://www.crc.id.au
>>
>> On Sun, Feb 7, 2021 at 12:49, James Smith <js5@sanger.acuk> wrote:
>>> As welsey said – try Registry, that was the standard way of using mod_perl to cache perl in the server – but your problem might be due to the note in PerlRun…
>>>
>>> https://perl.apache.org/docs/2.0/api/ModPerl/PerlRun.html#Description
>>> META: document that for now we don't chdir() into the script's dir, because it affects the whole process under threads. `ModPerl::PerlRunPrefork <https://perl.apache.org/docs/20/api/ModPerl/PerlRunPrefork.html>` should be used by those who run only under prefork MPM.
>>> {tbh most people don’t use mod perl under threads anyway as there isn’t really a gain from using them}
>>>
>>> It suggests you use ModPerl/PerlRunPrefork – as this does an additional step to cd to the script directory – which might be your issue….

>>>

>>> *From:* Steven Haigh <netwiz@crc.id.au>
>>> *Sent:* 07 February 2021 01:00
>>> *To:* modperl@perl.apache.org
>>> *Subject:* Moving ExecCGI to mod_perl - performance and custom 'modules' [EXT]

>>>

>>> Hi all,

>>>

>>> So for many years I've been slack and writing perl scripts to do various things - but never needed more than the normal apache +ExecCGI and Template Toolkit.

>>>

>>> One of my sites has become a bit more popular, so I'd like to spend a bit of time on performance. Currently, I'm seeing ~300-400ms of what I believe to be execution time of the script loading, running, and then blatting its output to STDOUT and the browser can go do its thing.

>>>

>>> I believe most of the delay would be to do with loading perl, its modules etc etc

>>>

>>> I know that the current trend would be to re-write the entire site in a more modern, daemon based solution - and I started down the Mojolicious path - but the amount of re-writing to save 1/3rd of a second seems to be excessive

>>>

>>> Would I be correct in thinking that mod_perl would help in this case?

>>>

>>> I did try a basic test, but I have a 'use functions' in all my scripts that loads a .pm with some global vars and a lot of common subs - and for whatever reason (can't find anything on Google as to why), none of the subs are recognised in the main script when loaded via ModPerl::PerlRun.

>>>

>>> So throwing it out to the list - am I on the right track? wasting my time? or just a simple mistake?

>>>

>>> --

>>> Steven Haigh ???? netwiz@crc.id.au ???? https://www.crc.id.au [crc.id.au] <https://urldefense.proofpoint.com/v2/url?u=https-3A__www.crc.id.au_&d=DwMFaQ&c=D7ByGjS34AllFgecYw0iC6Zq7qlm8uclZFI0SqQnqBo&r=oH2yp0ge1ecj4oDX0XM7vQ&m=bosoTbkecbnrPukObNK-5Duc1p3JTllIM7_FHhBYKW4&s=vQDi0ezyEZDOz86GVraerPdT76UjN2in3UdPh8fglRM&e=>

>>> -- The Wellcome Sanger Institute is operated by Genome Research Limited, a charity registered in England with number 1021457 and a company registered in England with number 2742969, whose registered office is 215 Euston Road, London, NW1 2BE.

On Sun, 07 Feb 2021 23:58:17 +1100
Steven Haigh <netwiz@crc.id.au> wrote:
>
> I haven't gotten into the preload or DBI sharing yet - as that'll end
> up needing a bit of a rewrite of code to take advantage of. I'd be open
> to suggestions here from those who have done it in the past to save me
> going down some dead ends :D

I use mod_perl handlers, so not sure how it mixes with PerlRegistry, but you probably want to have a look at connect_cached


--

Bien ? vous, Vincent Veyron

https://compta.libremen.com
Logiciel libre de comptabilit? g?n?rale en partie double

DBI sharing doesn't really gain you much - and can actually lead you into a whole world of pain. It isn't actually worth turning it on at all.

We use dedicated DB caching in the cases where we benefit from it as and when you need it (low level caching database)...

Although milage may vary - our problem was DB servers with 30 or 40 databases on them being connected from a number of approximately 50-100 child apaches, meant we ended up blowing up the connections to the database server really quickly...


-----Original Message-----
From: Vincent Veyron <vv.lists@wanadoo.fr>
Sent: 07 February 2021 19:06
To: Steven Haigh <netwiz@crc.id.au>
Cc: James Smith <js5@sanger.ac.uk>; modperl@perl.apache.org
Subject: Re: Moving ExecCGI to mod_perl - performance and custom 'modules' [EXT]

On Sun, 07 Feb 2021 23:58:17 +1100
Steven Haigh <netwiz@crc.id.au> wrote:
>
> I haven't gotten into the preload or DBI sharing yet - as that'll end
> up needing a bit of a rewrite of code to take advantage of. I'd be
> open to suggestions here from those who have done it in the past to
> save me going down some dead ends :D

I use mod_perl handlers, so not sure how it mixes with PerlRegistry, but you probably want to have a look at connect_cached


--

Bien ? vous, Vincent Veyron

https://urldefense.proofpoint.com/v2/url?u=https-3A__compta.libremen.com&d=DwIFAw&c=D7ByGjS34AllFgecYw0iC6Zq7qlm8uclZFI0SqQnqBo&r=oH2yp0ge1ecj4oDX0XM7vQ&m=C0OcuGbNbfxaSa8ASgV3uFXzejn7MpjIUH1aP1RbiyU&s=GPr8VuKQ3rZCzCPwggyAHdCOojK6ZThmShKk0Jb3maI&e=
Logiciel libre de comptabilit? g?n?rale en partie double



--
The Wellcome Sanger Institute is operated by Genome Research
Limited, a charity registered in England with number 1021457 and a
company registered in England with number 2742969, whose registered
office is 215 Euston Road, London, NW1 2BE.

We're in a similar situation with 20-40 databases per server. We use
PGBouncer for connection pooling.

On Sun, Feb 7, 2021 at 2:23 PM James Smith <js5@sanger.ac.uk> wrote:

> DBI sharing doesn't really gain you much - and can actually lead you into
> a whole world of pain. It isn't actually worth turning it on at all.
>
> We use dedicated DB caching in the cases where we benefit from it as and
> when you need it (low level caching database)...
>
> Although milage may vary - our problem was DB servers with 30 or 40
> databases on them being connected from a number of approximately 50-100
> child apaches, meant we ended up blowing up the connections to the database
> server really quickly...
>
>
> -----Original Message-----
> From: Vincent Veyron <vv.lists@wanadoo.fr>
> Sent: 07 February 2021 19:06
> To: Steven Haigh <netwiz@crc.id.au>
> Cc: James Smith <js5@sanger.ac.uk>; modperl@perl.apache.org
> Subject: Re: Moving ExecCGI to mod_perl - performance and custom 'modules'
> [EXT]
>
> On Sun, 07 Feb 2021 23:58:17 +1100
> Steven Haigh <netwiz@crc.id.au> wrote:
> >
> > I haven't gotten into the preload or DBI sharing yet - as that'll end
> > up needing a bit of a rewrite of code to take advantage of. I'd be
> > open to suggestions here from those who have done it in the past to
> > save me going down some dead ends :D
>
> I use mod_perl handlers, so not sure how it mixes with PerlRegistry, but
> you probably want to have a look at connect_cached
>
>
> --
>
> Bien à vous, Vincent Veyron
>
>
> https://urldefense.proofpoint.com/v2/url?u=https-3A__compta.libremen.com&d=DwIFAw&c=D7ByGjS34AllFgecYw0iC6Zq7qlm8uclZFI0SqQnqBo&r=oH2yp0ge1ecj4oDX0XM7vQ&m=C0OcuGbNbfxaSa8ASgV3uFXzejn7MpjIUH1aP1RbiyU&s=GPr8VuKQ3rZCzCPwggyAHdCOojK6ZThmShKk0Jb3maI&e=
> Logiciel libre de comptabilité générale en partie double
>
>
>
> --
> The Wellcome Sanger Institute is operated by Genome Research
> Limited, a charity registered in England with number 1021457 and a
> company registered in England with number 2742969, whose registered
> office is 215 Euston Road, London, NW1 2BE.
>


--
John Dunlap
*CTO | Lariat *

*Direct:*
*john@lariat.co <john@lariat.co>*

*Customer Service:*
877.268.6667
support@lariat.co

DBI sharing has it's own issues but in most use cases it does a pretty good
job and keeps the DBA's happy - that is very important ;)

On Sun, Feb 7, 2021 at 2:23 PM James Smith <js5@sanger.ac.uk> wrote:

> DBI sharing doesn't really gain you much - and can actually lead you into
> a whole world of pain. It isn't actually worth turning it on at all.
>
> We use dedicated DB caching in the cases where we benefit from it as and
> when you need it (low level caching database)...
>
> Although milage may vary - our problem was DB servers with 30 or 40
> databases on them being connected from a number of approximately 50-100
> child apaches, meant we ended up blowing up the connections to the database
> server really quickly...
>
>
> -----Original Message-----
> From: Vincent Veyron <vv.lists@wanadoo.fr>
> Sent: 07 February 2021 19:06
> To: Steven Haigh <netwiz@crc.id.au>
> Cc: James Smith <js5@sanger.ac.uk>; modperl@perl.apache.org
> Subject: Re: Moving ExecCGI to mod_perl - performance and custom 'modules'
> [EXT]
>
> On Sun, 07 Feb 2021 23:58:17 +1100
> Steven Haigh <netwiz@crc.id.au> wrote:
> >
> > I haven't gotten into the preload or DBI sharing yet - as that'll end
> > up needing a bit of a rewrite of code to take advantage of. I'd be
> > open to suggestions here from those who have done it in the past to
> > save me going down some dead ends :D
>
> I use mod_perl handlers, so not sure how it mixes with PerlRegistry, but
> you probably want to have a look at connect_cached
>
>
> --
>
> Bien à vous, Vincent Veyron
>
>
> https://urldefense.proofpoint.com/v2/url?u=https-3A__compta.libremen.com&d=DwIFAw&c=D7ByGjS34AllFgecYw0iC6Zq7qlm8uclZFI0SqQnqBo&r=oH2yp0ge1ecj4oDX0XM7vQ&m=C0OcuGbNbfxaSa8ASgV3uFXzejn7MpjIUH1aP1RbiyU&s=GPr8VuKQ3rZCzCPwggyAHdCOojK6ZThmShKk0Jb3maI&e=
> Logiciel libre de comptabilité générale en partie double
>
>
>
> --
> The Wellcome Sanger Institute is operated by Genome Research
> Limited, a charity registered in England with number 1021457 and a
> company registered in England with number 2742969, whose registered
> office is 215 Euston Road, London, NW1 2BE.
>

No it was one of our DBAs who told us to take if off…! It causes them all sorts of problems – the main one being it keeps connections open, which is the best way to completely foobar the database servers! We make them happy by turning it off. We’ve even had Oracle developers (sorry not people developing SQL in oracle but people developing Oracle) scratching their heads over things like this and not being able to come up with a solution…..

We’ve discovered lots of interesting side effects with connection pooling when using over a secure network – e.g. the Oracle client doesn’t like running through a firewall – as it can’t cope when the firewall drops the connection – both sides of the system the “pool” thinks the connection is open, the database thinks the connection is open, and the client just hangs….. We have seen similar things with other applications which use connection pooling and long duration database handles. We get round it with permanent MySQL connections by closing/re-opening them after 5 minutes of inactivity – hence the need to develop our own cache/pool….

From: Mithun Bhattacharya <mithnb@gmail.com>
Sent: 07 February 2021 20:36
To: James Smith <js5@sanger.ac.uk>
Cc: Vincent Veyron <vv.lists@wanadoo.fr>; Steven Haigh <netwiz@crc.id.au>; modperl@perl.apache.org
Subject: Re: Moving ExecCGI to mod_perl - performance and custom 'modules' [EXT]

DBI sharing has it's own issues but in most use cases it does a pretty good job and keeps the DBA's happy - that is very important ;)

On Sun, Feb 7, 2021 at 2:23 PM James Smith <js5@sanger.ac.uk<mailto:js5@sanger.ac.uk>> wrote:
DBI sharing doesn't really gain you much - and can actually lead you into a whole world of pain. It isn't actually worth turning it on at all.

We use dedicated DB caching in the cases where we benefit from it as and when you need it (low level caching database)...

Although milage may vary - our problem was DB servers with 30 or 40 databases on them being connected from a number of approximately 50-100 child apaches, meant we ended up blowing up the connections to the database server really quickly...


-----Original Message-----
From: Vincent Veyron <vv.lists@wanadoo.fr<mailto:vv.lists@wanadoo.fr>>
Sent: 07 February 2021 19:06
To: Steven Haigh <netwiz@crc.id.au<mailto:netwiz@crc.id.au>>
Cc: James Smith <js5@sanger.ac.uk<mailto:js5@sanger.ac.uk>>; modperl@perl.apache.org<mailto:modperl@perl.apache.org>
Subject: Re: Moving ExecCGI to mod_perl - performance and custom 'modules' [EXT]

On Sun, 07 Feb 2021 23:58:17 +1100
Steven Haigh <netwiz@crc.id.au<mailto:netwiz@crc.id.au>> wrote:
>
> I haven't gotten into the preload or DBI sharing yet - as that'll end
> up needing a bit of a rewrite of code to take advantage of. I'd be
> open to suggestions here from those who have done it in the past to
> save me going down some dead ends :D

I use mod_perl handlers, so not sure how it mixes with PerlRegistry, but you probably want to have a look at connect_cached


--

Bien à vous, Vincent Veyron

https://urldefense.proofpoint.com/v2/url?u=https-3A__compta.libremen.com&d=DwIFAw&c=D7ByGjS34AllFgecYw0iC6Zq7qlm8uclZFI0SqQnqBo&r=oH2yp0ge1ecj4oDX0XM7vQ&m=C0OcuGbNbfxaSa8ASgV3uFXzejn7MpjIUH1aP1RbiyU&s=GPr8VuKQ3rZCzCPwggyAHdCOojK6ZThmShKk0Jb3maI&e=
Logiciel libre de comptabilité générale en partie double



--
The Wellcome Sanger Institute is operated by Genome Research
Limited, a charity registered in England with number 1021457 and a
company registered in England with number 2742969, whose registered
office is 215 Euston Road, London, NW1 2BE.



--
The Wellcome Sanger Institute is operated by Genome Research
Limited, a charity registered in England with number 1021457 and a
company registered in England with number 2742969, whose registered
office is 215 Euston Road, London, NW1 2BE.

what's DBI sharing? do you mean Apache::DBI?
Does perl has Java similar DB connection pool?

Thanks.

On Mon, Feb 8, 2021, at 4:21 AM, James Smith wrote:
> DBI sharing doesn't really gain you much - and can actually lead you into a whole world of pain. It isn't actually worth turning it on at all.
>
> We use dedicated DB caching in the cases where we benefit from it as and when you need it (low level caching database)...
>
> Although milage may vary - our problem was DB servers with 30 or 40 databases on them being connected from a number of approximately 50-100 child apaches, meant we ended up blowing up the connections to the database server really quickly...
>
>
> -----Original Message-----
> From: Vincent Veyron <vv.lists@wanadoo.fr>
> Sent: 07 February 2021 19:06
> To: Steven Haigh <netwiz@crc.id.au>
> Cc: James Smith <js5@sanger.ac.uk>; modperl@perl.apache.org
> Subject: Re: Moving ExecCGI to mod_perl - performance and custom 'modules' [EXT]
>
> On Sun, 07 Feb 2021 23:58:17 +1100
> Steven Haigh <netwiz@crc.id.au> wrote:
> >
> > I haven't gotten into the preload or DBI sharing yet - as that'll end
> > up needing a bit of a rewrite of code to take advantage of. I'd be
> > open to suggestions here from those who have done it in the past to
> > save me going down some dead ends :D
>
> I use mod_perl handlers, so not sure how it mixes with PerlRegistry, but you probably want to have a look at connect_cached
>
>
> --
>
> Bien à vous, Vincent Veyron
>
> https://urldefense.proofpoint.com/v2/url?u=https-3A__compta.libremen.com&d=DwIFAw&c=D7ByGjS34AllFgecYw0iC6Zq7qlm8uclZFI0SqQnqBo&r=oH2yp0ge1ecj4oDX0XM7vQ&m=C0OcuGbNbfxaSa8ASgV3uFXzejn7MpjIUH1aP1RbiyU&s=GPr8VuKQ3rZCzCPwggyAHdCOojK6ZThmShKk0Jb3maI&e=
> Logiciel libre de comptabilité générale en partie double
>
>
>
> --
> The Wellcome Sanger Institute is operated by Genome Research
> Limited, a charity registered in England with number 1021457 and a
> company registered in England with number 2742969, whose registered
> office is 215 Euston Road, London, NW1 2BE.
>

I meant between requests in the same thread – it uses persistent connections which are bad in a number of ways {e.g. fun with locking if you use it and your script dies halfway through, keeps too many connections open and can block the database, has issues with secure set ups {firewall between web and servers for instance}

From: Wesley Peng <wesley@pengfamily.de>
Sent: 08 February 2021 00:29
To: modperl@perl.apache.org
Subject: Re: Moving ExecCGI to mod_perl - performance and custom 'modules' [EXT]

what's DBI sharing? do you mean Apache::DBI?
Does perl has Java similar DB connection pool?

Thanks.

On Mon, Feb 8, 2021, at 4:21 AM, James Smith wrote:
DBI sharing doesn't really gain you much - and can actually lead you into a whole world of pain. It isn't actually worth turning it on at all.

We use dedicated DB caching in the cases where we benefit from it as and when you need it (low level caching database)...

Although milage may vary - our problem was DB servers with 30 or 40 databases on them being connected from a number of approximately 50-100 child apaches, meant we ended up blowing up the connections to the database server really quickly...


-----Original Message-----
From: Vincent Veyron <vv.lists@wanadoo.fr<mailto:vv.lists@wanadoo.fr>>
Sent: 07 February 2021 19:06
To: Steven Haigh <netwiz@crc.id.au<mailto:netwiz@crc.id.au>>
Cc: James Smith <js5@sanger.ac.uk<mailto:js5@sanger.ac.uk>>; modperl@perl.apache.org<mailto:modperl@perl.apache.org>
Subject: Re: Moving ExecCGI to mod_perl - performance and custom 'modules' [EXT]

On Sun, 07 Feb 2021 23:58:17 +1100
Steven Haigh <netwiz@crc.id.au<mailto:netwiz@crc.id.au>> wrote:
>
> I haven't gotten into the preload or DBI sharing yet - as that'll end
> up needing a bit of a rewrite of code to take advantage of. I'd be
> open to suggestions here from those who have done it in the past to
> save me going down some dead ends :D

I use mod_perl handlers, so not sure how it mixes with PerlRegistry, but you probably want to have a look at connect_cached


--

Bien à vous, Vincent Veyron

https://urldefense.proofpoint.com/v2/url?u=https-3A__compta.libremen.com&d=DwIFAw&c=D7ByGjS34AllFgecYw0iC6Zq7qlm8uclZFI0SqQnqBo&r=oH2yp0ge1ecj4oDX0XM7vQ&m=C0OcuGbNbfxaSa8ASgV3uFXzejn7MpjIUH1aP1RbiyU&s=GPr8VuKQ3rZCzCPwggyAHdCOojK6ZThmShKk0Jb3maI&e=
Logiciel libre de comptabilité générale en partie double



--
The Wellcome Sanger Institute is operated by Genome Research
Limited, a charity registered in England with number 1021457 and a
company registered in England with number 2742969, whose registered
office is 215 Euston Road, London, NW1 2BE.





--
The Wellcome Sanger Institute is operated by Genome Research
Limited, a charity registered in England with number 1021457 and a
company registered in England with number 2742969, whose registered
office is 215 Euston Road, London, NW1 2BE.

On Sun, 7 Feb 2021 20:21:34 +0000
James Smith <js5@sanger.ac.uk> wrote:

Hi James,

> DBI sharing doesn't really gain you much - and can actually lead you into a whole world of pain. It isn't actually worth turning it on at all.
>

Never had a problem with it myself in years of using it, but I wrap my queries in an eval { } and check $@, so that the scripts are not left hanging; also I have a postgresql db ;-).

I ran some tests with ab, I do see an improvement in response speed :

my $dbh = DBI->connect()
Concurrency Level: 5
Time taken for tests: 22.198 seconds
Complete requests: 1000
Failed requests: 0
Total transferred: 8435000 bytes
HTML transferred: 8176000 bytes
Requests per second: 45.05 [#/sec] (mean)
Time per request: 110.990 [ms] (mean)
Time per request: 22.198 [ms] (mean, across all concurrent requests)
Transfer rate: 371.08 [Kbytes/sec] received

my $dbh = DBI->connect_cached()
Concurrency Level: 5
Time taken for tests: 15.133 seconds
Complete requests: 1000
Failed requests: 0
Total transferred: 8435000 bytes
HTML transferred: 8176000 bytes
Requests per second: 66.08 [#/sec] (mean)
Time per request: 75.664 [ms] (mean)
Time per request: 15.133 [ms] (mean, across all concurrent requests)
Transfer rate: 544.33 [Kbytes/sec] received


--

Bien ? vous, Vincent Veyron

https://compta.libremen.com
Logiciel libre de comptabilit? g?n?rale en partie double

Connection caching does work for most use cases - we have to accept James
works in scenarios most developers can't fathom :)

If you are just firing off simple SQL's without any triggers or named
temporary tables involved you should be good. The only times we recall
tripping on cached connection is when two different code snippets tried to
create the same temporary table. Another time the code was expecting the
disconnect to complete the connection cleanup.

On Tue, Feb 9, 2021 at 11:47 AM Vincent Veyron <vv.lists@wanadoo.fr> wrote:

> On Sun, 7 Feb 2021 20:21:34 +0000
> James Smith <js5@sanger.ac.uk> wrote:
>
> Hi James,
>
> > DBI sharing doesn't really gain you much - and can actually lead you
> into a whole world of pain. It isn't actually worth turning it on at all.
> >
>
> Never had a problem with it myself in years of using it, but I wrap my
> queries in an eval { } and check $@, so that the scripts are not left
> hanging; also I have a postgresql db ;-).
>
> I ran some tests with ab, I do see an improvement in response speed :
>
> my $dbh = DBI->connect()
> Concurrency Level: 5
> Time taken for tests: 22.198 seconds
> Complete requests: 1000
> Failed requests: 0
> Total transferred: 8435000 bytes
> HTML transferred: 8176000 bytes
> Requests per second: 45.05 [#/sec] (mean)
> Time per request: 110.990 [ms] (mean)
> Time per request: 22.198 [ms] (mean, across all concurrent requests)
> Transfer rate: 371.08 [Kbytes/sec] received
>
> my $dbh = DBI->connect_cached()
> Concurrency Level: 5
> Time taken for tests: 15.133 seconds
> Complete requests: 1000
> Failed requests: 0
> Total transferred: 8435000 bytes
> HTML transferred: 8176000 bytes
> Requests per second: 66.08 [#/sec] (mean)
> Time per request: 75.664 [ms] (mean)
> Time per request: 15.133 [ms] (mean, across all concurrent requests)
> Transfer rate: 544.33 [Kbytes/sec] received
>
>
> --
>
> Bien à vous, Vincent Veyron
>
> https://compta.libremen.com
> Logiciel libre de comptabilité générale en partie double
>
>
>
>

It doesn't matter what db - and whether you wrap it in eval it is a problem (postgres has a similar problem - the one with least problems is MySQL) - if you have a secure environment where your databases are in a firewalled zone it will happen to all of them... It's a nasty bit of networking - it does mean our meant to be secure enterprise level apps running against Oracle and less secure and less stable than the other apps we have (go figure!)...

-----Original Message-----
From: Vincent Veyron <vv.lists@wanadoo.fr>
Sent: 09 February 2021 17:47
To: modperl@perl.apache.org
Subject: Re: Moving ExecCGI to mod_perl - performance and custom 'modules' [EXT]

On Sun, 7 Feb 2021 20:21:34 +0000
James Smith <js5@sanger.ac.uk> wrote:

Hi James,

> DBI sharing doesn't really gain you much - and can actually lead you into a whole world of pain. It isn't actually worth turning it on at all.
>

Never had a problem with it myself in years of using it, but I wrap my queries in an eval { } and check $@, so that the scripts are not left hanging; also I have a postgresql db ;-).

I ran some tests with ab, I do see an improvement in response speed :

my $dbh = DBI->connect()
Concurrency Level: 5
Time taken for tests: 22.198 seconds
Complete requests: 1000
Failed requests: 0
Total transferred: 8435000 bytes
HTML transferred: 8176000 bytes
Requests per second: 45.05 [#/sec] (mean)
Time per request: 110.990 [ms] (mean)
Time per request: 22.198 [ms] (mean, across all concurrent requests)
Transfer rate: 371.08 [Kbytes/sec] received

my $dbh = DBI->connect_cached()
Concurrency Level: 5
Time taken for tests: 15.133 seconds
Complete requests: 1000
Failed requests: 0
Total transferred: 8435000 bytes
HTML transferred: 8176000 bytes
Requests per second: 66.08 [#/sec] (mean)
Time per request: 75.664 [ms] (mean)
Time per request: 15.133 [ms] (mean, across all concurrent requests)
Transfer rate: 544.33 [Kbytes/sec] received


--

Bien ? vous, Vincent Veyron

https://urldefense.proofpoint.com/v2/url?u=https-3A__compta.libremen.com&d=DwIFAw&c=D7ByGjS34AllFgecYw0iC6Zq7qlm8uclZFI0SqQnqBo&r=oH2yp0ge1ecj4oDX0XM7vQ&m=u0vYr2KXDAvFiif8YsX-7Uho_gsySe2x9Z3OHcD_Br4&s=7Hyp0l39Edk8cAZK0idIxVxKi3OQXhkR96T0T42b2tM&e=
Logiciel libre de comptabilit? g?n?rale en partie double





--
The Wellcome Sanger Institute is operated by Genome Research
Limited, a charity registered in England with number 1021457 and a
company registered in England with number 2742969, whose registered
office is 215 Euston Road, London, NW1 2BE.

> On 9 Feb 2021, at 18:45, James Smith <js5@sanger.ac.uk> wrote:
>
> It doesn't matter what db - and whether you wrap it in eval it is a problem (postgres has a similar problem - the one with least problems is MySQL) - if you have a secure environment where your databases are in a firewalled zone it will happen to all of them... It's a nasty bit of networking - it does mean our meant to be secure enterprise level apps running against Oracle and less secure and less stable than the other apps we have (go figure!)…
20 years ago I had exactly this argument with Amex when we wanted to use it for payment on the site I then worked for.

They said put a firewall between the app and db layerx

I said it's a dedicated nic/vlan on both sides and the ONLY port that is open is the db - what is a firewall going to add to that.

Eventually they agreed.

Security people who say firewall firewall firewall will solve all your security issues (or even some of them) are useless.

Most of them do it by the book - which should in all case just be the starting point.

Just my 2p

—
C

Another thing that can be done is keep the app server + DB inside your LAN and place a reverse proxy on your DMZ, that adds some level of protection.

Rafael


On Feb 9, 2021, 2:08 PM -0500, Clive Eisen <clive@hildebrand.co.uk>, wrote:

On 9 Feb 2021, at 18:45, James Smith <js5@sanger.ac.uk> wrote:

It doesn't matter what db - and whether you wrap it in eval it is a problem (postgres has a similar problem - the one with least problems is MySQL) - if you have a secure environment where your databases are in a firewalled zone it will happen to all of them... It's a nasty bit of networking - it does mean our meant to be secure enterprise level apps running against Oracle and less secure and less stable than the other apps we have (go figure!)…
20 years ago I had exactly this argument with Amex when we wanted to use it for payment on the site I then worked for.

They said put a firewall between the app and db layerx

I said it's a dedicated nic/vlan on both sides and the ONLY port that is open is the db - what is a firewall going to add to that.

Eventually they agreed.

Security people who say firewall firewall firewall will solve all your security issues (or even some of them) are useless.

Most of them do it by the book - which should in all case just be the starting point.

Just my 2p

—
C

> On 9 Feb 2021, at 19:16, Rafael Caceres <rcaceres@aasa.com.pe> wrote:
>
> Another thing that can be done is keep the app server + DB inside your LAN and place a reverse proxy on your DMZ, that adds some level of protection.

Not really - the only protection is if all your apis or web pages are secure - the reverse proxy does not help or hinder that.

—
C

Mithun,

I’m not sure on what scale you work – but these are from experience in sites with small to medium load – and we rarely see an appreciable gain in using cached or pooled connections, just the occasional heartache they cause.
If you are working on small applications with a minimal number of databases on the DB server then you may see some performance improvement (but tbh not as much as you used to – as the servers have changed) Unfortunately I don’t in both my main and secondary roles, and I know many others who come across these limitations as well.

I’m not saying don’t use persistent or cached connections – but leaving it to some hidden layers is not necessarily a good thing to do – it can have unforeseen side effects {and Apache::DBI & PHP pconnect have both shown these up}

If you are working with e.g. with MySQL the overhead of the (socket) connection is very small, but having more connections open to cope with persistent connections {memory wise} often needs specifying a much large database server – or not being able to do all the nice tricks to in memory indexes and queries [to increase query performance]. Being able to chose which connections you keep open and which you open/close on a per request basis gives you the benefits of caching without the risks involved [other than the “lock table” issue].


From: Mithun Bhattacharya <mithnb@gmail.com>
Sent: 09 February 2021 18:34
To: mod_perl list <modperl@perl.apache.org>
Subject: Re: Moving ExecCGI to mod_perl - performance and custom 'modules' [EXT]

Connection caching does work for most use cases - we have to accept James works in scenarios most developers can't fathom :)

If you are just firing off simple SQL's without any triggers or named temporary tables involved you should be good. The only times we recall tripping on cached connection is when two different code snippets tried to create the same temporary table. Another time the code was expecting the disconnect to complete the connection cleanup.

On Tue, Feb 9, 2021 at 11:47 AM Vincent Veyron <vv.lists@wanadoo.fr<mailto:vv.lists@wanadoo.fr>> wrote:
On Sun, 7 Feb 2021 20:21:34 +0000
James Smith <js5@sanger.ac.uk<mailto:js5@sanger.ac.uk>> wrote:

Hi James,

> DBI sharing doesn't really gain you much - and can actually lead you into a whole world of pain. It isn't actually worth turning it on at all.
>

Never had a problem with it myself in years of using it, but I wrap my queries in an eval { } and check $@, so that the scripts are not left hanging; also I have a postgresql db ;-).

I ran some tests with ab, I do see an improvement in response speed :

my $dbh = DBI->connect()
Concurrency Level: 5
Time taken for tests: 22.198 seconds
Complete requests: 1000
Failed requests: 0
Total transferred: 8435000 bytes
HTML transferred: 8176000 bytes
Requests per second: 45.05 [#/sec] (mean)
Time per request: 110.990 [ms] (mean)
Time per request: 22.198 [ms] (mean, across all concurrent requests)
Transfer rate: 371.08 [Kbytes/sec] received

my $dbh = DBI->connect_cached()
Concurrency Level: 5
Time taken for tests: 15.133 seconds
Complete requests: 1000
Failed requests: 0
Total transferred: 8435000 bytes
HTML transferred: 8176000 bytes
Requests per second: 66.08 [#/sec] (mean)
Time per request: 75.664 [ms] (mean)
Time per request: 15.133 [ms] (mean, across all concurrent requests)
Transfer rate: 544.33 [Kbytes/sec] received


--

Bien à vous, Vincent Veyron

https://compta.libremen.com [compta.libremen.com]<https://urldefense.proofpoint.com/v2/url?u=https-3A__compta.libremen.com&d=DwMFaQ&c=D7ByGjS34AllFgecYw0iC6Zq7qlm8uclZFI0SqQnqBo&r=oH2yp0ge1ecj4oDX0XM7vQ&m=CnIW-j3Bw_IfohZCciiwtkoqvr6nV2hHrNYMPpEOe8E&s=uf6Qi4tnTPryVuPvOKwfZQcFOksecWyn-LYPDVj44lY&e=>
Logiciel libre de comptabilité générale en partie double





--
The Wellcome Sanger Institute is operated by Genome Research
Limited, a charity registered in England with number 1021457 and a
company registered in England with number 2742969, whose registered
office is 215 Euston Road, London, NW1 2BE.

The advantage of the web proxy is not from securing your app - although there are things you can do on the reverse proxy to secure less secure apps

It's main advantage is that it doesn't run a large software stack - and so it makes it harder for people to compromise your front end and then compromise your internal networks, and even then they have to get from that DMZ into your main infrastructure.

We go a step further at work. We have a DMZ <- a web zone <- internal zone - so even if you can compromise the DMZ and the web servers you still don't have direct access to the other machines - taking servers + desktop machines - something like 30-50K cores.


-----Original Message-----
From: Clive Eisen <clive@hildebrand.co.uk>
Sent: 09 February 2021 19:23
To: Rafael Caceres <rcaceres@aasa.com.pe>
Cc: James Smith <js5@sanger.ac.uk>; Vincent Veyron <vv.lists@wanadoo.fr>; modperl@perl.apache.org
Subject: Re: Moving ExecCGI to mod_perl - performance and custom 'modules' [EXT]


> On 9 Feb 2021, at 19:16, Rafael Caceres <rcaceres@aasa.com.pe> wrote:
>
> Another thing that can be done is keep the app server + DB inside your LAN and place a reverse proxy on your DMZ, that adds some level of protection.

Not really - the only protection is if all your apis or web pages are secure - the reverse proxy does not help or hinder that.

—
C




--
The Wellcome Sanger Institute is operated by Genome Research
Limited, a charity registered in England with number 1021457 and a
company registered in England with number 2742969, whose registered
office is 215 Euston Road, London, NW1 2BE.

I would consider mine a small setup on an internal network and I have used
both Sybase and SQL Server. In our case the DBA's preferred us to remain
connected rather than make too many connections - we need DB access in
bursts - it could be quiet for more than an hour and then suddenly we might
need hundreds of connections within few minutes (if we didnt cache it).
Another thing was we were connecting from forked processes so at some point
everything gets reaped including the connections. Our style of coding has
been to connect to the DB wherever we actually need to fire one or more
SQLs and do connect_cached in the actual implementation (it is a separate
library since we had to place a wrapper to acquire credentials)

On Tue, Feb 9, 2021 at 2:34 PM James Smith <js5@sanger.ac.uk> wrote:

> Mithun,
>
> I’m not sure on what scale you work – but these are from experience in
> sites with small to medium load – and we rarely see an appreciable gain in
> using cached or pooled connections, just the occasional heartache they
> cause.
> If you are working on small applications with a minimal number of
> databases on the DB server then you may see some performance improvement
> (but tbh not as much as you used to – as the servers have changed)
> Unfortunately I don’t in both my main and secondary roles, and I know many
> others who come across these limitations as well.
>
> I’m not saying don’t use persistent or cached connections – but leaving it
> to some hidden layers is not necessarily a good thing to do – it can have
> unforeseen side effects {and Apache::DBI & PHP pconnect have both shown
> these up}
>
> If you are working with e.g. with MySQL the overhead of the (socket)
> connection is very small, but having more connections open to cope with
> persistent connections {memory wise} often needs specifying a much large
> database server – or not being able to do all the nice tricks to in memory
> indexes and queries [to increase query performance]. Being able to chose
> which connections you keep open and which you open/close on a per request
> basis gives you the benefits of caching without the risks involved [other
> than the “lock table” issue].
>
>
>
>
>
> *From:* Mithun Bhattacharya <mithnb@gmail.com>
> *Sent:* 09 February 2021 18:34
> *To:* mod_perl list <modperl@perl.apache.org>
> *Subject:* Re: Moving ExecCGI to mod_perl - performance and custom
> 'modules' [EXT]
>
>
>
> Connection caching does work for most use cases - we have to accept James
> works in scenarios most developers can't fathom :)
>
>
>
> If you are just firing off simple SQL's without any triggers or named
> temporary tables involved you should be good. The only times we recall
> tripping on cached connection is when two different code snippets tried to
> create the same temporary table. Another time the code was expecting the
> disconnect to complete the connection cleanup.
>
>
>
> On Tue, Feb 9, 2021 at 11:47 AM Vincent Veyron <vv.lists@wanadoo.fr>
> wrote:
>
> On Sun, 7 Feb 2021 20:21:34 +0000
> James Smith <js5@sanger.ac.uk> wrote:
>
> Hi James,
>
> > DBI sharing doesn't really gain you much - and can actually lead you
> into a whole world of pain. It isn't actually worth turning it on at all.
> >
>
> Never had a problem with it myself in years of using it, but I wrap my
> queries in an eval { } and check $@, so that the scripts are not left
> hanging; also I have a postgresql db ;-).
>
> I ran some tests with ab, I do see an improvement in response speed :
>
> my $dbh = DBI->connect()
> Concurrency Level: 5
> Time taken for tests: 22.198 seconds
> Complete requests: 1000
> Failed requests: 0
> Total transferred: 8435000 bytes
> HTML transferred: 8176000 bytes
> Requests per second: 45.05 [#/sec] (mean)
> Time per request: 110.990 [ms] (mean)
> Time per request: 22.198 [ms] (mean, across all concurrent requests)
> Transfer rate: 371.08 [Kbytes/sec] received
>
> my $dbh = DBI->connect_cached()
> Concurrency Level: 5
> Time taken for tests: 15.133 seconds
> Complete requests: 1000
> Failed requests: 0
> Total transferred: 8435000 bytes
> HTML transferred: 8176000 bytes
> Requests per second: 66.08 [#/sec] (mean)
> Time per request: 75.664 [ms] (mean)
> Time per request: 15.133 [ms] (mean, across all concurrent requests)
> Transfer rate: 544.33 [Kbytes/sec] received
>
>
> --
>
> Bien à vous, Vincent Veyron
>
> https://compta.libremen.com [compta.libremen.com]
> <https://urldefense.proofpoint.com/v2/url?u=https-3A__compta.libremen.com&d=DwMFaQ&c=D7ByGjS34AllFgecYw0iC6Zq7qlm8uclZFI0SqQnqBo&r=oH2yp0ge1ecj4oDX0XM7vQ&m=CnIW-j3Bw_IfohZCciiwtkoqvr6nV2hHrNYMPpEOe8E&s=uf6Qi4tnTPryVuPvOKwfZQcFOksecWyn-LYPDVj44lY&e=>
> Logiciel libre de comptabilité générale en partie double
>
>
> -- The Wellcome Sanger Institute is operated by Genome Research Limited, a
> charity registered in England with number 1021457 and a company registered
> in England with number 2742969, whose registered office is 215 Euston Road,
> London, NW1 2BE.
>

As a long time Oracle DBA and perl mangler I have refrained from joining
this conversation until now.

Connection caching is a configuration option in DBI, not a coding
requirement. For web servers it is a single line edit in startup.pl
import DBICache versus import DBI if I recall incorrectly.
NO CODE CHANGES REQUIRED

Now caching connections will require clean code. Code that was
acceptable on a short term connection
causes problems on long term connections. Cursors left open, virtual
connections not closed, etc
are non-issues using DBI but huge problems with DBICache

Applications doing heavy sql or batch work should not use DBICache. Just
from an administrative perspective.
Session less applications like web servers SHALL use it. And I use the
legal meaning for shall

I don't care what database is on the end of a DBI connection, I mangle
em all, poorly

YMMV
Dave

--
Dave Morgan
Senior Consultant, 1001111 Alberta Limited
dave.morgan@1001111.com
403 399 2442

Isn't this discussion about connection pools and firewalls etc getting a bit far from the
initial subject of the thread ?

On 09.02.2021 23:03, Mithun Bhattacharya wrote:
> I would consider mine a small setup on an internal network and I have used both Sybase and
> SQL Server. In our case the DBA's preferred us to remain connected rather than make too
> many connections - we need DB access in bursts - it could be quiet for more than an hour
> and then suddenly we might need hundreds of connections within few minutes (if we didnt
> cache it). Another thing was we were connecting from forked processes so at some point
> everything gets reaped including the connections. Our style of coding has been to connect
> to the DB wherever we actually need to fire one or more SQLs and do connect_cached in the
> actual implementation (it is a separate library since we had to place a wrapper to
> acquire credentials)
>
> On Tue, Feb 9, 2021 at 2:34 PM James Smith <js5@sanger.ac.uk <mailto:js5@sanger.ac.uk>> wrote:
>
> Mithun,
>
> I’m not sure on what scale you work – but these are from experience in sites with
> small to medium load – and we rarely see an appreciable gain in using cached or pooled
> connections, just the occasional heartache they cause.
> If you are working on small applications with a minimal number of databases on the DB
> server then you may see some performance improvement (but tbh not as much as you used
> to – as the servers have changed) Unfortunately I don’t in both my main and secondary
> roles, and I know many others who come across these limitations as well.
>
> I’m not saying don’t use persistent or cached connections – but leaving it to some
> hidden layers is not necessarily a good thing to do – it can have unforeseen side
> effects {and Apache::DBI & PHP pconnect have both shown these up}
>
> If you are working with e.g. with MySQL the overhead of the (socket) connection is
> very small, but having more connections open to cope with persistent connections
> {memory wise} often needs specifying a much large database server – or not being able
> to do all the nice tricks to in memory indexes and queries [to increase query
> performance]. Being able to chose which connections you keep open and which you
> open/close on a per request basis gives you the benefits of caching without the risks
> involved [other than the “lock table” issue].____
>
> __ __
>
> __ __
>
> *From:*Mithun Bhattacharya <mithnb@gmail.com <mailto:mithnb@gmail.com>>
> *Sent:* 09 February 2021 18:34
> *To:* mod_perl list <modperl@perl.apache.org <mailto:modperl@perl.apache.org>>
> *Subject:* Re: Moving ExecCGI to mod_perl - performance and custom 'modules' [EXT]____
>
> __ __
>
> Connection caching does work for most use cases - we have to accept James works in
> scenarios most developers can't fathom :) ____
>
> __ __
>
> If you are just firing off simple SQL's without any triggers or named temporary tables
> involved you should be good. The only times we recall tripping on cached connection is
> when two different code snippets tried to create the same temporary table. Another
> time the code was expecting the disconnect to complete the connection cleanup.____
>
> __ __
>
> On Tue, Feb 9, 2021 at 11:47 AM Vincent Veyron <vv.lists@wanadoo.fr
> <mailto:vv.lists@wanadoo.fr>> wrote:____
>
> On Sun, 7 Feb 2021 20:21:34 +0000
> James Smith <js5@sanger.ac.uk <mailto:js5@sanger.ac.uk>> wrote:
>
> Hi James,
>
> > DBI sharing doesn't really gain you much - and can actually lead you into a
> whole world of pain. It isn't actually worth turning it on at all.
> >
>
> Never had a problem with it myself in years of using it, but I wrap my queries in
> an eval { } and check $@, so that the scripts are not left hanging; also I have a
> postgresql db ;-).
>
> I ran some tests with ab, I do see an improvement in response speed :
>
> my $dbh = DBI->connect()
> Concurrency Level:      5
> Time taken for tests:   22.198 seconds
> Complete requests:      1000
> Failed requests:        0
> Total transferred:      8435000 bytes
> HTML transferred:       8176000 bytes
> Requests per second:    45.05 [#/sec] (mean)
> Time per request:       110.990 [ms] (mean)
> Time per request:       22.198 [ms] (mean, across all concurrent requests)
> Transfer rate:          371.08 [Kbytes/sec] received
>
> my $dbh = DBI->connect_cached()
> Concurrency Level:      5
> Time taken for tests:   15.133 seconds
> Complete requests:      1000
> Failed requests:        0
> Total transferred:      8435000 bytes
> HTML transferred:       8176000 bytes
> Requests per second:    66.08 [#/sec] (mean)
> Time per request:       75.664 [ms] (mean)
> Time per request:       15.133 [ms] (mean, across all concurrent requests)
> Transfer rate:          544.33 [Kbytes/sec] received
>
>
> --
>
>                                         Bien à vous, Vincent Veyron
>
> https://compta.libremen.com [compta.libremen.com]
> <https://urldefense.proofpoint.com/v2/url?u=https-3A__compta.libremen.com&d=DwMFaQ&c=D7ByGjS34AllFgecYw0iC6Zq7qlm8uclZFI0SqQnqBo&r=oH2yp0ge1ecj4oDX0XM7vQ&m=CnIW-j3Bw_IfohZCciiwtkoqvr6nV2hHrNYMPpEOe8E&s=uf6Qi4tnTPryVuPvOKwfZQcFOksecWyn-LYPDVj44lY&e=>
> Logiciel libre de comptabilité générale en partie double
>
>
> ____
>
> -- The Wellcome Sanger Institute is operated by Genome Research Limited, a charity
> registered in England with number 1021457 and a company registered in England with
> number 2742969, whose registered office is 215 Euston Road, London, NW1 2BE.
>

On Thu, Feb 11, 2021 at 09:52:16AM +0100, André Warnier (tomcat/perl) wrote:
> Isn't this discussion about connection pools and firewalls etc getting a bit
> far from the initial subject of the thread ?

Perhaps. But this has become a pretty low volume mailing list.
This "thread" has moved me to spend hours looking at changing and/or
better understanding the work I have done (pretty old code) and the
work I am now starting.

For me, I'm re-reading the manual pages for the DBI modules,
etc. I've also added another mailing list to follow about DBI.

And I will now have some threads to add in the near future.
Threads I wouldn't have thought of.
But this isn't my mailing list, so breaking these topics into new
threads is just fine. Not a problem at all. 8-)

Recently, something "clicked on" for me about mod_perl.
Which is pretty thrilling for me. ;-}

Chris


>
> On 09.02.2021 23:03, Mithun Bhattacharya wrote:
> > I would consider mine a small setup on an internal network and I have
> > used both Sybase and SQL Server. In our case the DBA's preferred us to
> > remain connected rather than make too many connections - we need DB
> > access in bursts - it could be quiet for more than an hour and then
> > suddenly we might need hundreds of connections within few minutes (if we
> > didnt cache it). Another thing was we were connecting from forked
> > processes so at some point everything gets reaped including the
> > connections. Our style of coding has been to connect to the DB wherever
> > we actually need to fire one or more SQLs and do connect_cached in the
> > actual implementation (it is a separate library since we had to place a
> > wrapper to acquire credentials)
> >
> > On Tue, Feb 9, 2021 at 2:34 PM James Smith <js5@sanger.ac.uk <mailto:js5@sanger.ac.uk>> wrote:
> >
> > Mithun,
> >
> > I’m not sure on what scale you work – but these are from experience in sites with
> > small to medium load – and we rarely see an appreciable gain in using cached or pooled
> > connections, just the occasional heartache they cause.
> > If you are working on small applications with a minimal number of databases on the DB
> > server then you may see some performance improvement (but tbh not as much as you used
> > to – as the servers have changed) Unfortunately I don’t in both my main and secondary
> > roles, and I know many others who come across these limitations as well.
> >
> > I’m not saying don’t use persistent or cached connections – but leaving it to some
> > hidden layers is not necessarily a good thing to do – it can have unforeseen side
> > effects {and Apache::DBI & PHP pconnect have both shown these up}
> >
> > If you are working with e.g. with MySQL the overhead of the (socket) connection is
> > very small, but having more connections open to cope with persistent connections
> > {memory wise} often needs specifying a much large database server – or not being able
> > to do all the nice tricks to in memory indexes and queries [to increase query
> > performance]. Being able to chose which connections you keep open and which you
> > open/close on a per request basis gives you the benefits of caching without the risks
> > involved [other than the “lock table” issue].____
> >
> > __ __
> >
> > __ __
> >
> > *From:*Mithun Bhattacharya <mithnb@gmail.com <mailto:mithnb@gmail.com>>
> > *Sent:* 09 February 2021 18:34
> > *To:* mod_perl list <modperl@perl.apache.org <mailto:modperl@perl.apache.org>>
> > *Subject:* Re: Moving ExecCGI to mod_perl - performance and custom 'modules' [EXT]____
> >
> > __ __
> >
> > Connection caching does work for most use cases - we have to accept James works in
> > scenarios most developers can't fathom :) ____
> >
> > __ __
> >
> > If you are just firing off simple SQL's without any triggers or named temporary tables
> > involved you should be good. The only times we recall tripping on cached connection is
> > when two different code snippets tried to create the same temporary table. Another
> > time the code was expecting the disconnect to complete the connection cleanup.____
> >
> > __ __
> >
> > On Tue, Feb 9, 2021 at 11:47 AM Vincent Veyron <vv.lists@wanadoo.fr
> > <mailto:vv.lists@wanadoo.fr>> wrote:____
> >
> > On Sun, 7 Feb 2021 20:21:34 +0000
> > James Smith <js5@sanger.ac.uk <mailto:js5@sanger.ac.uk>> wrote:
> >
> > Hi James,
> >
> > > DBI sharing doesn't really gain you much - and can actually lead you into a
> > whole world of pain. It isn't actually worth turning it on at all.
> > >
> >
> > Never had a problem with it myself in years of using it, but I wrap my queries in
> > an eval { } and check $@, so that the scripts are not left hanging; also I have a
> > postgresql db ;-).
> >
> > I ran some tests with ab, I do see an improvement in response speed :
> >
> > my $dbh = DBI->connect()
> > Concurrency Level:      5
> > Time taken for tests:   22.198 seconds
> > Complete requests:      1000
> > Failed requests:        0
> > Total transferred:      8435000 bytes
> > HTML transferred:       8176000 bytes
> > Requests per second:    45.05 [#/sec] (mean)
> > Time per request:       110.990 [ms] (mean)
> > Time per request:       22.198 [ms] (mean, across all concurrent requests)
> > Transfer rate:          371.08 [Kbytes/sec] received
> >
> > my $dbh = DBI->connect_cached()
> > Concurrency Level:      5
> > Time taken for tests:   15.133 seconds
> > Complete requests:      1000
> > Failed requests:        0
> > Total transferred:      8435000 bytes
> > HTML transferred:       8176000 bytes
> > Requests per second:    66.08 [#/sec] (mean)
> > Time per request:       75.664 [ms] (mean)
> > Time per request:       15.133 [ms] (mean, across all concurrent requests)
> > Transfer rate:          544.33 [Kbytes/sec] received
> >
> >
> > --
> >
> >                                         Bien à vous, Vincent Veyron
> >
> > https://compta.libremen.com [compta.libremen.com]
> > <https://urldefense.proofpoint.com/v2/url?u=https-3A__compta.libremen.com&d=DwMFaQ&c=D7ByGjS34AllFgecYw0iC6Zq7qlm8uclZFI0SqQnqBo&r=oH2yp0ge1ecj4oDX0XM7vQ&m=CnIW-j3Bw_IfohZCciiwtkoqvr6nV2hHrNYMPpEOe8E&s=uf6Qi4tnTPryVuPvOKwfZQcFOksecWyn-LYPDVj44lY&e=>
> > Logiciel libre de comptabilité générale en partie double
> >
> >
> > ____
> >
> > -- The Wellcome Sanger Institute is operated by Genome Research Limited, a charity
> > registered in England with number 1021457 and a company registered in England with
> > number 2742969, whose registered office is 215 Euston Road, London, NW1 2BE.
> >
>