Обсуждение: Wisconsin Circuit Court Access (WCCA) on PostgreSQL

The Consolidated Court Automation Programs (CCAP) of the Wisconsin Court
System has migrated to PostgreSQL for all of its Circuit Court web
operations.  Eight production databases have been converted, six of them
around 180 GB each, holding statewide information replicated real-time
from 72 county databases.  The central copies support audit functions,
statewide statistics and report generation, and this web site:

http://wcca.wicourts.gov/

Given the success of this effort, we expect to be converting the other
court databases to PostgreSQL.

We've been very happy with both the performance of the product and the
support we have received from the mailing lists.  Overall, PostgreSQL
has been faster than the commercial product from which we converted.
Even more important is the fast response we have had when posting
problems to the lists.  We have normally had a fix within 24 hours.
Frankly, the support has been amazing.

We wonder how widespread the use of PostgreSQL is within government
agencies; I would love to hear from anyone with experience with this.
With tight budgets, it seems likely that there may be others moving in
this direction.

Kevin Grittner wrote:
> The Consolidated Court Automation Programs (CCAP) of the Wisconsin Court
> System has migrated to PostgreSQL for all of its Circuit Court web
> operations.  Eight production databases have been converted, six of them
> around 180 GB each, holding statewide information replicated real-time
> from 72 county databases.  The central copies support audit functions,
> statewide statistics and report generation, and this web site:
>
> http://wcca.wicourts.gov/
>
> Given the success of this effort, we expect to be converting the other
> court databases to PostgreSQL.

Always good to hear. Two obvious questions:
1. Have you completed a case-study questionnaire (and if not, would you
be prepared to?)
2. People on the advocacy lists occasionally get asked for
quotes/interviewees by the press. Would you be up for this?

> We've been very happy with both the performance of the product and the
> support we have received from the mailing lists.  Overall, PostgreSQL
> has been faster than the commercial product from which we converted.
> Even more important is the fast response we have had when posting
> problems to the lists.  We have normally had a fix within 24 hours.
> Frankly, the support has been amazing.

You can spend more time on hold calling commercial vendors than it takes
to get a patch with PostgreSQL :-)

--
   Richard Huxton
   Archonet Ltd

Kevin Grittner wrote:
> Overall, PostgreSQL
> has been faster than the commercial product from which we converted.
>


Kevin,
Are you allowed to say what commercial product you converted from?

Thanks,

Tony Caduto
AM Software Design
Milwaukee WI
http://www.amsoftwaredesign.com

On Mon, 2006-03-13 at 15:16, Tony Caduto wrote:
> Kevin Grittner wrote:
> > Overall, PostgreSQL
> > has been faster than the commercial product from which we converted.
> >
>
>
> Kevin,
> Are you allowed to say what commercial product you converted from?

And whether he can or not, this would make a GREAT case study for the
advocacy folks.

On Mon, 2006-03-13 at 15:26, Scott Marlowe wrote:
> On Mon, 2006-03-13 at 15:16, Tony Caduto wrote:
> > Kevin Grittner wrote:
> > > Overall, PostgreSQL
> > > has been faster than the commercial product from which we converted.
> > >
> >
> >
> > Kevin,
> > Are you allowed to say what commercial product you converted from?
>
> And whether he can or not, this would make a GREAT case study for the
> advocacy folks.

Yeah, anyone who says installing postgresql is hard should have to
install Oracle first.  Or compile MySQL from source.  :)

>>> On Mon, Mar 13, 2006 at  3:16 pm, in message
<4415E12D.70104@amsoftwaredesign.com>, Tony Caduto
<tony_caduto@amsoftwaredesign.com> wrote:
> Kevin Grittner wrote:
>> Overall, PostgreSQL
>> has been faster than the commercial product from which we converted.

>
> Are you allowed to say what commercial product you converted from?

The license for the commercial product contains a clause which
prohibits disclosing benchmarks of their product without their written
permission.  (Heaven only knows why they would include such a clause.)
My comment is not based on any formal benchmarks, but on the graphs
produced by our monitoring of the production software under real loads,
and on the query metrics from our middle tier software in the production
environment.  Even though it would be a huge stretch to call the
comparison a benchmark under these conditions, this is a litigious
society.  I'm sure you understand my concern.

Short of being compelled by law to open our records, I'm not
comfortable providing any performance comparison which names the
vendor.

-Kevin

On Mon, 2006-03-13 at 13:27 -0600, Kevin Grittner wrote:
> Even more important is the fast response we have had when posting
> problems to the lists.  We have normally had a fix within 24 hours.
> Frankly, the support has been amazing.

Kevin, well done. We've all watched your progress with interest.

The reason you've got excellent support is because of the detailed
postings you've made, together with responses to all replies. Doing all
your homework before posting is essential; unfortunately many people
don't do this and then leave disappointed.

Best Regards, Simon Riggs

One doesn't 'install' oracle. That implies you have control of the
situation. One attempts to convince it to condescend to install itself onto
your machine.

Of course, this is like convincing my 3 year old to go to bed on time. Such
powers of persuasion are not common.

On 3/13/06 5:41 PM, "Scott Marlowe" <smarlowe@g2switchworks.com> wrote:

> On Mon, 2006-03-13 at 15:26, Scott Marlowe wrote:
>> On Mon, 2006-03-13 at 15:16, Tony Caduto wrote:
>>> Kevin Grittner wrote:
>>>> Overall, PostgreSQL
>>>> has been faster than the commercial product from which we converted.
>>>>
>>>
>>>
>>> Kevin,
>>> Are you allowed to say what commercial product you converted from?
>>
>> And whether he can or not, this would make a GREAT case study for the
>> advocacy folks.
>
> Yeah, anyone who says installing postgresql is hard should have to
> install Oracle first.  Or compile MySQL from source.  :)
>
> ---------------------------(end of broadcast)---------------------------
> TIP 3: Have you checked our extensive FAQ?
>
>                http://www.postgresql.org/docs/faq

On Tue, 2006-14-03 at 07:45 -0500, Andrew Rawnsley wrote:
> One doesn't 'install' oracle. That implies you have control of the
> situation. One attempts to convince it to condescend to install itself onto
> your machine.
>
> Of course, this is like convincing my 3 year old to go to bed on time. Such
> powers of persuasion are not common.
>
> On 3/13/06 5:41 PM, "Scott Marlowe" <smarlowe@g2switchworks.com> wrote:
>
> > On Mon, 2006-03-13 at 15:26, Scott Marlowe wrote:
> >> On Mon, 2006-03-13 at 15:16, Tony Caduto wrote:
> >>> Kevin Grittner wrote:
> >>>> Overall, PostgreSQL
> >>>> has been faster than the commercial product from which we converted.
> >>>>
> >>>
> >>>
> >>> Kevin,
> >>> Are you allowed to say what commercial product you converted from?
> >>
> >> And whether he can or not, this would make a GREAT case study for the
> >> advocacy folks.
> >
> > Yeah, anyone who says installing postgresql is hard should have to
> > install Oracle first.  Or compile MySQL from source.  :)
> >
I'll agree with that, we had a tech who tried for a week to install
Oracle, only having to resort to buying third party books, to
figure out how to configure it. Managing it, is another scary task
when that guy left, I soon discovered the jumble of tools required
to administer it. We stopped supporting Oracle when our last
customer using it stopped using it.

I use PostgreSQL for most projects but have been supporting MySQL
for customers who request it, and usually build any new libraries
to be able to support either transparently, just by changing the
driver and user credentials in the config file. The hard part is
usually getting the MySQL to do what I expect, and what PostgreSQL
does by default.

On 3/13/06 5:50 PM, "Kevin Grittner" <Kevin.Grittner@wicourts.gov> wrote:

> The license for the commercial product contains a clause which
> prohibits disclosing benchmarks of their product without their written
> permission.  (Heaven only knows why they would include such a clause.)
> My comment is not based on any formal benchmarks, but on the graphs
> produced by our monitoring of the production software under real loads,
> and on the query metrics from our middle tier software in the production
> environment.  Even though it would be a huge stretch to call the
> comparison a benchmark under these conditions, this is a litigious
> society.  I'm sure you understand my concern.
>
> Short of being compelled by law to open our records, I'm not
> comfortable providing any performance comparison which names the
> vendor.

Hmm..  What vendor throws such a clause into all their licensing agreements.
Sounds suspiciously like Microsoft...

I'm not a lawyer, but I believe things such as what vendors a government
entity is using is required by law to be public information.

Wes

>> Short of being compelled by law to open our records, I'm not
>> comfortable providing any performance comparison which names the
>> vendor.
>>
>
> Hmm..  What vendor throws such a clause into all their licensing agreements.
> Sounds suspiciously like Microsoft...
>

Or Oracle, DB2, Sybase, Progress ...
> I'm not a lawyer, but I believe things such as what vendors a government
> entity is using is required by law to be public information.
>


> Wes
>
>
>
> ---------------------------(end of broadcast)---------------------------
> TIP 5: don't forget to increase your free space map settings
>
>

Wes wrote:
> Hmm..  What vendor throws such a clause into all their licensing agreements.
> Sounds suspiciously like Microsoft...
>
> I'm not a lawyer, but I believe things such as what vendors a government
> entity is using is required by law to be public information.
>
> Wes
>
>
>

I think we can safely assume it's either Oracle or M$ SQL server.

I am leaning towards oracle because the state of WI just had a huge
SNAFU with Oracle as their email provider.
They spent several million dollars on the Oracle groupware/email package
only to find it did not work(disapearing emails etc).
Guess they didn't know about postfix and Open Exchange :-)

Tony

On 14/3/06 20:43, "Wes" <wespvp@syntegra.com> wrote:

> On 3/13/06 5:50 PM, "Kevin Grittner" <Kevin.Grittner@wicourts.gov> wrote:
>
> I'm not a lawyer, but I believe things such as what vendors a government
> entity is using is required by law to be public information.

Maybe in some jurisdictions, but that's not the same as Kevin naming the
vendor in the same message or even thread as he has compared their
perfomance with ours.

Anyway - whilst I'm emailing I'd like to congratulate Kevin on a successful
project, and thank him for telling us about it.

Regards, Dave

>>> On Tue, Mar 14, 2006 at  2:08 am, in message
<1142323733.11178.15.camel@localhost.localdomain>, Simon Riggs
<simon@2ndquadrant.com> wrote:
> On Mon, 2006- 03- 13 at 13:27 - 0600, Kevin Grittner wrote:
>> Even more important is the fast response we have had when posting
>> problems to the lists.  We have normally had a fix within 24 hours.

>> Frankly, the support has been amazing.
>
> Kevin, well done. We've all watched your progress with interest.

Thanks to all who have offered congratulations.

> The reason you've got excellent support is because of the detailed
> postings you've made, together with responses to all replies. Doing
all
> your homework before posting is essential; unfortunately many people
> don't do this and then leave disappointed.

Here I think you underestimate how well the community helps people in
these lists.  I have witnessed remarkable patience here when people post
vague messages asking for help.  You (as a community) generally succeed
in drawing out sufficient detail to provide good advice, and / or
identify areas for product improvement.  I do try to give as much
information as I can, including reproducible test cases where
practicable; but, I have done so with commercial vendors to whom my
clients have paid big money for support, and been very disappointed.

With one commercial vendor we've routinely been told by first line
support staff that the product was functioning as intended.  After days
of effort, sometimes involving calls from top management, we've gotten
through to someone who can actually understand the problem and
acknowledge the bug; only to have it take months (sometimes over a year)
to get a fix,

With another open source vendor, from whom no support is available
without a paid license and a paid support contract, we (after paying for
a commercial license and a support contract) have been told that such
things as using an OR predicate within the ON clause of a JOIN was an
"unimplemented feature" (even though it worked in simple cases).  They
said they might "add the feature" in the next major release, but that
wouldn't be for at least a year, and no guarantees.

It was unexpected and quite refreshing to provide the same level of
detail in a post to a PostgreSQL list, and get a patch file fast enough
to be running a fixed version within 24 hours of posting the problem.
When we have been able to provide sufficient detail and / or a test
case, this has usually been the result.  When we participated in the
beta test phase, people were quite helpful in leading me through the use
of unfamiliar tools to capture the information they needed to identify
and fix problems before the official release.

After decades of working as an independent consultant, I've recently
(eight days ago) accepted employment with the Wisconsin Court System as
a DBA, and I'm told that as a court employee I'm not allowed to endorse
one product over another; but, I can speak of my experiences with
products so long as I don't violate any constraints of the license
agreements.  I have worked with quite a few database products in my
career and can say unequivocally that the support I've seen provided for
PostgreSQL is superior to that which I've seen provided for any other
database product.

I don't want to name any names, because I would undoubtedly forget
several very helpful people here, but I have to admit that my personal
favorite was when I posted information about a bug in the JDBC driver
shortly before I left for the day, and while I was sleeping a user in
Germany created a program to cause the race condition, tracked down the
cause, and posted a patch with a suggested fix.  By the time I'd
finished my coffee the next morning, the patch had been reviewed, scaled
back to the minimum change required to effect a fix, applied to CVS, and
a new jar file deployed for download.  Wow.

I can't really accept congratulations for this successful deployment
without offering it right back to the community for all the help you've
provided, as well as the product itself.  Absolutely fantastic, all
around!

-Kevin

On Mon, Mar 13, 2006 at 17:50:44 -0600,
  Kevin Grittner <Kevin.Grittner@wicourts.gov> wrote:
>
> Short of being compelled by law to open our records, I'm not
> comfortable providing any performance comparison which names the
> vendor.

An open records request inquiring about vendors and contracts with the
State of Wisconsin could probably turn up the answer if anyone really
wanted to know.

* Bruno Wolff III (bruno@wolff.to) wrote:
> On Mon, Mar 13, 2006 at 17:50:44 -0600,
>   Kevin Grittner <Kevin.Grittner@wicourts.gov> wrote:
> > Short of being compelled by law to open our records, I'm not
> > comfortable providing any performance comparison which names the
> > vendor.
>
> An open records request inquiring about vendors and contracts with the
> State of Wisconsin could probably turn up the answer if anyone really
> wanted to know.

About which vendors they use and what contracts they have and you might
be able to figure out which vendors have such a clause.  I don't know
that such a request could compel the performance data out associated
with a specific vendor when that's clearly against a license the state
is currently under.

    Enjoy,

        Stephen

On Tue, Mar 14, 2006 at 23:27:24 -0500,
  Stephen Frost <sfrost@snowman.net> wrote:
> * Bruno Wolff III (bruno@wolff.to) wrote:
> > On Mon, Mar 13, 2006 at 17:50:44 -0600,
> >   Kevin Grittner <Kevin.Grittner@wicourts.gov> wrote:
> > > Short of being compelled by law to open our records, I'm not
> > > comfortable providing any performance comparison which names the
> > > vendor.
> >
> > An open records request inquiring about vendors and contracts with the
> > State of Wisconsin could probably turn up the answer if anyone really
> > wanted to know.
>
> About which vendors they use and what contracts they have and you might
> be able to figure out which vendors have such a clause.  I don't know
> that such a request could compel the performance data out associated
> with a specific vendor when that's clearly against a license the state
> is currently under.

But we already have an unofficial comment on the performance, we just don't
know what database postgres is being compared to.

* Bruno Wolff III (bruno@wolff.to) wrote:
> On Tue, Mar 14, 2006 at 23:27:24 -0500,
>   Stephen Frost <sfrost@snowman.net> wrote:
> > About which vendors they use and what contracts they have and you might
> > be able to figure out which vendors have such a clause.  I don't know
> > that such a request could compel the performance data out associated
> > with a specific vendor when that's clearly against a license the state
> > is currently under.
>
> But we already have an unofficial comment on the performance, we just don't
> know what database postgres is being compared to.

We've probably got a pretty good idea already. :)  Besides, all you'd be
able to get down to would be: what database vendors the state uses
(probably more than one) filtered by which of those have such a clause
in their license (also probably more than one), so in the end all you
know is that it one of a set.

Besides, I don't think it's a good move for us to go digging around
trying to force the state to tell us and then assuming we can corrolate
that to what Kevin was talking about.  Both from a "it's really not that
big a deal" and a "Kevin's a nice guy, let's not get him into trouble
and make him feel like he can't say anything" perspective. :)

    Thanks,

        Stephen

On Tuesday 14 March 2006 16:00, Kevin Grittner wrote:
> >>> On Tue, Mar 14, 2006 at  2:08 am, in message
>
> <1142323733.11178.15.camel@localhost.localdomain>, Simon Riggs
>
> <simon@2ndquadrant.com> wrote:
> > On Mon, 2006- 03- 13 at 13:27 - 0600, Kevin Grittner wrote:
> >> Even more important is the fast response we have had when posting
> >> problems to the lists.  We have normally had a fix within 24 hours.
> >>
> >> Frankly, the support has been amazing.
> >
> > Kevin, well done. We've all watched your progress with interest.
>
> Thanks to all who have offered congratulations.
>
> > The reason you've got excellent support is because of the detailed
> > postings you've made, together with responses to all replies. Doing
>
> all
>
> > your homework before posting is essential; unfortunately many people
> > don't do this and then leave disappointed.
>
> Here I think you underestimate how well the community helps people in
> these lists.  I have witnessed remarkable patience here when people post
> vague messages asking for help.  You (as a community) generally succeed
> in drawing out sufficient detail to provide good advice, and / or
> identify areas for product improvement.  I do try to give as much
> information as I can, including reproducible test cases where
> practicable; but, I have done so with commercial vendors to whom my
> clients have paid big money for support, and been very disappointed.
>
> With one commercial vendor we've routinely been told by first line
> support staff that the product was functioning as intended.  After days
> of effort, sometimes involving calls from top management, we've gotten
> through to someone who can actually understand the problem and
> acknowledge the bug; only to have it take months (sometimes over a year)
> to get a fix,
>
> With another open source vendor, from whom no support is available
> without a paid license and a paid support contract, we (after paying for
> a commercial license and a support contract) have been told that such
> things as using an OR predicate within the ON clause of a JOIN was an
> "unimplemented feature" (even though it worked in simple cases).  They
> said they might "add the feature" in the next major release, but that
> wouldn't be for at least a year, and no guarantees.
>
> It was unexpected and quite refreshing to provide the same level of
> detail in a post to a PostgreSQL list, and get a patch file fast enough
> to be running a fixed version within 24 hours of posting the problem.
> When we have been able to provide sufficient detail and / or a test
> case, this has usually been the result.  When we participated in the
> beta test phase, people were quite helpful in leading me through the use
> of unfamiliar tools to capture the information they needed to identify
> and fix problems before the official release.
>
> After decades of working as an independent consultant, I've recently
> (eight days ago) accepted employment with the Wisconsin Court System as
> a DBA, and I'm told that as a court employee I'm not allowed to endorse
> one product over another; but, I can speak of my experiences with
> products so long as I don't violate any constraints of the license
> agreements.  I have worked with quite a few database products in my
> career and can say unequivocally that the support I've seen provided for
> PostgreSQL is superior to that which I've seen provided for any other
> database product.
>
> I don't want to name any names, because I would undoubtedly forget
> several very helpful people here, but I have to admit that my personal
> favorite was when I posted information about a bug in the JDBC driver
> shortly before I left for the day, and while I was sleeping a user in
> Germany created a program to cause the race condition, tracked down the
> cause, and posted a patch with a suggested fix.  By the time I'd
> finished my coffee the next morning, the patch had been reviewed, scaled
> back to the minimum change required to effect a fix, applied to CVS, and
> a new jar file deployed for download.  Wow.
>
> I can't really accept congratulations for this successful deployment
> without offering it right back to the community for all the help you've
> provided, as well as the product itself.  Absolutely fantastic, all
> around!
>
> -Kevin

If we do not have a testimonilas page, then  this is the perfect example of
why we should, and what should be on it.  As good as we are in features, text
like this can be far more efective in getting a foot in the door with PHB.

(just my 2 bits worth)


>
>
> ---------------------------(end of broadcast)---------------------------
> TIP 9: In versions below 8.0, the planner will ignore your desire to
>        choose an index scan if your joining column's datatypes do not
>        match

--
Darcy Buskermolen
Wavefire Technologies Corp.

http://www.wavefire.com
ph: 250.717.0200
fx: 250.763.1759

> We've probably got a pretty good idea already. :)  Besides, all you'd be
> able to get down to would be: what database vendors the state uses
> (probably more than one) filtered by which of those have such a clause
> in their license (also probably more than one), so in the end all you
> know is that it one of a set.

Well, no, you could get at least RFPs, the winning bid, and the actual
contract for any or all specific projects, because they're all public
records. Local laws vary, but basically no government entity is allowed to
enter into secret contracts, except where national security is a concern...

I'm not interested in pushing the issue, just wanted to point out that when
public funds are spent, the paper trail as to who receives them is public
info.


--
Scott Ribe
scott_ribe@killerbytes.com
http://www.killerbytes.com/
(303) 722-0567 voice