hibernate-devel

[Hibernate] hibernate performace improvements and dynamic mappings generation using heuristics

[Mahen P. <ma...@gm...>]

Hi All,

I am a postgraduate student from Imperial College, London,,

I am doing a research on performace optimisations in the peristence layer of
a web
application...

just want to get your ideas on likely improvements that can be done in
hibernate to
optimise the performance,,
these changes should be changes to the hibernate core...

yeah,, i have seen various articles on how to increase performace in
hibernate and all,, but
what i want to know is that what are the areas that i can explore in order
to increase
the performance more..at the code level

i am hoping to try out various optmisations and evaluate the likely impact
of each
optimisation and stuff like that and would like to contribute to hibernate
with my
findings,,,

greatly appreciate some pointers from you all,,

mainly i am concerned about the following disadvantage of hibernate:


"Developers must know enough information about data usage at compile time to
be able to optimise queries and choose the appropriate object-relational
mappings"
so therefore, i am thinking about the repercussions and advantages of
having  dynamically generated mapping files based on heuristics of the usage
pattern of the data,, in other words dynamically deciding on the mapping
based on what queries are used more frquently in a system,,

would appreciate some feedback from you all

Thanks
Mahen

Re: [Hibernate] hibernate performace improvements and dynamic mappings generation using heuristics

[Max R. A. <max...@jb...>]

Hi Mahen,

some have previously talked about utilizing the statistics and aop to  
figure
out more optimal ways of querying/mapping so it is not a new idea; just  
that
noone have yet had the time to investigate into it.

I would assume classic database query optimizing techniques and program  
flow analysis
is applicable in this area so if you know those then simply learn  
hibernate and
apply them ;)

/max

> Hi All,
>
> I am a postgraduate student from Imperial College, London,,
>
> I am doing a research on performace optimisations in the peristence  
> layer of
> a web
> application...
>
> just want to get your ideas on likely improvements that can be done in
> hibernate to
> optimise the performance,,
> these changes should be changes to the hibernate core...
>
> yeah,, i have seen various articles on how to increase performace in
> hibernate and all,, but
> what i want to know is that what are the areas that i can explore in  
> order
> to increase
> the performance more..at the code level
>
> i am hoping to try out various optmisations and evaluate the likely  
> impact
> of each
> optimisation and stuff like that and would like to contribute to  
> hibernate
> with my
> findings,,,
>
> greatly appreciate some pointers from you all,,
>
> mainly i am concerned about the following disadvantage of hibernate:
>
>
> "Developers must know enough information about data usage at compile  
> time to
> be able to optimise queries and choose the appropriate object-relational
> mappings"
> so therefore, i am thinking about the repercussions and advantages of
> having  dynamically generated mapping files based on heuristics of the  
> usage
> pattern of the data,, in other words dynamically deciding on the mapping
> based on what queries are used more frquently in a system,,
>
> would appreciate some feedback from you all
>
> Thanks
> Mahen



-- 
--
Max Rydahl Andersen
callto://max.rydahl.andersen

Hibernate
ma...@hi...
http://hibernate.org

JBoss a division of Red Hat
max...@jb...

[Hibernate] Fwd: hibernate performace improvements and dynamic mappings generation using heuristics

[Mahen P. <ma...@gm...>]

hi,

Appreciate if someone can give me some pointers on where i can find a
Hibernate Developers guide, as in some document(s) which will help in
developing and analysing the hibernate codebase.

Thanks
Mahen

---------- Forwarded message ----------
From: Max Rydahl Andersen <max...@jb...>
Date: Oct 19, 2006 9:24 AM
Subject: Re: [Hibernate] hibernate performace improvements and dynamic
mappings generation using heuristics
To: Mahen Perera <ma...@gm...>, hib...@li...

Hi Mahen,

some have previously talked about utilizing the statistics and aop to
figure
out more optimal ways of querying/mapping so it is not a new idea; just
that
noone have yet had the time to investigate into it.

I would assume classic database query optimizing techniques and program
flow analysis
is applicable in this area so if you know those then simply learn
hibernate and
apply them ;)

/max

> Hi All,
>
> I am a postgraduate student from Imperial College, London,,
>
> I am doing a research on performace optimisations in the peristence
> layer of
> a web
> application...
>
> just want to get your ideas on likely improvements that can be done in
> hibernate to
> optimise the performance,,
> these changes should be changes to the hibernate core...
>
> yeah,, i have seen various articles on how to increase performace in
> hibernate and all,, but
> what i want to know is that what are the areas that i can explore in
> order
> to increase
> the performance more..at the code level
>
> i am hoping to try out various optmisations and evaluate the likely
> impact
> of each
> optimisation and stuff like that and would like to contribute to
> hibernate
> with my
> findings,,,
>
> greatly appreciate some pointers from you all,,
>
> mainly i am concerned about the following disadvantage of hibernate:
>
>
> "Developers must know enough information about data usage at compile
> time to
> be able to optimise queries and choose the appropriate object-relational
> mappings"
> so therefore, i am thinking about the repercussions and advantages of
> having  dynamically generated mapping files based on heuristics of the
> usage
> pattern of the data,, in other words dynamically deciding on the mapping
> based on what queries are used more frquently in a system,,
>
> would appreciate some feedback from you all
>
> Thanks
> Mahen



--
--
Max Rydahl Andersen
callto://max.rydahl.andersen

Hibernate
ma...@hi...
http://hibernate.org

JBoss a division of Red Hat
max...@jb...

Re: [Hibernate] hibernate performace improvements and dynamic mappings generation using heuristics

[Mahen P. <ma...@gm...>]

Hi Max and ppl in the hibernate dev list,

Talking about the dynamic generation of mappings, one of my Colleague has
found out that there is no real performance improvement by doing this since
the underlying database will anyway to the necessary optimisations at the
physical layer: Say for example there is this query which does multiple
joins between tables and say that this query is used heavilly by the
application,,, then the database will identify this and aggregate the
necessary tables so that the query will not be involved in any join
operations.

appreciate you guys feedback/comments on this

thanks
mahen


On 10/19/06, Max Rydahl Andersen <max...@jb...> wrote:
>
> Hi Mahen,
>
> some have previously talked about utilizing the statistics and aop to
> figure
> out more optimal ways of querying/mapping so it is not a new idea; just
> that
> noone have yet had the time to investigate into it.
>
> I would assume classic database query optimizing techniques and program
> flow analysis
> is applicable in this area so if you know those then simply learn
> hibernate and
> apply them ;)
>
> /max
>
> > Hi All,
> >
> > I am a postgraduate student from Imperial College, London,,
> >
> > I am doing a research on performace optimisations in the peristence
> > layer of
> > a web
> > application...
> >
> > just want to get your ideas on likely improvements that can be done in
> > hibernate to
> > optimise the performance,,
> > these changes should be changes to the hibernate core...
> >
> > yeah,, i have seen various articles on how to increase performace in
> > hibernate and all,, but
> > what i want to know is that what are the areas that i can explore in
> > order
> > to increase
> > the performance more..at the code level
> >
> > i am hoping to try out various optmisations and evaluate the likely
> > impact
> > of each
> > optimisation and stuff like that and would like to contribute to
> > hibernate
> > with my
> > findings,,,
> >
> > greatly appreciate some pointers from you all,,
> >
> > mainly i am concerned about the following disadvantage of hibernate:
> >
> >
> > "Developers must know enough information about data usage at compile
> > time to
> > be able to optimise queries and choose the appropriate object-relational
> > mappings"
> > so therefore, i am thinking about the repercussions and advantages of
> > having  dynamically generated mapping files based on heuristics of the
> > usage
> > pattern of the data,, in other words dynamically deciding on the mapping
> > based on what queries are used more frquently in a system,,
> >
> > would appreciate some feedback from you all
> >
> > Thanks
> > Mahen
>
>
>
> --
> --
> Max Rydahl Andersen
> callto://max.rydahl.andersen
>
> Hibernate
> ma...@hi...
> http://hibernate.org
>
> JBoss a division of Red Hat
> max...@jb...
>

Re: [Hibernate] hibernate performace improvements and dynamic mappings generation using heuristics

[Max R. A. <max...@jb...>]

yes, for some scenario's this make sense.

But do remember that what the db optimize's in this case is that
you actually queried these objects (either via a query or lazy loading  
them)
even without you actually using the data.

A "dynamic mapping generation using heuristics" could detect if these  
objects were actually
used and wether it would actually be necessary to do the query or not.

So yes, the heuristics should not find out what the optimal joins are at  
the db-level,
but what access pattern that should be used at the object/query level.

/max

> Hi Max and ppl in the hibernate dev list,
>
> Talking about the dynamic generation of mappings, one of my Colleague has
> found out that there is no real performance improvement by doing this  
> since
> the underlying database will anyway to the necessary optimisations at the
> physical layer: Say for example there is this query which does multiple
> joins between tables and say that this query is used heavilly by the
> application,,, then the database will identify this and aggregate the
> necessary tables so that the query will not be involved in any join
> operations.
>
> appreciate you guys feedback/comments on this
>
> thanks
> mahen
>
>
> On 10/19/06, Max Rydahl Andersen <max...@jb...> wrote:
>>
>> Hi Mahen,
>>
>> some have previously talked about utilizing the statistics and aop to
>> figure
>> out more optimal ways of querying/mapping so it is not a new idea; just
>> that
>> noone have yet had the time to investigate into it.
>>
>> I would assume classic database query optimizing techniques and program
>> flow analysis
>> is applicable in this area so if you know those then simply learn
>> hibernate and
>> apply them ;)
>>
>> /max
>>
>> > Hi All,
>> >
>> > I am a postgraduate student from Imperial College, London,,
>> >
>> > I am doing a research on performace optimisations in the peristence
>> > layer of
>> > a web
>> > application...
>> >
>> > just want to get your ideas on likely improvements that can be done in
>> > hibernate to
>> > optimise the performance,,
>> > these changes should be changes to the hibernate core...
>> >
>> > yeah,, i have seen various articles on how to increase performace in
>> > hibernate and all,, but
>> > what i want to know is that what are the areas that i can explore in
>> > order
>> > to increase
>> > the performance more..at the code level
>> >
>> > i am hoping to try out various optmisations and evaluate the likely
>> > impact
>> > of each
>> > optimisation and stuff like that and would like to contribute to
>> > hibernate
>> > with my
>> > findings,,,
>> >
>> > greatly appreciate some pointers from you all,,
>> >
>> > mainly i am concerned about the following disadvantage of hibernate:
>> >
>> >
>> > "Developers must know enough information about data usage at compile
>> > time to
>> > be able to optimise queries and choose the appropriate  
>> object-relational
>> > mappings"
>> > so therefore, i am thinking about the repercussions and advantages of
>> > having  dynamically generated mapping files based on heuristics of the
>> > usage
>> > pattern of the data,, in other words dynamically deciding on the  
>> mapping
>> > based on what queries are used more frquently in a system,,
>> >
>> > would appreciate some feedback from you all
>> >
>> > Thanks
>> > Mahen
>>
>>
>>
>> --
>> --
>> Max Rydahl Andersen
>> callto://max.rydahl.andersen
>>
>> Hibernate
>> ma...@hi...
>> http://hibernate.org
>>
>> JBoss a division of Red Hat
>> max...@jb...
>>



-- 
--
Max Rydahl Andersen
callto://max.rydahl.andersen

Hibernate
ma...@hi...
http://hibernate.org

JBoss a division of Red Hat
max...@jb...