1

Topic: Changeover DWH by business model. How?

All greetings;
I work many years with DWH and the reporting. Here it is interesting, whether it is possible to make a superstructure over DWH which allows to work with , instead of with tables? Something similar is at C#, is called EntityFramework.
Let's tell, instead of 50 tables, to make essence: the client.
Further at the client a dial-up of characteristics which imbibe the data of all tables, and further samplings, reports and analytics become already by means of essence, instead of by means of tables.
At me the question becomes complicated, since clients millions and at them it is a lot of characteristics. + samplings  interest on every week and on the end of month.
Whether the such is possible generally?
Thanks.

2

Re: Changeover DWH by business model. How?

User2155 wrote:

we Tell, instead of 50 tables, to make essence: the client.
Further at the client a dial-up of characteristics which imbibe the data of all tables, and further samplings, reports and analytics become already by means of essence, instead of by means of tables.

?

3

Re: Changeover DWH by business model. How?

User2155 wrote:

All greetings;
I work many years with DWH and the reporting. Here it is interesting, whether it is possible to make a superstructure over DWH which allows to work with , instead of with tables? Something similar is at C#, is called EntityFramework.
Let's tell, instead of 50 tables, to make essence: the client.
Further at the client a dial-up of characteristics which imbibe the data of all tables, and further samplings, reports and analytics become already by means of essence, instead of by means of tables.
At me the question becomes complicated, since clients millions and at them it is a lot of characteristics. + samplings  interest on every week and on the end of month.
Whether the such is possible generally?
Thanks.

Is.  is called. The client is branchy avro object in parquet a file. A hogwash that all the same from it to have BI to build data marts everyone since they relational do not read nothing.

4

Re: Changeover DWH by business model. How?

User2155 wrote:

All greetings;
I work many years with DWH and .

It as so a long time I work.
All is not enough  system system have client applications for the description of a metalayer, transfer of terms DWH in business terms.
In Cognos BI, SAP BO, Oracle BI it is exact.
Though now I can in Cognos BI, open framework and make clients of 50 different systems with different DB, publish the project on a portal and the user the essence of clients will work about one business.

5

Re: Changeover DWH by business model. How?

SpellBuilder wrote:

it is passed...
It as so a long time I work.
All is not enough  system system have client applications for the description of a metalayer, transfer of terms DWH in business terms.
In Cognos BI, SAP BO, Oracle BI it is exact.
Though now I can in Cognos BI, open framework and make clients of 50 different systems with different DB, publish the project on a portal and the user the essence of clients will work about one business.

To make that it is possible, here is how it will work))))) is other question, especially if clients millions and 50 bases)))
Because all these  will become not on a database engine, and on an engine most Cognos which, on  to my observations all the same is rather weak.

6

Re: Changeover DWH by business model. How?

sergeyns wrote:

to Make that it is possible, here is how it will work))))) is other question, especially if clients millions and 50 bases)))
...

Let's divide tasks:
To make outswapping of 50 systems in data store. For in 50 systems the life, requirements, productivity, and you here come also requests rummage on  clients. A disorder...
To draw model of entities, and then to write requests to it. The same SAP BO quite it allows, hides physical essence from the user, and then broadcasts requests from logical essence to a physical DB. Also it is any DB, not SAP HANA.

7

Re: Changeover DWH by business model. How?

Let's assume that basis one.
It is a lot of tables.
does not rescue, since the data in a tabular type you will not present.
Present the client:
- It has an address, and there can be 2 addresses, and can also three;
- And it are accessories for payment. And there can be some accessories;
- It has a heap of managers (from our side).
....
More shortly it not , plus certainly very much a considerable quantity of records. I.e. the data on the client should be aggregated somehow in advance that the subsequent requests worked quickly.
Model  at me and so is. I.e. descriptions of tables, the data are well cleared....
Only  it is somehow curve. If the request from a manual comes, I in the old manner think where the necessary data lies and I do requests. It is a lot of requests, since the data different. Then it is all lays down in a new label and on it reports/cubes are under construction.
If it was possible at once to model  requests to do, here then would be abruptly, only the model should not brake from a considerable quantity of the data.

8

Re: Changeover DWH by business model. How?

User2155;
I.e. you want, that at appearance of new attribute at "client" (for example 4 addresses)
The object model itself was rebuilt and  the data on this new attribute?
So cubes after attribute adding in the circuit and processing work
What problem to collect your client in a snowflake?
Want  "dynamics" leave on   and rolap
Want still  "dynamics" leave on mpp and direct query

9

Re: Changeover DWH by business model. How?

The grandfather;
How it can itself will be set up?)) certainly it is necessary to give the description and the data somehow to tighten.
Cube not a variant. Present a cube, which hours 6 . To that what in this case to do with finishings for 1-2 months when the manual wants to trace any time an index and then about it to forget - I do not want to saw a difficult cube here there.
Kolonochnye indexes already put. Places simply given now less began to occupy, and here their obtaining strongly became complicated.
They well work, if it is necessary ALL heap given to process and if at you any part (i.e. filters are put) or God forbid on this table it is necessary to make Join - to a server badly.
My question in once to construct any data scheme/model, then it to fill and already on its basis to do samplings. Yes, cubes are strongly similar that is necessary, however there it is difficult to project. If at you on the screen 30-50 tables cubed hang, it is what that . Classes in C# it, however they do not store in themselves the data, i.e. will do requests to basis...
It is possible more in detail about rolap mpp and direct query?

10

Re: Changeover DWH by business model. How?

User2155 wrote:

the Grandfather;
How it can itself will be set up?)) certainly it is necessary to give the description and the data somehow to tighten.
Cube not a variant. Present a cube, which hours 6 . To that what in this case to do with finishings for 1-2 months when the manual wants to trace any time an index and then about it to forget - I do not want to saw a difficult cube here there.
Kolonochnye indexes already put. Places simply given now less began to occupy, and here their obtaining strongly became complicated.
They well work, if it is necessary ALL heap given to process and if at you any part (i.e. filters are put) or God forbid on this table it is necessary to make Join - to a server badly.
My question in once to construct any data scheme/model, then it to fill and already on its basis to do samplings. Yes, cubes are strongly similar that is necessary, however there it is difficult to project. If at you on the screen 30-50 tables cubed hang, it is what that . Classes in C# it, however they do not store in themselves the data, i.e. will do requests to basis...
It is possible more in detail about rolap mpp and direct query?

1.
We did  an index + some normal indexes on foreign keys, on popular  fields +
2.
For example, is available DWH on simple Oracle it is already heavy to them to cope with loading.
It is bought Exadata and Oracl over it already flies.

11

Re: Changeover DWH by business model. How?

Thanks.
I seem answered the question.
1) Tables with the initial data at me are structured. Here problems are not present.
2) Over them it is necessary to put data marts which are as much as possible enriched by any information. And here these show-windows without problems can already sit on  indexes.
Can stupid outputs, I apologize.
Simply at me at a head sat down that the data mart is a certain table which contains a part of the necessary information. For example, segmentation of clients for the named date, or the list of contracts for the named date. And in my case a show-window another - there a motley crew from everyones different necessary to me (!!) indexes which can be pushed in single line and to unite a sign.
For example the client + + segmentation of the client + different parameters of the client + residuals + an amount   + channels chavo-nibud + all that it is possible to hammer in the contract into single line.
Such 5-6 show-windows rescue the father of Russian democracy. And the most important will close percent of 70 % of my needs.
too it is necessary, since creation of show-windows business .

12

Re: Changeover DWH by business model. How?

Many years worked with the reporting, but from hundreds millions/billions lines faced recently))) therefore such .

13

Re: Changeover DWH by business model. How?

User2155;
IMHO;
The data mart is NOT one plane table in which that is tightened much.
It is a dial-up (logical sharing) tables with a different level , devoted to certain subjects. It is balance between convenience, ease of usage and support of productivity (rate of fire) of requests about reading. Besides, it is necessary not to admit "duplication" of tables in various show-windows.
Living example of one of big bank:
The table of residuals (the main debt delayed the main debt) in rbl. under the credit contracts with depth stories of ~10 years as of the end of each month
The table of residuals (all types of residuals and charges) in rbl. under retail credit contracts as of every day in depth 1-2 months
The table of residuals (some types of debts) in the main currencies on some subset of credit contracts with depth of history of ~10 years as of the end of each month
And some more variations on the same subject in different subdividings
Though if to think, it is possible to unify in one table "States of contracts for each date", adding other often used indexes
p.s.
Exadata and to it similar cost much

14

Re: Changeover DWH by business model. How?

Alex_496;
By experience it turns out, the more the data, the it is necessary to do normalization less.
I.e. to tighten all that it is necessary in the general tables and to superimpose on them  indexes that counterparts to compress.
Otherwise then it is necessary to do join 300 million on 900 million that defeats any server.
Exadata it is expensive? And why?
If to compare to my salary it is finite  hellish but if it is the large company which should work with terabytes of the data for it it is quite normal price. You want to work - buy, you do not want - let's look at other variants of the reporting, but they will be absolutely-absolutely others :-)

15

Re: Changeover DWH by business model. How?

User2155 wrote:

All greetings;
I work many years with DWH and the reporting. Here it is interesting, whether it is possible to make a superstructure over DWH which allows to work with , instead of with tables? Something similar is at C#, is called EntityFramework.

Greetings! It seems to me a question not about BI more, and about business process management. But it is a separate class of systems - BPM, esteem, can, their ideology you on any thoughts pushes. Because Business process implies control of events and information streams.
And, by the way, to idea some tens years... Remember Rational Rose?? Also Soda to document the code tried smile))
, dreams.... Though, do not buy them IBM who knows. And, probably, soon there will be new systems and we will watch growth of interest to systems of a similar class.
Yours faithfully;
George

16

Re: Changeover DWH by business model. How?

George Nordic;
George, do not import confusion. He only needs to stretch the objective basis, a database able to work about objective .. To PORRIDGE (for example), on tasks  (storage and the analysis). Or to invent as it guys made of Diasoft the objective superstructure over a database.
It in the correct direction digs since basis  is business processes the enterprises which have been sewn up in registration systems, instead of reports of users as some think.

17

Re: Changeover DWH by business model. How?

User2155;
On SAP BW similar, but I with it a sign only under articles, in the live did not work.
Sometimes it use as , without having anything remaining of SAP-a, article from German log about advantages of such choice before traditional decisions came across to me - at once receive : model, it would be not necessary to care of handling of the same SCD at loading-SAP like does it.