Create database

[Vinod S]

Hi all, i know it will sound funny to the senior techies in this forum, but its a serious issue for me that i need to create a database of our own for one of our application. Can any one show me a way to start with. Please do give importance to this and help me as i am in a puzzlement now.

[HairyMonkeyMan]

Firstly, which DBMS are you using?? :confused:

[davide++]

Firstly, which DBMS are you using?? :confused:

I agree.
If you're going to use an Access database this task is the simpliest thing you have to do.
If you want to create an Oracle instance this's a very hard job.

[Vinod S]

Dear all, you may get surprised by my answer to your above questions, but the thing is that we need to create our own data storage. Our own database in other way.

As i told you earlier it will be funny to you guys but the client requirement is so and we need to give them a solution not using conventional database, they need to store the data on their on proprietary format and file. Thats why we are behind it. So its not a jock for us now and became a real headache. Can you help me in handling this situation.

[PeejAvery]

Not use an already existing database platform??? Are your clients idiots? The time it would take to develop a database platform would be such a waste of money. Also, the features in it would be so far behind the abilities of the already existing ones...it would take too long to catch up. By that time, your need for it would be way past deadline.

Why waste your time? Tell them no!

With that said, there is no way that we here at CG have the time to dig into such a project of great size. But, even if it was a paid proposition, it is still an idiotic gesture.

:confused:

[HairyMonkeyMan]

Does the new system has to interface with their proprietary system?

We have to deal with something similar in my current job. We are slowly phasing out the old system (while still supporting the active parts).

The old system runs on openvms and uses flat files for data storage (ddf). The data fields are fixed width. We have an 'Attunity' engine that can convert the data to xml on the fly - which we can then connect to with oledb.

We have a nice new mysql server which we synchronize with the old flat files every 5 mins (with a custom program).

Would this setup suit you?

If they just want everything in the old format for the sake of consistency, then I agree with PeejAvery

[davide++]

Dear all, you may get surprised by my answer to your above questions, but the thing is that we need to create our own data storage. Our own database in other way.

As i told you earlier it will be funny to you guys but the client requirement is so and we need to give them a solution not using conventional database, they need to store the data on their on proprietary format and file. Thats why we are behind it. So its not a jock for us now and became a real headache. Can you help me in handling this situation.

Well.
I confess I didn't read carefully your post, so I thought that you've been forgetting to write what db you're going to use.
Actually you have to create a new database, but as PeejAvery said this's a very very very hard task, that requires time.
What your system have to do?
If it must to perform the database operations (adding, updating, deleting and searching records) on files in a particular format what you can do depend on the specific format, but in general you have to read and write the files.
HairyMonkeyMan's solution is an interesting one, but I see some problems, above all sinchronization.

[cjard]

Actually you have to create a new database, but as PeejAvery said this's a very very very hard task, that requires time.

indeed, it's taken oracle more than a decade, and they are still working on it..

I'd be tempted to avoid a phased transition from the old system in this case.. Develop a new system on a proper existing RDBMS, replicate all the used functionality of the old system, and then port the data across and dump the old system..

[George1111]

The "essence" of a database is a direct access file, plus and indexing system to those records

eg,

Main File Contains records like

RNO DATA
1 Jones 125 Main Street New York ................... data .......
2 Smith 25 Henry Rd Charlestown .................. data ......
3 Adams 55 Observation Drive Boston .............. data ......

If the file contained thousands of records and you were attempting to find "SMITH" you would sequentially read the file until you found it - this could be slow if each record was 3000 bytes long

So then we add an index which can be sorted on a criteria we choose - eg, Name

We create an Index as follows

Adams 3
Jones 1
Smith 2

We can now read the index until we find the match we want, and using the record number associated with it, perform a direct hit on the main record

If the index is perfectly sorted, you can also do a Binary search in the index file which most times will allow you to find the index in less than 10 hits

Ie, Point to the centre of the index file - Is the record I want less than or greater than the half way point ? LESS. Take the 1st half, and point to the centre of that, and so on

Prior to the advent of hard disks in personal computers in 1981, I developed the above file system which gave me return of a list of matches, from 4000 records in less than 1 second.

That was using an 8" Floppy disk


The point is that you can create your own file system (I wont call it a database) but as the other guys say "at what price"

If you have a need to go outside of SQL Server and Access databases which can be FREE, then you are looking for a mountain (Everest) of work for no real return

If you really want to be nonconfirmist, you will end up with a system which will rapidly became unmaintainable, because only a very few people will understand the concept of your system (its all very well untill you suddenly encounter a bug in some essoteric part of the indexing routine )

A nice variation to main stream databases can be to use ISAM Indexed Files - these may even perform faster than traditional databases but will need addition maintenance to keep the indexes well organised - they are simply files with indexes, not a database (which can be differentiated by the fact that One File name contains many tables)

For me, I would go with SQL Server 2005 Express which will pay great dividends in terms of code standardisation, Reporting Capabilities, Data Ware Housing Capabilities (Analysis Services) ALL of which are vital to an organisation if it is going to succeed in business in the next 20 years

Any other approach is really not too bright

[Teranoz]

Not use an already existing database platform??? Are your clients idiots? The time it would take to develop a database platform would be such a waste of money. Also, the features in it would be so far behind the abilities of the already existing ones...it would take too long to catch up. By that time, your need for it would be way past deadline.

Why waste your time? Tell them no!

With that said, there is no way that we here at CG have the time to dig into such a project of great size. But, even if it was a paid proposition, it is still an idiotic gesture.

:confused:

Maybe they need to create a new programming language also for that :-)

[Shuja Ali]

If your client insists on not using any of the existing RDBMS then you could try your hands on XML. As others have already pointed created your own DBMS would be time consuming, however if you want to write a data driven application that does not use any RDBMS then I will have to say that it is possible.