Database record format that encourages optimal retrieve(read, search, etc) & store(write, sort, etc)

[seim...@gmail.com]

Hi,

I am researching data that is optimal in how many
distinct read operations are performed over a period
of time and how many distinct write operations are performed
over a period of time.

I have one example(bank account) which for generality is text only:

Name: John Smith

Age: 50

Amount: + 1000


Name: John Smith Jr

Age: 31

Amount: +100

In the abobe example, an optimal write to the second record
would consider
the redundancy in the Name and Amount.

Likewise, while retrieving the record, an optimal read
will not read all the bytes of the second record rather
it will retrive the "Jr", & subtract a 0 from the Amount.

Lest readers of this Post be confused, I admit that my
example does acknowledge the influence of
Information
theory.(not compression, though).

In the example I intend to have a "Live Database" that serves the purpose
of a reference point. This Live Database will have a lot
of generality(or as much as theory will permit). It is
against this reference that updates and retrieval of
the actual databases will be done.

The above paragraph is the only use of information theory and
thereby the only limitation for my database theory.

My questions that are relevant to database theory are:

i) What are good formats for data that encourage Optimal read
and optimal writes ? Sacling from intra-records(1 bit) to
intra-record(a few bytes) to entire databases.


ii) Does database theory *have to* depend upon information theory or
parallelism or other outside domain technologies for
efficiency ?

iii) In my example, what could the "Live Database" contain ?

Sincerely,
Seima Rao.

[Ed Prochak]

On Sunday, September 13, 2015 at 4:23:50 AM UTC-4, seim...@gmail.com wrote:
> Hi,
>
> I am researching data that is optimal in how many
> distinct read operations are performed over a period
> of time and how many distinct write operations are performed
> over a period of time.

This isn't a theory question exactly. Once you start talking about record formats and reads and writes to files, you are talking about implementation.

>
> I have one example(bank account) which for generality is text only:
>
> Name: John Smith
>
> Age: 50
>
> Amount: + 1000
>
>
> Name: John Smith Jr
>
> Age: 31
>
> Amount: +100
>
> In the abobe example, an optimal write to the second record
> would consider
> the redundancy in the Name and Amount.

NO. John Smith Jr is not John Smith. The information is not Redundant!

>
> Likewise, while retrieving the record, an optimal read
> will not read all the bytes of the second record rather
> it will retrive the "Jr", & subtract a 0 from the Amount.

I'll just stop here and remind you:
Premature optimization is the root of all evil.

I think you are chasing the wrong solution.
[]