Oracle integrating with MySQL

[ccguedes]

Guys, good morning.
is a doubt, even if anyone knows details, could not only heal my doubt, as from other members here in the forum.

I was talking to a friend about using mysql in e-commerce, facebook, and sites of great retailers .. So far, okay, but the doubts emerged.

The architecture is usually done as follows:

MySQL is the front-end seat .. Bank that is first requested, made the requests is printed on the Screen for customer / user, but mass itself is in Oracle. Why this?. Such that?
Can not the Oracle the DB Front? Are there limitations that MySQL better meets Oracle in this sense?

Who can sketch a simple architecture of this sense, and TVz, explain in detail because one and not another, cost, performance, embedded technology and the like, would appreciate it.

Thank you very much.

[adrianoturbo]

Brother, to remedy your questions: Let's give a brush stroke in the concept of data replication.
Replication is exactly this process that you described there in the example that has a MySQL bank in the Front End and Oracle in the back end, that is, it is made a replication or synchronous or asynchronous load in the Oracle that treats the " thick "data.
The Oracle is able to treat this data better than the banks that are in the Front End, nor will I enter the merit of algorithms, high availability, among other factors.
In production environments that work with large data masses is interesting that we have heterogeneous banks of various flavors, many times the chosen ones to do these role in Front End is MySQL and Postgre.
but watches for non factors that can contribute to the use of these different banks:
- Performance;
- Licensing;
- Business model;
- $$$$$$$$$;
- Among other factors.
Obviously when we are talking about replication, we are talking about high availability.
A very simple replication case are those supermarket PDVs, on the front usually has a Postgre (in most cases by being open source) and mysql.
When we should have heterogeneous databases ? The answer depends, depends on your business model.
The more data, more processing and storage will need and this is not cheap.
I will cite a small example of the banks: the vast majority of banks, if not all, use mainframes to make processors of their data, to where I know, these mainframes are rented and this rent is charged CPU TIME, usually by You always have a DB2 or Oracle and in front of a bank that does not require as much performance as MySQL and Postgre.
As you say Brother the Fábio Telles from the Postgre Community: "Postgre is equal lover, everyone has, but no one has the courage to say that it has" rsrs ...
good is this, I hope have helped in some way.

[ccguedes]

Oops, thank you very much for the answer.
I will search Better on Data Replication!

Li on the net, on the issue that Oracle is limited to the number of connections, while those mentioned above, no longer has this restriction, proceeds? One of the reasons would this be this too?

I understand that while the purchase was not effective, the data is being treated on MySQL / Postg .. after effective go to the master base ... correct?

[adrianoturbo]

Brother, the number of connections is limited by the licensing number, this does not mean that Oracle is limited by number of connections, which will set this is the number of licenses and the server processing power to get an idea Oracle Engine can make 2048 connections simultaneously.
The TSE in the elections does not synchronously replication of the urns, the data is stored on a memory card and then made a load in the Oracle of this data, imagine all the million votes of Brazilians being replicated in real time, I would have to have a NASA server to transact as many data in real time.
[color=# 0040FF] "I understand that while the purchase was not effective, the data is being treated in MySQL / Postg .. after effected go to the master base ... correct? [/color] "
are being pre-stored and then challenges in the Oracle River.

[ccguedes]

Show, then see my statement, and if this is correct or wrong.

I have 500 tables in Oracle, these same 500 tables have to be on mysql / postgre? And only a few, the main ... copies
A customer that his last purchase was in 2000 and he only bought now in 2015, these data are recovered from where ...? Oracle? For it is an old customer, and as the base of the front bank is the "fresh", you will not have it there, TVz .. maybe.

[adrianoturbo]

Brother, I will answer your questions separately by topics:
[color=# 0040bf] "I understand that while the purchase has not been effective, the data is being treated in MySQL / Postg .. after effected go to the base Master ... correct? " [/color]
Brother, following the ACID properties, the data were saved and yes with the purchase made in the Front Ends banks, understand that the Front Ends banks are temporary storage bank for later They are replicated in Oracle, DB2 another bank that can work with large data mass.

[color=# 0040bf] "I have 500 tables in Oracle, these same 500 tables have to be in MySQL / Postgre? .Ou only a few, the main ... Copy" [/ color ]

Not necessarily all 500 these tables.

[color=# 0040bf]
"If a client that the last purchase of him was in 2000 and he only bought now in 2015, these data are recovered from where. .. from Oracle? Because it is an old customer, and as the base of Banco Front is the "fresh", it will not have it there, TVz .. maybe. "[/color]

Here we will enter another concept: the concept of DataWarehouse (DW) that stores historical data, making a cartesian product between table suit and table dimension.
Usually this information can be extracted directly from the Oracle database in production or through a separate DataWarehouse environment that is the most advisable.

[souldeath]

Looking from the database architecture point of view, I know that in Oracle is more expensive to open and close hundreds or thousands of concurrent connections, each connection represents a set of processes that are created in the operating system, in MySQL are several threads, each Open connection A new thread is created in case an e-commerce system is faster.

[ccguedes]

Thanks for the explanations