Talk:JDBC driver

Latest comment: 13 years ago by SJK in topic type 5

Type 2 or 4 faster? edit

What's faster? This article suggests that type 4 (pure Java/thin) is faster, however some oracle guys claim type2/oci to be way faster and type 4 is only to be used for applets where you can't ship oci with.

Also maybe a list of type 2 drivers for promiment databases should be featured (is there really anything other than oci?) —Preceding unsigned comment added by 81.189.19.54 (talk) 19:03, 3 April 2008 (UTC)Reply

Regarding speed: It really depends on the driver and the DB you're connecting to. OCI may be the fastest for Oracle but there are certainly many more RDBMS out there!

In MS SQL Server, my company's testing has shown jTDS (a free type 4 driver) to be faster than other drivers (type 2, type 3, etc.). However, we have seen limited cases where a type 3 driver is faster with some queries. Across a wide range of queries jTDS is either slightly better or vastly superior to the type 3 driver. But, the cases where it's worse, it is orders of magnitude worse. For us, that is a problem since those areas are frequently accessed. As a result our software uses BOTH drivers to connect and manages connections so that the type 3 driver is used where the type 4 driver is slower. 24.62.81.239 (talk) 03:38, 20 May 2008 (UTC)Reply

Since the question of which driver type is faster depends on the database & driver implementation at least, the statement that Type 2 is faster than Type 4 is incorrect and should be changed? 210.18.83.173 (talk) 08:09, 15 September 2009 (UTC)Reply

Merger edit

Type 4, bullet point 1 edit

This bullet point implies that there the following necessary:

  1. That a 'server' be involved.
  2. That a network connection must be made.

I disagree - a type 4 JDBC driver does not necessarily have to be reliant on an external, network DB. Am I wrong? 67.23.168.161 02:06, 16 December 2006 (UTC)Reply

You're correct. While most Type 4 drivers are going to open a Socket and communicate that way to the external DB, there is nothing that actually mandates that this be the case. In fact, it is not uncommon the DB to be on the same box. I've cleaned the section up.24.62.81.239 (talk) 03:42, 20 May 2008 (UTC)Reply

Type-3 terminology edit

In both the description and diagram under type-3, the middle-tier piece of this driver is referred to as an 'Application Server'. I don't believe this is a correct usage of this term. The middle-tier of a type-3 archetecture is typically concerned with servicing driver connections to datasources and from clients. It's not a full blown application server, it isn't even necessarily Java-based.

I suggest keep using a term-like 'middle-tier', and not use 'application server' for this component. --208.204.155.241 16:15, 18 December 2006 (UTC)Reply

Sparse Introduction edit

The introduction doesn't even mention that this is for the Java programming language! --81.171.161.2 (talk) 09:58, 29 February 2008 (UTC)Reply

Have started to tidy this article up. Minkythecat (talk) 12:08, 9 June 2008 (UTC)Reply

Type 2 disadvantages correction edit

One of the disadvantages cited of the type 2 drivers is 'Cannot be used in web-based application due the client side software needed'. This should be changed to be more specific: 'Cannot be used in applet-based application due to the client side software needed' since a type 2 driver will work with server-based web application.Quietgenie (talk) 21:27, 10 February 2009 (UTC)Reply

List of JDBC Drivers edit

I have included a list of drivers and an opensource perf banchmark. May be we can start a list of results run on different machines and rdbms with jdbc drivers. —Preceding unsigned comment added by Hermandr (talkcontribs) 09:29, 5 January 2010 (UTC)Reply

type 5 edit

Deleted section on "type 5". There is no clear definition of what a type 5 architecture is, unlike types 1-4. This is not officially terminology adopted through the JCP. It seems to be to be mostly an invention of Progress Software's marketing department. --SJK (talk) 10:54, 30 October 2010 (UTC)Reply