In Java/j2ee you can use whatever design you want, you can follow all the big core J2EE patterns or use "looser" ones.
In fact it's a matter of design, not platform. In Java you have much MORE design/idioms choices, because of the maturity:
- You can use (or not) IoC stuff: Spring, Hivemind, Pico/Nano, your own - You can use (or not) AOP: AspectJ, AspectWerkz, Jboss AOP, etc. - You can use (or not) OR mapping: from EJB to hibernate, to iBatis, to your own. - you can use (or not) MVC frameworks: Struts, Webwork, JSF, etc, your own - You can persist on RDBMS, ODBMS, persistent collections (Berkeley DB), prevaylence system (prevayler), file system
And you can usually mix and match all this stuff. Some of this stuff (this is by far complete... only an example) is being ported to.net.
It's Prevayler, but is not running in Ramdisk, it's simple Java objects in RAM.
Basically it's based on "serializing" commands (transactions) vs the live object and logging the commands themselves on disk to survive RAM failures.
The assumption here is that the total RAM is enough to contain the whole "database". Being the "queries" simple in-vm method calls is in some situations thousands times faster that ORCL.
I have used it in 2 different projects and it's quite impressive.
Here you can find more infos: http://www.prevayler.org
Slashdot Mirror
too structured doesn't mean anything.
.net.
In Java/j2ee you can use whatever design you want, you can follow all the big core J2EE patterns or use "looser" ones.
In fact it's a matter of design, not platform. In Java you have much MORE design/idioms choices, because of the maturity:
- You can use (or not) IoC stuff: Spring, Hivemind, Pico/Nano, your own
- You can use (or not) AOP: AspectJ, AspectWerkz, Jboss AOP, etc.
- You can use (or not) OR mapping: from EJB to hibernate, to iBatis, to your own.
- you can use (or not) MVC frameworks: Struts, Webwork, JSF, etc, your own
- You can persist on RDBMS, ODBMS, persistent collections (Berkeley DB), prevaylence system (prevayler), file system
And you can usually mix and match all this stuff.
Some of this stuff (this is by far complete... only an example) is being ported to
It's Prevayler, but is not running in Ramdisk, it's simple Java objects in RAM.
Basically it's based on "serializing" commands (transactions) vs the live object and logging the commands themselves on disk to survive RAM failures.
The assumption here is that the total RAM is enough to contain the whole "database". Being the "queries" simple in-vm method calls is in some situations thousands times faster that ORCL.
I have used it in 2 different projects and it's quite impressive.
Here you can find more infos: http://www.prevayler.org
There should be also portings in other languages.