Extending the Java Runtime: Plug-in Capacity and Availability for Java Technology
TS-4219
Presenter: Jonas Boner and Ari Zilka, Terracotta
Today most clustering and caching products on the market are API based - basically designed around map.get() and map.put(). Unfortunately a design like that has some negative side effects.
For example it:
- Breaks Object Identity (e.g. Java technology's pass-by reference semantics)
- Perturbs your Domain Model - developers are forced to layer some kind of primary-key mechanism onto their domain objects, which forces them to think like relational database designers
- Tangles the business logic with clustering code
- Developers can not trust regular Java thread coordination anymore (e.g. synchronized(), notify(), wait() etc. - but need additional coordination constructs)
- Requires serialization of whole object graphs - cannot do fine-grained updates, which affects scalability
Can it be done better?
Come to the talk and see that by preserving Object Identity and extending Java technology's Memory Model and thread coordination semantics, we can actually address all these problems transparently.
Watch The Session
You need to be a registered Sun Developer Network member to view this multimedia session. If you are a registered SDN member, please click on "Watch Multimedia" button to log-in to view the multimedia session. If you wish to join SDN, please click here.
|
