org.apache.geronimo.clustering
Class Data
java.lang.Object
org.apache.geronimo.clustering.Data
- public class Data
- extends Object
An object capable of holding the state of multiple applications and
tiers. This will be a Map of tiers (web, ejb, ...). Each tier may
choose how to organise it's own Map. e.g. Web tier will use a Map
of webapps. Each webapp being represented by a Map of
HttpSessionID:HttpSession. Assuming that all IDs where GUIDs, it
would be posible to collapse all webapps together (and maybe all
tiers), thus avoiding dehashing overhead with each lookup coming
over the Cluster. I've considered this and decided that the extra
partitioning of the data that I am doing, will result in much less
contention on the Map used (particularly upon e.g. session
creation), furthermore, by retaining references into the
above-described structure, and distributing e.g. deltas across
e.g. webapp specific Channels, this dehashing overhead could be
avoided.
By using a Map of tiers, we avoid closing the set of
clusterable services, but incur a little more synchronisation
overhead on lookups, consider...
- Version:
- $Revision: 1.5 $ $Date: 2004/03/10 09:58:21 $
Field Summary |
protected static org.apache.commons.logging.Log |
_log
|
protected Map |
_tiers
|
Constructor Summary |
Data()
|
_log
protected static org.apache.commons.logging.Log _log
_tiers
protected Map _tiers
Data
public Data()
getTiers
public Map getTiers()
toString
public String toString()
Copyright © 2003-2005 Apache Software Foundation. All Rights Reserved.