improve performance and should be scalable
I have written a code snippet, this code will be access from thousands of object and get method look up also will be done from these objects. I need help in improving performance of this code and should be scalable so that if more then 10000 objects are accessing it it should not effect performance
import java.util.Set;
import java.util.HashSet;
/**
* Implementation of the WidgetCatalog interface. Please see the interface
* for the specification.
*/
public class WidgetCatalogImpl implements WidgetCatalog
{
// using JDK 1.5
Set<Widget> set = new HashSet<Widget>();
public void add(Widget widget) {
set.add(widget);
}
/**
* Return the widget for the specified version of the named widget. If
* the exact flag is true then this will only return a widget for an
* exact name/version match. If the exact flag is false then this will
* return the version with a matching major version but the highest
* available minor number.
*
* @param name the name of the widget
* @param major the major version
* @param minor the minor version
* @param exact whether we need an exact match
*/
public Widget get(String name,int major,int minor,boolean exact) {
Widget widgetToReturn = null;
if(exact) {
Widget w = new Widget(name, major, minor);
// for loop using JDK 1.5 version
for(Widget wid : set) {
if((w.getName().equals(wid.getName())) && (wid.getVersion()).equals(w.getVersion())) {
widgetToReturn = w;
break;
}
}
} else {
Widget w = new Widget(name, major, minor);
WidgetVersion widgetVersion = new WidgetVersion(major, minor);
// for loop using JDK 1.5 version
for(Widget wid : set) {
WidgetVersion wv = wid.getVersion();
if((w.getName().equals(wid.getName())) && major == wv.getMajor() && WidgetVersion.isCompatibleAndNewer(wv, widgetVersion)) {
widgetToReturn = wid;
} else if((w.getName().equals(wid.getName())) && wv.equals(widgetVersion.getMajor(), widgetVersion.getMinor())) {
widgetToReturn = w;
}
}
}
return widgetToReturn;
}
/**
* Return the widget for the specified version of the named widget.
* This will return the newest version of the widget that is compatible,
* meaning the major version is the same but the minor version is equal to
* or higher than the desired version.
*
* @param name the name of the widget
* @param major the major version
* @param minor the minor version
*/
public Widget getBest(String name,int major,int minor) {
// THIS METHOD IS NOT GETTING USED ANY WHERE
return null;
}
}
Assuming you have more than ten widgets, instead of using a HashSet to store your widgets consider using a different data structure, such as a HashMap. Currently you are searching for widgets by brute-force, i.e. iterating over the set with average search times of n/2 (where the set contains n items). With a HashMap, search times are of the order of 1. A huge improvement.
Also, have you considered the effects of concurrent access? Would it be a problem if two or more clients added and requested the same widget type simultaneously? If so, then you must use a thread-safe data structure and synchronise method access.
Hope this helps.
Also, have you considered the effects of concurrent access? Would it be a problem if two or more clients added and requested the same widget type simultaneously? If so, then you must use a thread-safe data structure and synchronise method access.
Hope this helps.
Well, several ways of combining keys...
1. (Crude.) Use a concatenation (name and version?) as your key.
2. (Better.) Create a map of sets. I.e. A map that maps a widget name to a set, where a set retrieved from the first map contains widgets with the same name E.g.
HashMap<String, Set<Widget>> map;
This reduces the size of your brute-force search.
3. (Sophisticated.) Create a map of maps, or a map of lists. Where the inner maps use version number as a key or are sorted by version number!
1. (Crude.) Use a concatenation (name and version?) as your key.
2. (Better.) Create a map of sets. I.e. A map that maps a widget name to a set, where a set retrieved from the first map contains widgets with the same name E.g.
HashMap<String, Set<Widget>> map;
This reduces the size of your brute-force search.
3. (Sophisticated.) Create a map of maps, or a map of lists. Where the inner maps use version number as a key or are sorted by version number!
Correction for point 3. above:
3. (Sophisticated.) Create a map of maps, or a map of lists. Where the inner maps use version number as a key or are the inner list is sorted by version number!
3. (Sophisticated.) Create a map of maps, or a map of lists. Where the inner maps use version number as a key or are the inner list is sorted by version number!
Okay, here are some suggestions for a Map of Lists...
Modify the internal set and add method as code below. Then, in your get method, each for loop can be rewritten from
for(Widget wid : set) {
// ...
}
to
for(Widget wid : map.get(name)) {
// ...
}
Now, the time-consuming brute-force search has been restricted to widgets with the correct name.
Modify the internal set and add method as code below. Then, in your get method, each for loop can be rewritten from
for(Widget wid : set) {
// ...
}
to
for(Widget wid : map.get(name)) {
// ...
}
Now, the time-consuming brute-force search has been restricted to widgets with the correct name.
Map<String, List<Widget>> map = new HashMap<String, List<Widget>>();
public void add(Widget widget) {
if(map.containsKey(widget.getName())
map.get(widget.getName()).add(widget);
else {
List<Widget> newList = new Vector<Widget>(widget);
map.add(newList);
}
}
Premium Content
You need an Expert Office subscription to comment.Start Free Trial