Featured White Papers

• Java Web Hosting | DailyRazor.com - Over 10yrs Of Java Web Hosting Exp!

Sponsored Links

Sponsored Links

Optimize with a SATA RAID Storage Solution
Range of capacities as low as $1250 per TB. Ideal if you currently rely on servers/disks/JBODs

Make room for JavaSpaces, Part 6

Build and use distributed data structures in your JavaSpaces programs

By Susanne Hupfer, JavaWorld.com, 10/02/00

As Figure 1 below shows, message passing and remote method invocation systems usually sequester data structures behind a centralized manager process; therefore, any process that wants to manipulate the structures will have to wait its turn and ask the manager to perform the operation on its behalf. In other words, multiple processes can't truly access a data structure simultaneously in these conventional systems.

Figure 1. Conventional systems barricade data behind a centralized
manager process (graphic courtesy of Sun Microsystems)

Distributed data structures take a very different approach that decouples the data from any one process. Distributed data structures are represented as collections of objects that, as I said earlier, can be accessed and modified by multiple processes concurrently. As Figure 2 suggests, processes can work on different pieces of the structure at the same time without getting in each other's way.

Figure 2. JavaSpaces supports concurrent access to distributed
data structures. (graphic courtesy of Sun Microsystems)

The design of any distributed data structure requires both a representation of the data structure and a protocol that processes follow to ensure that they can manipulate the structure safely and fairly. Now I'll explain how you can use entries to represent distributed data structures, and how you can use space operations to implement distributed protocols.

TEXTBOX: TEXTBOX_HEAD: Make room for JavaSpaces: Read the whole series!

• Part 1. Ease the development of distributed apps with JavaSpaces

• Part 2. Build a compute server with JavaSpaces

• Part 3. Coordinate your Jini apps with JavaSpaces

• Part 4. Explore Jini transactions with JavaSpaces

• Part 5. Make your compute server robust and scalable

• Part 6. Build and use distributed data structures in your JavaSpaces programs
• :END_TEXTBOX
  
  

Build distributed data structures with entries

Let's say you want to store an array of values in a space, and that you'd like the array to be accessible and modifiable by multiple processes simultaneously. For example, suppose your array holds today's highest temperatures for all 50 US states, and that weather stations in the various states will need to read and alter the values throughout the day.

The first approach you might think up is to store the array in an entry, like this:

public class HighTemps implements Entry {
    public Integer[] values;
    public HighTemps() {
    }
}

If a weather station wants to read or alter the high temperature for its state, it must read or remove the entire array. With this scheme, you haven't yet created a truly distributed array, since multiple processes cannot access or modify the array values simultaneously. The entry itself can become a bottleneck if many processes are trying to access the array concurrently.