Project Darkstar's Roadmap Core Development Team Seeking Community Feedback
Project Darkstar has published a roadmap covering the next year or so to summarize the work currently being done for the next release, and the core technology development team wants to hear what the community at large thinks about its projected development path.
"If something is missing, or if some feature is more important than we think it is, tell us that we got the priorities wrong," writes Jim Waldo. "If we are working on something that isn't really important at all, tell us. We are always happy to not do something. And tell us what you think we should be doing after the work on the current map is done."
The roadmap for the basic multi-node Project Darkstar with performance and predictability feature set has a target release for version 1.0 of May 15, 2008.
Basic multi-node feature set to include:
Multiple machines can run the same game, sharing state
Data is shared across all nodes
Channels can include players located on different nodes
No load balancing or player movement; players are assigned to a node when they first log in and remain on that node until disconnected
New simplified channel interface
New wire protocol specification that reflects the simplified channel interface
Update simple client to reflect the new wire protocol and channel interface
Changes to the kernel to support only a single game in each VM/stack
Performance and predictability feature set to include:
IO throttling, allowing the IO system to put back-pressure on channels and sessions when the amount of IO threatens to overwhelm the system
Design and implementation of other throttling when needed for other resources.
Identification of performance bottlenecks in the multi-node system and the correction of those bottlenecks when possible
Construction of a set of benchmark programs that can be used to measure performance and identify any performance regressions
Optimization of the wire protocol
Updating of the client to reflect changes in the wire protocol
The team has also put together a future plan that includes:
Automatic Load Balancing/Multi-Node Scaling Feature Set: Administrative tools to allow tracking the load on individual machines; Tracking of what players are interacting with what objects and of communication patterns; Ability to move a player from one node to another without the client being involved; and Data caching and node-local optimizations for high concurrency
Full Fault-Tolerance Feature Set: Ability to dynamically add and remove machines from a Darkstar cluster; Redundant data store implementation; and Fault-tolerant implementations of the watchdog and node-mapping services
[...read more...]
Customized news reports about Sun Microsystems. Just the news you need, none of what you don't. 50,000+ Members. 20,000+ Articles Published since 1998.
Print Friendly
Email this article
tell us what you think we should be doing 
Project Darkstar's Roadmap
