pluto
pluto copied to clipboard
snuspl
MIST: High-performance IoT Stream Processing
This PR resolves #1139 by : - Adding `ReplayServerAddress` and `ReplayServerPort` parameters - Modifying `start` method of `QueryStarter` class to `startOrSetupForReplay` because most of the code for both are the...
This PR resolves #1131 via * Implement recovery-based scaling out. * Add necessary avro formats and function calls.
I've found that enabling checkpointing leads to large memory use. A possible memory leak could exist in checkpointing and we need to solve this problem.
When a group is recovered, events must be replayed before the execution DAGs start receiving new events. Therefore the current implementation of the query `create` method must be separated into...
Current implementation of scale in/out is based on fault recovery, which can cause long latency. We need to make it more efficient in the later version.
In current PR, #1114, we don't support task recovery when a master is not recovered yet. We need to fix this issue after merging #1114.
We need to change query allocation to query assignment
In our current implementation, we store each query as a file to DFS. We should reduce the overhead.
We need to measure the recovery time to identify recovery speed of each policy.