Progress Tracker

Dates	Progress
6 - 13 Sep	Literature review Data generator
13 - 20 Sep	Working to build the interface
20 - 27 Sep	Grace Hopper Conference
27 Sep - 4 Oct	Dynamic/live graphs from performance metrics Separate thread to continuously listen for tuning parameters
4 - 11 Oct	CMakeLists tie the pipeline together code cleanup
11 - 18 Oct	separate the tuning thread and workload thread
18 - 25 Oct	shared memory - for IPC between tuning interface and tuning params processes
25 Oct - 1 Nov	establish IPC between tuning_params.cc(tune rocksdb) and calculating_params.py(output from mlos) - used named pipes
1 - 8 Nov	Performance Graph correction Workload modification Integrate MLOS Draft Thesis Proposal
8 - 15 Nov	Establish communication between tune parameters and MLOS to signal tuning complete performance_graph modification Reduce Range workload
15 - 22 Nov	Restart workload in case of bad values for parameters
22 - 29 Nov	Thanksgiving Recess
29 Nov - 6 Dec	Handle background jobs when restarting the database because of bad values
Final Exams + Winter break
11 - 18 Jan	Restructure pipeline to handle intermittent failures during db restart
18 - 25 Jan	Add multiple tuning parameters
25 Jan - 1 Feb	Integrate different parameter types - Numeric, Categorical, Float
1 - 8 Feb	Bug fixes(dtype bug fix for pandas)
8 - 15 Feb	Setup project on Artemis machine
15 - 22 Feb	Bug Fix - write stalls Run basic experiments
22 - 29 Feb	Experiments - continued
29 Feb - 7 Mar	separate dynamic option setting via Options API and DbOptions API Experiments(continued) - add evaluation phase why do restarts happen? throughput metrics
7 - 14 Mar	change problem to maximize throughput debug negative compaction score

Full Progress Report

Thesis Draft