7月14日-Heat Optimization-林冠宇(EN).pdf

Heat Optimization Rico Lin Chief OpenStack Technologist, inwinSTACK IRC: ricolin rico.l@ @ricolintw Placeholder Footer Copy / BU Logo or Name Goes Here Why NOT 80% adoption rates Orchestrates multiple composite cloud applications by using either the native HOT template format or the AWS CloudFormation template format, through both an OpenStack-native REST API and a CloudFormation-compatible Query API. ? Was not parallel excution ? Database monster ? Memory monster ? Don’t know how we can use it? Because! Convergence : Truly using every power we have Template Heat-API AMQP Heat-engine AMQP (engine_w orker) Worker DB B A Template Heat-API AMQP Heat-engine DB stackA stackA stackA stackA ? ? Start Newton, `convergence_engine` default to `True` ? num_engine_workers = max(4, multiprocessing.cpu_count()) ? (WIP)Migrate your old stack to convergence_stack convergence_engine = True Template Heat-API AMQP Heat-engine AMQP (engine_w orker) Worker DB More Database comsume when we change to Convergence Eager and Lazy Fetch ? What always been query right next to each other?(Eager fetch it!) – raw_template always after stack ? What we can took a chance and read if required(Lazy fetch it!) – stack_tag not always after stack Contexts Cache ? Only apply when `refresh = False` ? Replace database access Batch Task ? Accept Queue to process tasks in batch Template Heat-API AMQP Heat-engineAMQP (engine_worker) Worker DB TaskTaskTask TaskTask Task Task Task Task Task Task Optimize Sync Table C D BA SyncPoint leaves ? Resolving back to back retry – Random retry ? max_wait=conflict_count*0.01 ? wait=random(0, max_wait) – Exponential backoff ? wait=random(max_wait)*2^N+jitter Resolving Update Cancel ? Under develop ? handle_%s_cancel Reduce dramatically Placeholder Footer Copy / BU Logo or Name Goes Here Optimize your script More operate = More mistake 1 Operation 20 Min. 100 Operations 100 Min. VS Stack Stack = Parameters + Resources parameters: flavor: type: string ne