class Strategy(strategy.StrategyBase):
def __init__(self, scoreboard, pump, model):
super(Strategy, self).__init__(scoreboard, pump, model, INTERVAL, START_WAIT)
# Setup migration queue
simple = True
self.migration_queue = MigrationQueue(self, simple, not simple)
# Build allocations
self.placement = initial_placement_dsap_helper.DSAPPlacement()
self.initial_migrations, self.active_server_info = self.placement.execute(NUM_BUCKETS,
lambda x: np.percentile(x, PERCENTILE),
MIGRATION_LIMIT)
# Current bucket
self.curr_bucket = 0
def start(self):
# Initialization time
self.time_null = self.pump.sim_time()
# Super call
super(Strategy, self).start()
def dump(self):
print 'DSAP controller - Dump configuration...'
logger.info('Strategy Configuration: %s' % json.dumps({'name' : 'DSAP-Strategy',
'start_wait' : START_WAIT,
'interval' : INTERVAL,
'num_bucketsl' : NUM_BUCKETS,
}))
def __run_migrations(self, bucket_index):
# Assignment
curr_assignment = self.placement.assignment_list[bucket_index]
# Previous assignment (based on model data - not uncertain calculation data)
# Calculated data might be different from model data due to failed migrations
# prev_assignment = self.placement.assignment_list[(bucket_index - 1) % NUM_BUCKETS]
prev_assignment = self.model.get_assignment_list()
for index_domain in curr_assignment.keys():
# Get data
domain_name = conf_domains.initial_domains[index_domain].name
source_node = conf_nodes.get_node_name(prev_assignment[index_domain])
target_node = conf_nodes.get_node_name(curr_assignment[index_domain])
# Find current node for domain
source_node = self.model.get_host_for_domain(domain_name).name
# Trigger migration
model_domain = self.model.get_host(domain_name)
model_source = self.model.get_host(source_node)
model_target = self.model.get_host(target_node)
self.migration_queue.add(model_domain, model_source, model_target)
def __run_optimized_migrations(self, bucket_index):
# Create allocations lists for GOAP
# Assignment
curr_assignment = self.placement.assignment_list[bucket_index]
# Previous assignment
prev_assignment = self.placement.assignment_list[(bucket_index - 1) % NUM_BUCKETS]
as_current = [0 for _ in xrange(conf_domains.count())]
as_next = [0 for _ in xrange(conf_domains.count())]
for index_domain in xrange(conf_domains.count()):
as_current[index_domain] = prev_assignment[index_domain]
as_next[index_domain] = curr_assignment[index_domain]
# Get current domain loads
domain_load = []
for mapping in conf_domains.initial_domains:
domain_name = mapping.domain
load = self.model.get_host(domain_name).mean_load(20)
domain_load.append(conf_nodes.to_node_load(load, conf_domainsize.DEFAULT))
# Schedule migrations
from ai import astar
migrations = astar.plan(conf_nodes.NODE_COUNT, as_current, as_next, domain_load)
# Trigger migrations
dep = None
for migration in migrations:
domain_name = conf_domains.initial_domains[migration[0]].domain
source_node = conf_nodes.get_node_name(migration[1])
target_node = conf_nodes.get_node_name(migration[2])
print 'domain %s - source %s - target %s' % (domain_name, source_node, target_node)
model_domain = self.model.get_host(domain_name)
model_source = self.model.get_host(source_node)
model_target = self.model.get_host(target_node)
# dep = self.migration_queue.add(model_domain, model_source, model_target, dep)
dep = self.migration_queue.add(model_domain, model_source, model_target)
return
def balance(self):
# Current bucket index
bucket_duration = TOTAL_EXPERIMENT_DURATION / NUM_BUCKETS
time = self.pump.sim_time() - self.time_null # relative time since end of ramp-up and start of steady-state
timeg = time # global time in TS data
bucket_index = int(timeg / bucket_duration) # always floor this value
bucket_time = bucket_duration - (timeg - bucket_duration * bucket_index)
# bucket_index %= NUM_BUCKETS
print 'bucket index %i' % bucket_index
print 'time till next bucket %i' % (bucket_time)
if bucket_index >= NUM_BUCKETS:
return
# Schedule migrations only once per bucket
if self.curr_bucket == bucket_index:
return
# Wait for migration queue to finish up all migrations
if not self.migration_queue.empty():
return
# Update current bucket status
self.curr_bucket = bucket_index
# Trigger migrations to get new bucket allocation
self.__run_migrations(self.curr_bucket)
# self.__run_optimized_migrations(self.curr_bucket)
def post_migrate_hook(self, success, domain, node_from, node_to, end_time):
node_from.blocked = self.pump.sim_time() - 1
node_to.blocked = self.pump.sim_time() - 1
self.migration_queue.finished(success, domain, node_from, node_to)
class Strategy(strategy.StrategyBase):
def __init__(self, scoreboard, pump, model):
super(Strategy, self).__init__(scoreboard, pump, model, INTERVAL, START_WAIT)
# Setup migration queue
simple = True
self.migration_queue = MigrationQueue(self, simple, not simple, True)
def start(self):
# Super call
super(Strategy, self).start()
def dump(self):
print 'DSAP controller - Dump configuration...'
logger.info('Strategy Configuration: %s' % json.dumps({'name' : 'DSAP-Strategy',
'start_wait' : START_WAIT,
'interval' : INTERVAL,
}))
def __run_migrations(self):
conversion_table = {}
prev_assignment = {}
i = 0
for node in self.model.get_hosts(types.NODE):
for domain_name in node.domains.keys():
conversion_table[i] = domain_name
prev_assignment[i] = conf_nodes.index_of(node.name)
i +=1
inverse_conversion_table = {domain_name : i for i, domain_name in conversion_table.iteritems()}
demand_cpu = {}
demand_mem = {}
for domain in self.model.get_hosts(types.DOMAIN):
if domain in conf_domains.initial_domains:
domain_size = conf_domains.initial_domains[conf_domains.initial_domain_index(domain.name)].size
else:
domain_size = conf_domains.available_domains[conf_domains.available_domain_index(domain.name)].size
domain_index = inverse_conversion_table[domain.name]
cpu_readings = domain.get_readings()
domain_load = conf_nodes.to_node_load(np.mean(cpu_readings[-NUM_CPU_READINGS:]), domain_size)
demand_cpu[domain_index] = domain_load
demand_mem[domain_index] = domain.domain_configuration.get_domain_spec().total_memory()
print 'domain : %d, demand_cpu : %d, demand_memory : %d' % (domain_index, domain_load, domain.domain_configuration.get_domain_spec().total_memory())
# Assignment
try:
_, curr_assignment = dsapp.solve(conf_nodes.NODE_COUNT,
conf_nodes.UTIL,
conf_nodes.NODE_MEM,
demand_cpu,
demand_mem,
prev_assignment,
MIG_OVERHEAD_SOURCE,
MIG_OVERHEAD_TARGET)
except:
print 'invalid solution #######################'
# don't change anything and just return in case the model was infeasible
return
assignment_changed = dsapp.AssignmentChanged(prev_assignment, curr_assignment)
print 'CHANGE in the Assignment : %s' % assignment_changed
if not assignment_changed:
# there is no change in the assignment, we can just return
logger.info("Returning because the previous assignment was optimal...")
return
for index_domain in curr_assignment.keys():
domain_name = conversion_table[index_domain]
source_node = conf_nodes.get_node_name(prev_assignment[index_domain])
target_node = conf_nodes.get_node_name(curr_assignment[index_domain])
# Find current node for domain
source_node = self.model.get_host_for_domain(domain_name).name
# Trigger migration
model_domain = self.model.get_host(domain_name)
model_source = self.model.get_host(source_node)
model_target = self.model.get_host(target_node)
self.migration_queue.add(model_domain, model_source, model_target)
def balance(self):
# Wait for migration queue to finish up all migrations
if not self.migration_queue.empty():
return
# Trigger migrations
self.__run_migrations()
def post_migrate_hook(self, success, domain, node_from, node_to, end_time):
node_from.blocked = self.pump.sim_time() - 1
node_to.blocked = self.pump.sim_time() - 1
self.migration_queue.finished(success, domain, node_from, node_to)