Execution Profiles (experimental)

Execution profiles are an experimental API aimed at making it easier to execute requests in different ways within a single connected Session. Execution profiles are being introduced to deal with the exploding number of configuration options, especially as the database platform evolves more complex workloads.

The Execution Profile API is being introduced now, in an experimental capacity, in order to take advantage of it in existing projects, and to gauge interest and feedback in the community. For now, the legacy configuration remains intact, but legacy and Execution Profile APIs cannot be used simultaneously on the same client Cluster.

This document explains how Execution Profiles relate to existing settings, and shows how to use the new profiles for request execution.

Mapping Legacy Parameters to Profiles

Execution profiles can inherit from cluster.ExecutionProfile, and currently provide the following options, previously input from the noted attributes:

When using the new API, these parameters can be defined by instances of cluster.ExecutionProfile.

Using Execution Profiles

Default

from dse.cluster import Cluster
cluster = Cluster()
session = cluster.connect()
local_query = 'SELECT rpc_address FROM system.local'
for _ in cluster.metadata.all_hosts():
    print session.execute(local_query)[0]
Row(rpc_address='127.0.0.2')
Row(rpc_address='127.0.0.1')

The default execution profile is built from Cluster parameters and default Session attributes. This profile matches existing default parameters.

Initializing cluster with profiles

from dse.cluster import ExecutionProfile
from dse.policies import WhiteListRoundRobinPolicy

node1_profile = ExecutionProfile(load_balancing_policy=WhiteListRoundRobinPolicy(['127.0.0.1']))
node2_profile = ExecutionProfile(load_balancing_policy=WhiteListRoundRobinPolicy(['127.0.0.2']))

profiles = {'node1': node1_profile, 'node2': node2_profile}
session = Cluster(execution_profiles=profiles).connect()
for _ in cluster.metadata.all_hosts():
    print session.execute(local_query, execution_profile='node1')[0]
Row(rpc_address='127.0.0.1')
Row(rpc_address='127.0.0.1')
for _ in cluster.metadata.all_hosts():
    print session.execute(local_query, execution_profile='node2')[0]
Row(rpc_address='127.0.0.2')
Row(rpc_address='127.0.0.2')
for _ in cluster.metadata.all_hosts():
    print session.execute(local_query)[0]
Row(rpc_address='127.0.0.2')
Row(rpc_address='127.0.0.1')

Note that, even when custom profiles are injected, the default TokenAwarePolicy(DCAwareRoundRobinPolicy()) is still present. To override the default, specify a policy with the EXEC_PROFILE_DEFAULT key.

from dse.cluster import EXEC_PROFILE_DEFAULT
profile = ExecutionProfile(request_timeout=30)
cluster = Cluster(execution_profiles={EXEC_PROFILE_DEFAULT: profile})

Adding named profiles

New profiles can be added constructing from scratch, or deriving from default:

locked_execution = ExecutionProfile(load_balancing_policy=WhiteListRoundRobinPolicy(['127.0.0.1']))
node1_profile = 'node1_whitelist'
cluster.add_execution_profile(node1_profile, locked_execution)

for _ in cluster.metadata.all_hosts():
    print session.execute(local_query, execution_profile=node1_profile)[0]
Row(rpc_address='127.0.0.1')
Row(rpc_address='127.0.0.1')

See Cluster.add_execution_profile() for details and optional parameters.

Passing a profile instance without mapping

We also have the ability to pass profile instances to be used for execution, but not added to the mapping:

from dse.query import tuple_factory

tmp = session.execution_profile_clone_update('node1', request_timeout=100, row_factory=tuple_factory)

print session.execute(local_query, execution_profile=tmp)[0]
print session.execute(local_query, execution_profile='node1')[0]
('127.0.0.1',)
Row(rpc_address='127.0.0.1')

The new profile is a shallow copy, so the tmp profile shares a load balancing policy with one managed by the cluster. If reference objects are to be updated in the clone, one would typically set those attributes to a new instance.