COA Performance Optimization 2 — Questions and Answers
Question 1: Which Nova scheduler filter ensures instances are placed on hosts with sufficient free RAM before considering CPU and disk resources?
- RamFilter (Correct answer)
- ComputeFilter
- CoreFilter
- DiskFilter
Correct answer: RamFilter
RamFilter is applied first in the default scheduler filter chain to ensure hosts meet the instance's RAM requirements.
Question 2: What is the recommended Ceph client-side caching mode for OpenStack Cinder volumes to maximize write performance while maintaining data integrity?
- writeback (Correct answer)
- writethrough
- none
- directsync
Correct answer: writeback
Writeback mode caches writes locally before flushing to Ceph, providing the best write performance while the OSDs still ensure durability.
Question 3: Which parameter in nova.conf controls the maximum number of simultaneous live migrations per compute host?
- max_concurrent_live_migrations (Correct answer)
- live_migration_bandwidth
- live_migration_completion_timeout
- max_live_migration_workers
Correct answer: max_concurrent_live_migrations
max_concurrent_live_migrations limits parallel live migrations per compute node to prevent network and CPU saturation.
Question 4: An OpenStack environment shows high latency on Neutron API calls. Which component should be scaled horizontally first to address this?
- neutron-server (Correct answer)
- neutron-openvswitch-agent
- neutron-dhcp-agent
- neutron-metadata-agent
Correct answer: neutron-server
neutron-server handles all API requests and is the primary bottleneck for API latency; scaling it horizontally behind a load balancer reduces response times.
Question 5: Which Oslo messaging transport protocol offers the lowest latency for RPC calls between OpenStack services?
- RabbitMQ with direct exchange
- Kafka
- ZeroMQ (Correct answer)
- HTTP-based oslo.messaging
Correct answer: ZeroMQ
ZeroMQ (zmq) eliminates the broker entirely, resulting in the lowest RPC latency for direct service-to-service communication.
Question 6: What does enabling 'reserved_host_memory_mb' in Nova accomplish for performance optimization?
- Reserves host RAM so the hypervisor and OS retain enough memory, preventing instance scheduling on overloaded hosts (Correct answer)
- Allocates memory to instance swap space
- Sets the maximum memory per tenant
- Reserves NUMA-local memory for SR-IOV
Correct answer: Reserves host RAM so the hypervisor and OS retain enough memory, preventing instance scheduling on overloaded hosts
reserved_host_memory_mb subtracts a fixed amount from the host's reported available RAM so Nova won't overcommit the hypervisor OS itself.
Question 7: Which Keystone token provider type has the best performance for large-scale deployments because it requires no database lookup on validation?
- fernet (Correct answer)
- uuid
- pki
- saml
Correct answer: fernet
Fernet tokens are cryptographically self-contained, so Keystone can validate them without a database round-trip, dramatically reducing validation latency.
COA Performance Optimization 2
Which Nova scheduler filter ensures instances are placed on hosts with sufficient free RAM before considering CPU and disk resources?