Performance Optimization Flashcards
7 cards from real COA practice questions. Tap to flip, then mark Knew It or Still Learning — missed cards come back until you master them.
Read the first 7 Performance Optimization flashcards as text
Which qemu-kvm disk cache mode provides the best sequential write throughput for ephemeral disks when data durability on host crash is not required?
Answer: unsafe
The 'unsafe' cache mode skips all fsync calls and host page-cache flushing, yielding maximum write throughput but losing data on host crash.
What Nova feature allows pinning specific instance vCPUs to dedicated physical CPU cores to guarantee compute performance?
Answer: CPU pinning via hw:cpu_policy=dedicated
Setting hw:cpu_policy=dedicated in a flavor or image metadata causes Nova to pin each vCPU to an exclusive physical core, eliminating CPU contention.
Which OpenStack service is responsible for consolidating live migration of instances to free up underutilized compute hosts?
Answer: Watcher
Watcher is the OpenStack resource optimization service that analyzes cluster usage and triggers live migrations to consolidate workloads and free hosts.
In Neutron, which setting reduces agent polling overhead on large deployments with thousands of ports?
Answer: polling_interval in neutron-openvswitch-agent.ini
Increasing polling_interval reduces how frequently the OVS agent polls for port state changes, lowering CPU and RPC overhead at the cost of slightly slower reaction time.
Which Cinder volume type backend configuration option enables thin provisioning to optimize storage utilization?
Answer: lvm_type=thin in cinder.conf
Setting lvm_type=thin configures the LVM driver to use LVM thin pools, allowing volumes to consume only the storage they actually use rather than pre-allocating the full size.
What is the effect of setting 'scheduler_max_attempts' in Nova to a higher value on performance?
Answer: It increases the number of retry attempts for scheduling, which can increase latency but reduces NoValidHost failures
A higher scheduler_max_attempts lets Nova retry placement on more hosts before giving up, trading scheduling latency for a higher chance of successful placement.
Which protocol should be used for Cinder volume access when maximum IOPS and lowest latency for database workloads are required?
Answer: NVMe-oF (NVMe over Fabrics)
NVMe-oF provides the lowest latency and highest IOPS of any network block storage protocol by preserving the NVMe command set across the network fabric.