This example presents a scenario, in which:
- 2 NVME devices exist, to be partitioned into 4 partitions each.
- 8 EBS devices exist, to be used as shadows for the 8 total NVME partitions.
- Create a cluster:
# aws
aerolab cluster create -n clusterName -c 2 -I r5ad.4xlarge -E 20,30,30,30,30,30,30,30,30
# gcp
aerolab cluster create -n clusterName -c 2 --instance c2d-standard-16 --zone us-central1-a --disk=pd-ssd:20 --disk=local-ssd --disk=local-ssd --disk=pd-ssd:30 --disk=pd-ssd:30 --disk=pd-ssd:30 --disk=pd-ssd:30 --disk=pd-ssd:30 --disk=pd-ssd:30 --disk=pd-ssd:30 --disk=pd-ssd:30
blkdiscard
and partition NVME: 25% disk space on each partition.
aerolab cluster partition create -n clusterName --filter-type=nvme -p 25,25,25,25
- prepare EBS by zeroing out the first 8MiB as required by Aerospike:
aerolab cluster partition create -n clusterName --filter-type=ebs
- Configure Aerospike to use devices for
test
namespace
aerolab cluster partition conf -n clusterName --namespace=test --filter-type=nvme --configure=device
- Configure EBS to be used as shadow devices, no partitions:
aerolab cluster partition conf -n clusterName --namespace=test --filter-type=ebs --filter-partitions=0 --configure=shadow
- Restart Aerospike:
aerolab aerospike stop -n clusterName
aerolab aerospike start -n clusterName