Query is slow - large result - cassandra

Question

This table:
CREATE TABLE buckets (
year_month text,
source_id uuid,
nano_since_epoch bigint,
max double,
min double,
num_values int,
sum double,
value double,
PRIMARY KEY ((year_month, source_id), nano_since_epoch)
)
and this query:
select value from buckets where year_month = '2020_9' AND source_id=699ddcc0-5896-4b35-a901-8eec2f221499
It works, but the query takes ~4 sec. to complete. Result set has ~48.000 rows.
Cluster has 3 nodes running on GCP, n1-standard-2 (2 vCPUs, 7.5 GB memory).
Not using SSDs. Almost same performance on the second DC with only 1 node, on a much larger -bare metal- machine.
Table histograms:
Percentile SSTables Write Latency Read Latency Partition Size Cell Count
(micros) (micros) (bytes)
50% 0.00 51.01 73.46 182785 17084
75% 0.00 61.21 182.79 1629722 126934
95% 3.00 182.79 654.95 4055269 315852
98% 3.00 263.21 785.94 4866323 454826
99% 4.00 315.85 943.13 5839588 454826
Min 0.00 4.77 20.50 43 0
Max 4.00 2816.16 943.13 5839588 454826
Is it possible to speed up fetching of the results or is the size of the result the limiting factor here?
EDIT:
Cassandra is running in docker on the VMS. Memory utilization on the VM is ~ 50%
Dockerfile:
services:
elassandra:
image: strapdata/elassandra:6.2.3.23
container_name: elassandra
environment:
- CASSANDRA_BROADCAST_ADDRESS=10.164.0.58
- CASSANDRA_SEEDS=10.164.0.58
- CASSANDRA_CLUSTER_NAME=Prod Cluster 2
- CASSANDRA_NUM_TOKENS=256
- CASSANDRA_DC=datacenter-prod
- CASSANDRA_RACK=rack1
- LOCAL_JMX=no
volumes:
- '/opt/elassandra/data:/opt/elassandra/data'
- '/var/log/cassandra:/var/log/cassandra'
- '/var/lib/cassandra:/var/lib/cassandra'
ports:
- 7000:7000
- 7001:7001
- 7199:7199
- 8080:8080
- 8081:8081
- 9042:9042
- 9160:9160
- 9200:9200
- 9300:9300
ulimits:
memlock: -1
nproc: 32768
nofile: 100000
network_mode: "host"
logging:
driver: syslog
options:
tag: cassandra-0