Changed disk provisioning threshold for adjusting cassandra

service_capacity_modeling/models/org/netflix/cassandra.py

@@ -314,7 +314,12 @@ def _estimate_cassandra_cluster_zonal(
         ),
         # C* requires ephemeral disks to be 25% full because compaction
         # and replacement time if we're underscaled.
-        required_disk_space=lambda x: x * 4,
+        required_disk_space=lambda x: x * 2
+        if instance.drive is None
+        or "local" in instance.drive.drive_type.name
+        or required_cluster_size is None
+        or (x / required_cluster_size) < 1000
+        else x + (1000 * required_cluster_size),
         # C* clusters cannot recover data from neighbors quickly so we
         # want to avoid clusters with more than 1 TiB of local state
         max_local_disk_gib=max_local_disk_gib,

tests/netflix/test_cassandra.py

@@ -199,7 +199,7 @@ def test_high_write_throughput():
     assert (
         300_000
         > high_writes_result.count * high_writes_result.attached_drives[0].size_gib
-        >= 100_000
+        >= 70_000
     )
 
     cluster_cost = cap_plan.candidate_clusters.annual_costs["cassandra.zonal-clusters"]
@@ -353,3 +353,51 @@ def test_plan_certain():
     lr_clusters = cap_plan[0].candidate_clusters.zonal[0]
     assert lr_clusters.count == 8
     assert lr_clusters.instance.cpu == 16
+
+
+def test_plan_certain_disk_space():
+    """
+    Disk space is adjusted based on EBS and ephemeral. Takes lesser vCPUs. Correct
+    adjustment of disk space attributes to lower node density and hence provides
+    lower vCPUs.
+    """
+    cluster_capacity = CurrentClusterCapacity(
+        cluster_instance_name="i4i.8xlarge",
+        cluster_instance_count=Interval(low=32, mid=32, high=32, confidence=1),
+        cpu_utilization=Interval(low=2.96, mid=4.36, high=7.47, confidence=1),
+    )
+
+    worn_desire = CapacityDesires(
+        service_tier=1,
+        current_clusters=CurrentClusters(zonal=[cluster_capacity]),
+        query_pattern=QueryPattern(
+            access_pattern=AccessPattern(AccessPattern.latency),
+            estimated_read_per_second=Interval(
+                low=583, mid=933, high=2790, confidence=0.98
+            ),
+            estimated_write_per_second=Interval(
+                low=11310, mid=16491, high=20501, confidence=0.98
+            ),
+        ),
+        data_shape=DataShape(
+            estimated_state_size_gib=Interval(
+                low=71734.83, mid=72232.16, high=72385.33, confidence=0.98
+            ),
+            estimated_compression_ratio=Interval(low=1, mid=1, high=1, confidence=1),
+        ),
+    )
+    cap_plan = planner.plan_certain(
+        model_name="org.netflix.cassandra",
+        region="us-east-1",
+        num_results=3,
+        num_regions=4,
+        desires=worn_desire,
+        extra_model_arguments={
+            "current_asg_size": 3,
+            "max_disk_used_gib": 2520.8339739371695,
+            "max_local_disk_gib": 16384,
+            "required_cluster_size": 32,
+        },
+    )
+    lr_clusters = cap_plan[0].candidate_clusters.zonal[0]
+    assert lr_clusters.instance.cpu < 32

tests/netflix/test_cassandra_uncertain.py

@@ -101,7 +101,7 @@ def test_increasing_qps_simple():
         if lr.instance.drive is None:
             assert sum(dr.size_gib for dr in lr.attached_drives) >= 100
         else:
-            assert lr.instance.drive.size_gib >= 100
+            assert lr.instance.drive.size_gib >= 60
 
         result.append(
             (lr_family, lr_cpu, lr_cost, cap_plan.least_regret[0].requirements.zonal[0])
@@ -147,6 +147,7 @@ def test_worn_dataset():
         extra_model_arguments={
             "max_regional_size": 200,
             "copies_per_region": 2,
+            "require_attached_disks": True,
         },
     )
 

tests/netflix/test_key_value.py

@@ -53,7 +53,7 @@ def test_kv_increasing_qps_simple():
         if zlr.instance.drive is None:
             assert sum(dr.size_gib for dr in zlr.attached_drives) >= 100
         else:
-            assert zlr.instance.drive.size_gib >= 100
+            assert zlr.instance.drive.size_gib >= 60
 
         zonal_result.append(
             (

tests/netflix/test_time_series.py

@@ -108,7 +108,7 @@ def test_timeseries_increasing_qps_simple():
         if zlr.instance.drive is None:
             assert sum(dr.size_gib for dr in zlr.attached_drives) >= 100
         else:
-            assert zlr.instance.drive.size_gib >= 100
+            assert zlr.instance.drive.size_gib >= 60
 
         zonal_result.append(
             (