Delete individual points or batches of points from your collection. Deleted points can be recovered only before you compact the collection.
Before you begin, make sure you have a running VectorAI DB instance and an existing collection with points. See Insert points to add data first. Delete a single point
The following example removes a single point by ID.
from actian_vectorai import VectorAIClient
COLLECTION = "products"
# Connect to VectorAI DB server
with VectorAIClient("localhost:50051") as client:
# Delete point with ID 11
client.points.delete_by_ids(COLLECTION, ids=[11])
print("Point deleted successfully")
Delete multiple points
Deleting a nonexistent point does not cause an error — the operation completes successfully regardless of whether the point exists. The following example removes multiple points at once.
from actian_vectorai import VectorAIClient
COLLECTION = "products"
# Connect to VectorAI DB server
with VectorAIClient("localhost:50051") as client:
# Delete multiple points
delete_ids = [9, 10, 12, 13]
client.points.delete_by_ids(COLLECTION, ids=delete_ids)
print(f"Deleted {len(delete_ids)} points")
Compact a collection
After deleting points, run a compaction operation to reclaim storage space and optimize the index. The compact() method is an alias for optimize().
Compaction permanently removes deleted points from storage, rebuilds the index structure, and reclaims disk space. Run compaction after large-scale deletions to maintain optimal performance.
Compaction is a resource-intensive operation that may temporarily impact collection performance. Run during low-traffic periods when possible.
from actian_vectorai import VectorAIClient
COLLECTION = "products"
# Connect to VectorAI DB server
with VectorAIClient("localhost:50051") as client:
# Delete points
client.points.delete_by_ids(COLLECTION, ids=[9, 10, 11, 12])
print("Deleted 4 points")
# Compact to reclaim storage
print("Compacting collection...")
client.vde.optimize(COLLECTION) # Optimize and reclaim space
print("Collection compacted successfully")
