Selecting an Index Best Suited for Your Scenario
Most of the vector index types supported by Milvus use approximate nearest neighbors search (ANNS). Compared with accurate retrieval, which is usually very timeconsuming, the core idea of ANNS is no longer limited to returning the most accurate result, but only searching for neighbors of the target. ANNS improves retrieval efficiency by sacrificing accuracy within an acceptable range.
According to the implementation methods, the ANNS vector index can be divided into four categories:
 Treebased index
 Graphbased index
 Hashbased index
 Quantizationbased index
The following table classifies the indexes that Milvus supports:
Supported index  Classification  Scenario 

FLAT  N/A 

IVF_FLAT  Quantizationbased index 

IVF_SQ8  Quantizationbased index 

IVF_PQ  Quantizationbased index 

RNSG  Graphbased index 

HNSW  Graphbased index 

ANNOY  Treebased index 

Supported vector indexes
FLAT
For vector similarity search applications that require perfect accuracy and depend on relatively small (millionscale) datasets, the FLAT index is a good choice. FLAT does not compress vectors, and is the only index that can guarantee exact search results. Results from FLAT can also be used as a point of comparison for results produced by other indexes that have less than 100% recall.
FLAT is accurate because it takes an exhaustive approach to search, which means for each query the target input is compared to every vector in a dataset. This makes FLAT the slowest index on our list, and poorly suited for querying massive vector data. There are no parameters for the FLAT index in Milvus, and using it does not require data training or additional storage.

Search parameters
Parameter Description Range metric_type
[Optional] The chosen distance metric. See Supported Metrics.
IVF_FLAT
IVF_FLAT divides vector data into nlist
cluster units, and then compares distances between the target input vector and the center of each cluster. Depending on the number of clusters the system is set to query (nprobe
), similarity search results are returned based on comparisons between the target input and the vectors in the most similar cluster(s) only — drastically reducing query time.
By adjusting nprobe
, an ideal balance between accuracy and speed can be found for a given scenario. Results from the IVF_FLAT performance test demonstrate that query time increases sharply as both the number of target input vectors (nq
), and the number of clusters to search (nprobe
), increase.
IVF_FLAT is the most basic IVF index, and the encoded data stored in each unit is consistent with the original data.

Index building parameters
Parameter Description Range nlist
Number of cluster units [1, 65536] 
Search parameters
Parameter Description Range nprobe
Number of units to query CPU: [1, nlist]
GPU: [1, min(2048, nlist)]
IVF_SQ8
IVF_FLAT does not perform any compression, so the index files it produces are roughly the same size as the original, raw nonindexed vector data. For example, if the original 1B SIFT dataset is 476 GB, its IVF_FLAT index files will be slightly larger (~470 GB). Loading all the index files into memory will consume 470 GB of storage.
When disk, CPU, or GPU memory resources are limited, IVF_SQ8 is a better option than IVF_FLAT. This index type can convert each FLOAT (4 bytes) to UINT8 (1 byte) by performing scalar quantization. This reduces disk, CPU, and GPU memory consumption by 70–75%. For the 1B SIFT dataset, the IVF_SQ8 index files require just 140 GB of storage.

Index building parameters
Parameter Description Range nlist
Number of cluster units [1, 65536] 
Search parameters
Parameter Description Range nprobe
Number of units to query CPU: [1, nlist]
GPU: [1, min(2048, nlist)]
IVF_PQ
PQ
(Product Quantization) uniformly decomposes the original highdimensional vector space into Cartesian products of m
lowdimensional vector spaces, and then quantizes the decomposed lowdimensional vector spaces. Instead of calculating the distances between the target vector and the center of all the units, product quantization enables the calculation of distances between the target vector and the clustering center of each lowdimensional space and greatly reduces the time complexity and space complexity of the algorithm.
IVF_PQ performs IVF index clustering before quantizing the product of vectors. Its index file is even smaller than IVF_SQ8, but it also causes a loss of accuracy during searching vectors.

Index building parameters
Parameter Description Range nlist
Number of cluster units [1, 65536] m
Number of factors of product quantization dim ≡ 0 (mod m) nbits
[Optional] Number of bits in which each lowdimensional vector is stored. [1, 16] (8 by default) 
Search parameters
Parameter Description Range nprobe
Number of units to query [1, nlist]
RNSG
RNSG (Refined Navigating Spreadingout Graph) is a graphbased indexing algorithm. It sets the center position of the whole image as a navigation point, and then uses a specific edge selection strategy to control the outdegree of each point (less than or equal to out_degree
). Therefore, it can reduce memory usage and quickly locate the target position nearby during searching vectors.
The graph construction process of RNSG is as follows:
 Find
knng
nearest neighbors for each point.  Iterate at least
search_length
times based onknng
nearest neighbor nodes to selectcandidate_pool_size
possible nearest neighbor nodes.  Construct the outedge of each point in the selected
candidate_pool_size
nodes according to the edge selection strategy.
The query process is similar to the graph building process. It starts from the navigation point and iterate at least search_length
times to get the final result.

Index building parameters
Parameter Description Range out_degree
Maximum outdegree of the node [5, 300] candidate_pool_size
Candidate pool size of the node [50, 1000] search_length
Number of query iterations [10, 300] knng
Number of nearest neighbors [5, 300] 
Search parameters
Parameter Description Range search_length
Number of query iterations [10, 300]
HNSW
HNSW (Hierarchical Navigable Small World Graph) is a graphbased indexing algorithm. It builds a multilayer navigation structure for an image according to certain rules. In this structure, the upper layers are more sparse and the distances between nodes are farther; the lower layers are denser and the distances between nodes are closer. The search starts from the uppermost layer, finds the node closest to the target in this layer, and then enters the next layer to begin another search. After multiple iterations, it can quickly approach the target position.
In order to improve performance, HNSW limits the maximum degree of nodes on each layer of the graph to M
. In addition, you can use efConstruction
(when building index) or ef
(when searching targets) to specify a search range.

Index building parameters
Parameter Description Range M
Maximum degree of the node [4, 64] efConstruction
Search scope [8, 512] 
Search parameters
Parameter Description Range ef
Search scope [ top_k
, 32768]
ANNOY
ANNOY (Approximate Nearest Neighbors Oh Yeah) is an index that uses a hyperplane to divide a highdimensional space into multiple subspaces, and then stores them in a tree structure.
When searching for vectors, ANNOY follows the tree structure to find subspaces closer to the target vector, and then compares all the vectors in these subspaces (The number of vectors being compared should not be less than search_k
) to obtain the final result. Obviously, when the target vector is close to the edge of a certain subspace, sometimes it is necessary to greatly increase the number of searched subspaces to obtain a high recall rate. Therefore, ANNOY uses n_trees
different methods to divide the whole space, and searches all the dividing methods simultaneously to reduce the probability that the target vector is always at the edge of the subspace.

Index building parameters
Parameter Description Range n_trees
The number of methods of space division. [1, 1024] 
Search parameters
Parameter Description Range search_k
The number of nodes to search. 1 means 5% of the whole data. {1} ∪ [ top_k
, n ×n_trees
]
FAQ
What is the difference between FLAT index and IVF_FLAT index?
IVF_FLAT index divides a vector space into nlist
clusters. If you keep the default value of nlist
as 16384, Milvus compares the distances between the target vector and the centers of all 16384 clusters to get nprobe
nearest clusters. Then Milvus compares the distances between the target vector and the vectors in the selected clusters to get the nearest vectors. Unlike IVF_FLAT, FLAT directly compares the distances between the target vector and each and every vector.
Therefore, when the total number of vectors approximately equals nlist
, IVF_FLAT and FLAT has little difference in the way of calculation required and search performance. But as the number of vectors grows to two times, three times, or n times of nlist
, IVF_FLAT index begins to show increasingly greater advantages.
See How to Choose an Index in Milvus for more information.