This document describes concepts related to Pinecone indexes. To learn how to create or modify an index, see Manage indexes.
An index is the highest-level organizational unit of vector data in Pinecone. It accepts and stores vectors, serves queries over the vectors it contains, and does other vector operations over its contents. Each index runs on at least one pod.
Pods are pre-configured units of hardware for running a Pinecone service. Each index runs on one or more pods. Generally, more pods mean more storage capacity, lower latency, and higher throughput. You can also create pods of different sizes.
Once an index is created using a particular pod type, you cannot change the pod type for that index. However, you can create a new index from that collection with a different pod type.
Users on the Starter (free) plan are limited to 1 p1 pod or 1 s1 pod.
Different pod types are priced differently. See pricing for more details.
These storage-optimized pods provide large storage capacity and lower overall costs with slightly higher query latencies than p1 pods. They are ideal for very large indexes with moderate or relaxed latency requirements.
Each s1 pod has enough capacity for around 5M vectors of 768 dimensions.
These performance-optimized pods provide very low query latencies, but hold fewer vectors per pod than s1 pods. They are ideal for applications with low latency requirements (<100ms).
Each p1 pod has enough capacity for around 1M vectors of 768 dimensions.
This is a Public Preview ("Beta") feature. Test thoroughly before using this feature for production workloads. No SLAs or technical support commitments are provided for this feature.
The p2 pod type provides greater query throughput with lower latency. p2 pods support 200 QPS per replica and return queries in less than 10ms. This means that query throughput and latency are better than s1 and p1, especially for low dimension vectors (<512D).
However, unlike with p1 and s1 pods, filtering searches of indexes on p2 pods does not improve query latency.
Each p2 pod has enough capacity for around 1M vectors of 768 dimensions. However, capacity may vary with dimensionality.
Data ingestion for p2 pods is significantly slower than p1 and varies depending on dimensionality. Upsert at most 50 updates per second per p2 pod. Because query latency and throughput for p2 pods vary from p1 pods, test p2 pod performance with your dataset.
The p2 pod type does not support the
dotproduct distance metric.
The s1 and p1 pod type come in four sizes:
index storage and compute capacity doubles for each size step. The default
pod size is
x1. You can increase the size of a pod after index creation.
To learn about changing the pod size of an index, see Manage indexes.
You can choose from different metrics when creating a vector index:
- This is used to calculate the distance between two data points in a plane. It is one of the most commonly used distance metric. For an example, see our image similarity search example.
- When you use
metric='euclidean', the most similar results are those with the lowest score.
- This is often used to find similarities between different documents. The advantage is that the scores are normalized to [-1,1] range.
- This is used to multiply two vectors. You can use it to tell us how similar the two vectors are. The more positive the answer is, the closer the two vectors are in terms of their directions.
For the full list of parameters available to customize an index, see the create_index API reference.
Depending on your application, some metrics have better recall and precision performance than others. For more information, see: What is Vector Similarity Search?