Skip to content

Latest commit

 

History

History
46 lines (32 loc) · 1.71 KB

File metadata and controls

46 lines (32 loc) · 1.71 KB
title Migrating from Memory Stores to Vector Stores
description Describes how to migrate from using memory stores to vector stores in semantic kernel.
zone_pivot_groups programming-languages
author westey-m
ms.topic conceptual
ms.author westey
ms.date 03/06/2025
ms.service semantic-kernel

::: zone pivot="programming-language-csharp"

Migrating from Memory Stores to Vector Stores

Semantic Kernel provides two distinct abstractions for interacting with vector stores.

  1. A set of legacy abstractions where the primary interface is Microsoft.SemanticKernel.Memory.IMemoryStore.
  2. A new and improved set of abstractions where the primary abstract base class is Microsoft.Extensions.VectorData.VectorStore.

The Vector Store abstractions provide more functionality than what the Memory Store abstractions provide, e.g. being able to define your own schema, supporting multiple vectors per record (database permitting), supporting more vector types than ReadOnlyMemory<float>, etc. We recommend using the Vector Store abstractions instead of the Memory Store abstractions.

Tip

For a more detailed comparison of the Memory Store and Vector Store abstractions see here.

Migrating from Memory Stores to Vector Stores

See the Legacy Semantic Kernel Memory Stores page for instructions on how to migrate.

::: zone-end ::: zone pivot="programming-language-python"

Coming soon

More info coming soon.

::: zone-end ::: zone pivot="programming-language-java"

Coming soon

More info coming soon.

::: zone-end