Object storage paths

更新时间:
复制 MD 格式

GanosBase supports reading vector data stored in Alibaba Cloud Object Storage Service (OSS) buckets, self-managed MinIO buckets, and Hadoop Distributed File System (HDFS) clusters. This topic describes the path format for each storage type.

OSS paths

Use the following format to specify a vector data path in an OSS bucket:

oss://<access_id>:<secret_key>@[<endpoint>]/<bucket>/path_to/file

Parameters

ParameterDescription
access_idThe AccessKey ID of the AccessKey pair used to access the OSS bucket.
secret_keyThe AccessKey secret of the AccessKey pair. To get your AccessKey pair, see Obtain an AccessKey pair.
endpoint(Optional) The endpoint of the OSS bucket. For available endpoints, see OSS endpoints.
bucketThe name of the OSS bucket.
path_to/fileThe path to the object in the OSS bucket.

Usage notes

  • Use the internal endpoint to make sure GanosBase can access the OSS bucket. The database cluster and the OSS bucket must be in the same region.

  • To use a Shapefile, upload at least the .shp, .shx, and .dbf files to the same folder in the OSS bucket.

MinIO paths

Note

MinIO paths are only supported in hybrid cloud environments.

Use the following format to specify a vector data path in a MinIO bucket:

mio://<id>:<key>@<host>[:<port>]/<bucket>/path_to/file

Parameters

ParameterDescription
idThe access ID for the MinIO bucket.
keyThe access key for the MinIO bucket.
hostThe host address of the MinIO server.
port(Optional) The port of the MinIO server. If not specified, the HTTPS port is used.
bucketThe name of the MinIO bucket.
path_to/fileThe path to the file in the MinIO bucket.

Usage notes

  • To achieve high data transfer performance, the database cluster and the MinIO bucket must be in the same region and communicate over an internal network.

HDFS paths

Note

HDFS paths are only supported in hybrid cloud environments.

Use the following format to specify a vector data path in an HDFS cluster:

HDFS://<user_name>@<host>[:<port>]/path_to/file

Parameters

ParameterDescription
user_name(Optional) The HDFS username. Defaults to hadoop if not specified.
hostThe host address of the Remote Procedure Call (RPC) endpoint, as defined in the HDFS configuration file.
port(Optional) The RPC port, as defined in the HDFS configuration file.
path_to/fileThe path to the file in HDFS.

Usage notes

  • To achieve high data transfer performance, the database cluster and the HDFS cluster must be in the same Virtual Private Cloud (VPC) and communicate over an internal network.