Introduction to Vertex AI Feature Store
1. Introduction to Vertex AI Feature Store
In this section, we will discuss the different storage methods available in Vertex AI Feature Store, and learn how to create, list, describe, update, and delete feature stores. Let's start with storage methods. Vertex AI Feature Store uses two storage methods, online storage and offline storage. All feature stores have offline storage and optionally online storage. Online storage retains the latest timestamp values of your features to efficiently handle online serving requests. Offline storage stores data until the data reaches the retention limit or until you delete the data. You can control offline storage costs by managing how much data you keep. You can also override the default online store data retention limit for your feature store and the offline data retention limit for an entity type. You can always view the amount of online and offline storage you are currently using through the Google Cloud Console. Another important concept here is online serving nodes, which are a type of virtual machine used to serve feature values to online requests. Online serving notes are always running and can handle a large number of requests per second. Online serving nodes are an important part of Vertex AI Feature Store because they offer many benefits, including low latency, scalability, and reliability. The number of online serving nodes that you require is directly proportional to the following two factors. The number of online serving requests, queries per second that the feature store receives, the number of ingestion jobs that write to online storage. You can switch between the following options to configure your number of online serving nodes. Auto-scaling. The feature store automatically changes the number of nodes based on CPU utilization. Allocating a fixed node count. Vertex AI Feature Store maintains a consistent number of nodes regardless of the traffic patterns. If you choose auto-scaling, you must consider four additional points. After adding online serving notes, the online store needs time to re-balance the data. If you've submit online serving requests to the feature store without online serving nodes, the operation returns an error. If you don't require online serving and want to prevent incurring charges for online serving nodes, set the number of online serving nodes to zero. If you set the number of online serving nodes to zero, the entire online store, including its data, is deleted. Now, let's look at how to create, list, describe, update, and delete feature stores. Note that the instructions provided here are just a general overview and it is recommended to refer to the official documentation for more detailed guidance on each operation. Create a feature store. To store entity types and features, you can create a feature store using the Google Cloud Console if a feature store isn't already created in the Google Cloud project for the selected region. If a feature store already exists for the project and region, you can use Terraform or send a post request by using the feature store create method, or client libraries such as Python, Java, or Node.js. Here are a couple of important notes about creating a feature store. The feature store location must match the location of your source data. You can ingest data from Cloud Storage buckets in the same location as the feature store, or in the US multi-region location. For big query, you can ingest data from tables in the feature stores location or in the US multi-region location. Ensure that your source data meets the requirements specified in the documentation. Create a feature store with customer managed encryption key, CMEK. You can also create a feature store with customer managed encryption key or CMEK, if you want to have full control over the encryption of your data. With CMEK, you can create and manage your own encryption keys which are used to encrypt data at rest and in transit. This gives you the peace of mind, knowing that your data is protected, even if Google systems are compromised. If you don't have an existing CMEK, set up a customer managed encryption key by using Cloud Key Management Service, KMS, and configure your appropriate permissions. Here are some specific reasons why you should use CMEK for your feature store. Compliance requirements. If you're required to encrypt their data in order to comply with regulations such as HIPAA or GDPR, CMEK can help you meet these requirements. Data sovereignty. If you want to keep their data within their own control or within the control of a specific region, CMEK might help you achieve this goal. Security. CMEK gives you the ability to manage your own encryption keys, which can provide an additional layer of security for your data. View feature store details. You can get details about a feature stores such as its name, an online serving configuration through the Vertex AI section of the Google Cloud Console or REST API. If you use the Google Cloud Console, you can also view Cloud monitoring metrics for feature stores. Delete a feature store. To delete a store that contains existing entity types and features, use the force query parameter. However, be mindful when using this parameter because it will irreversibly delete the feature store and its data. Before we move on to the next section, where we will review the main concepts you learned throughout this course, let's practice what we learned.2. Let's practice!
Create Your Free Account
or
By continuing, you accept our Terms of Use, our Privacy Policy and that your data is stored in the USA.