GenAIHub
← Back to Technical Section

Local SSD

Ultra-fast, ephemeral storage for high-performance workloads on Google Cloud

What is Local SSD?

Local SSD is a high-performance, low-latency ephemeral storage option available for Google Cloud Compute Engine virtual machines (VMs). Unlike persistent disks, Local SSDs are physically attached to the host server that runs your VM, providing extremely fast input/output operations per second (IOPS) and minimal latency. This makes Local SSD ideal for workloads that require rapid access to temporary data, such as caching, scratch processing, and high-throughput analytics.

Local SSDs are designed for temporary storage: data stored on these disks does not persist after the VM is stopped, deleted, or migrated. Each Local SSD partition is 375 GB, and you can attach up to 24 partitions (9 TB total) to a single VM, depending on the machine type. Local SSDs are commonly used for scenarios where performance is critical and data durability is not required.

Architecture

Compute Engine VM (Runs your workload) Local SSD (Ephemeral, high-speed storage) Physical Host (Server hardware)

Key Components

Local SSD Device

Physically attached SSD hardware providing ultra-fast, ephemeral storage directly to the VM instance.

Compute Engine VM

The virtual machine that utilizes Local SSD for high-performance, temporary data processing and caching.

Ephemeral Storage Layer

A non-persistent storage layer optimized for speed, ideal for workloads where data durability is not required.

Key Capabilities

Ultra-low Latency

Delivers extremely low latency and high IOPS, making it ideal for demanding workloads.

High Throughput

Supports rapid data transfer rates, suitable for analytics, media rendering, and AI/ML workloads.

Ephemeral Storage

Data is temporary and is deleted when the VM is stopped or deleted, making it perfect for scratch space and caches.

Common Use Cases

High-performance caching
Scratch disk for data analytics
Media rendering and transcoding
Temporary storage for AI/ML workloads
Database acceleration (e.g., NoSQL, in-memory DBs)
Big data processing (e.g., Hadoop, Spark)

Related Topics

Test Your Knowledge

Score 8/10 or higher to pass