Compare Cloud Storage Filestore Persistent Disk

What this storage architecture is

This design maps each application workload to its optimal storage abstraction in Google Cloud: Cloud Storage for object storage, Filestore for fully managed network-attached file storage (NAS), and Persistent Disk for high-performance virtual block storage.

How it meets the requirements

• Workload 1 (Unstructured Archives): Cloud Storage provides scalable object storage optimized for unstructured media files and backups. It supports multi-region redundancy and automatic lifecycle management rules that seamlessly transition objects between Standard, Nearline, Coldline, and Archive tiers based on age or access criteria.
• Workload 2 (Shared POSIX File System): Filestore provides a managed NFS export that multiple Linux virtual machine instances can mount concurrently in read-write mode (ReadWriteMany), fully preserving standard POSIX file permissions and folder hierarchies without code refactoring.
• Workload 3 (Transactional Database): Persistent Disk provides durable network-attached raw block storage attached directly to a single VM instance. It can be formatted with journaling file systems like ext4 to ensure data integrity and maximize IOPS and low-latency performance for transactional database engines.

Key advantages and capabilities

• Workload-to-protocol optimization: Implements object APIs (REST/HTTP) for media, NFSv3/NFSv4 for shared file structures, and native block I/O for database storage.
• Cost efficiency: Lifecycle tiering in Cloud Storage lowers retention expenses for aging media assets.
• POSIX compliance: Filestore provides directory browsing, file locking, and standard Unix permissions required by multi-instance application servers.
• High durability: Persistent Disks offer transparent checksumming, scheduled snapshots, and multi-attach/regional replication options for high availability.

Why this is optimal

This combination aligns access protocols, latency expectations, and sharing semantics precisely with each workload's architectural profile, avoiding performance degradation or application redevelopment.