Block, File, and Object Storage Types (Including
Protocols)
1. Introduction
Storage architectures are central to modern computing, each optimized for specific types of
workloads. Block, file, and object storage represent three fundamental types. They differ in
how data is organized, accessed, and transmitted—each using specific protocols and file
systems to meet unique performance, scalability, and availability needs.
2. Block Storage
2.1 Overview
Block storage divides data into equal-sized chunks (blocks), each addressed individually.
The operating system handles the organization of these blocks, which makes it function like
a traditional disk drive. It is ideal for high-performance and low-latency scenarios.
2.2 Protocols and Interfaces
1. SCSI (Small Computer System Interface)
Type: Command protocol (not a transport)
Function: Defines a standard set of commands and responses between host and
storage.
Usage: Still foundational; SCSI commands are encapsulated in iSCSI, FC, SAS.
Strengths: Mature, universal, highly compatible.
Variants:
o Parallel SCSI (legacy)
o SAS (Serial Attached SCSI)
o iSCSI (over TCP/IP)
o FCP (Fibre Channel Protocol)
2. iSCSI (Internet SCSI)
Type: Transport protocol
Layer: Works over TCP/IP
Encapsulates: SCSI commands into IP packets
Port: TCP 3260
Used in: SANs over standard Ethernet
Pros:
o Leverages existing infrastructure
o Widely supported by OSs and hypervisors
Cons:
o Relatively higher latency than native FC
o Performance depends on network quality
3. Fibre Channel (FC)
Type: Transport protocol
Layer: Data Link / Physical (OSI Layers 1–2)
Encapsulates: SCSI commands into FC frames
Used in: Enterprise SANs
Pros:
o High throughput (8/16/32/64 Gbps)
o Very low latency and reliable
Cons:
o Requires dedicated hardware (HBA, switches)
o Expensive and complex
4. FCoE (Fibre Channel over Ethernet)
Type: Converged transport protocol
Encapsulates: FC frames into Ethernet frames
Used in: Unified data and storage networks
Pros:
o Reduces cabling and hardware footprint
Cons:
o Requires Data Center Bridging (DCB) features
o Operational complexity
5. SAS (Serial Attached SCSI)
Type: Direct-attach protocol (internal)
Usage: High-speed interface for hard drives/SSDs inside servers/storage arrays
Pros:
o Full-duplex and faster than SATA
o Supports multiple devices on a single bus
Cons:
o Cable length limitations
6. NVMe-oF (NVMe over Fabrics)
Type: Transport for NVMe commands over network
Fabric Types:
o NVMe/TCP
o NVMe/RDMA
o NVMe/FC
Used in: High-performance, flash-optimized SANs
Pros:
o Ultra-low latency
o Designed for SSDs
Cons:
o Requires compatible infrastructure
7. RDMA (Remote Direct Memory Access)
Type: Communication method (used by iSER, NVMe-oF)
Function: Enables direct memory transfer between systems, bypassing CPU
Protocols Using RDMA:
o iSER (iSCSI Extensions for RDMA)
o NVMe/RDMA
Pros:
o Minimal CPU overhead
o Lowest possible latency
3. File Storage
3.1 Overview
File storage arranges data in directories and subdirectories. Users access it using traditional
file paths. Common in NAS systems.
3.2 Protocols and Systems
1. NFS (Network File System)
Platform: Unix/Linux
Developed by: Sun Microsystems
Versions:
o NFSv3: Stateless, fast, widely used
o NFSv4: Stateful, adds ACLs, encryption
Access Method: Mount remote file systems locally
Pros:
o Lightweight and efficient
Cons:
o Security limitations in earlier versions
2. SMB/CIFS (Server Message Block / Common Internet File System)
Platform: Windows (but also supported on Linux/macOS)
Developed by: Microsoft
Versions:
o SMB1 (legacy, insecure)
o SMB2 (improved performance)
o SMB3 (encryption, multichannel)
Use Cases: Home directories, print sharing, collaborative work
Pros:
o File locking, access control
Cons:
o Performance penalty on non-Windows systems
3. AFP (Apple Filing Protocol)
Platform: macOS (legacy)
Use: Apple’s native file sharing before SMB adoption
Status: Deprecated
4. ZFS
Type: File system and logical volume manager
Developed by: Sun Microsystems
Features:
o Snapshots, cloning, checksumming, deduplication
Protocol Integration: Often used with NFS, SMB for sharing
Use: NAS systems, enterprise backup, Proxmox/TrueNAS
5. HDFS (Hadoop Distributed File System)
Use Case: Big Data workloads
Characteristics:
o Optimized for large sequential reads/writes
o Data split into large blocks (e.g., 128MB)
Pros:
o Scalable across thousands of nodes
Cons:
o High latency
o Not ideal for small file access
6. Lustre
Type: Parallel file system
Use Case: HPC (High Performance Computing)
Features:
Concurrent access to multiple servers
Petabyte-scale capability
4. Object Storage
4.1 Overview
Object storage breaks data into objects that include the data, metadata, and a unique
identifier. It is accessed via APIs and is highly scalable.
4.2 Protocols
1. Amazon S3 API
De facto standard: Used by AWS and many on-premise solutions (MinIO, Ceph,
etc.)
API Operations: PUT, GET, DELETE, HEAD, LIST
Security: IAM roles, signed URLs, bucket policies
Use Case: Backup, archive, logs, cloud-native apps
2. OpenStack Swift
Type: Open-source object storage
Protocol: RESTful
Integration: OpenStack cloud suite
Use: Private cloud deployments
3. CDMI (Cloud Data Management Interface)
Standardized by: SNIA
Purpose: Standard API for cloud vendors and users
Use Case: Vendor-neutral storage integration
Adoption: Limited compared to S3
