iSCSI vs. NFS: 5 Key Comparisons

iSCSI works by transmitting block-level data between an iSCSI initiator (placed on a server) and an iSCSI target (placed on a storage device). Once the packet reaches the iSCSI target, it is disassembled, and the SCSI commands are separated by the protocol. This allows the storage to be visible using any operating system.

Unlike its alternatives (such as fiber channels),  iSCSI can work on existing IP infrastructure without dedicated cabling. As a result, it can serve as a low-cost SAN option.

iSCSI can establish communications with arbitrary SCSI device types. This protocol is widely used by system administrators to set up servers for disk volume access on storage arrays. However, performance issues may arise if iSCSI is not deployed on a dedicated network or subnet.

The client-side may issue two types of requests: read requests and write requests. Read requests are issued when the client wants to read the data on the server. Write requests are issued to the server when the client computer needs to write over the existing data. The read and write requests are implemented using the standard read/write operations. The server computer completes the request by leveraging the corresponding protocol. The data is then returned to the client computer.

Data requests from NFS clients are transmitted through the NFS server. The server retrieves the requested data from the storage and transmits it back to the clients.

Shared file locking is a key software feature of NFS. Shared file access can be implemented by properly specifying both file locking and caching parameters. If the user fails to specify these parameters and file data is only retained in a host cache, all NFS storage clients use the same locking and caching parameters for mounted files.

In cases where multiple computers or threads attempt to access one file simultaneously, the shared file access feature may malfunction. The file locking mechanism was developed to improve the efficiency of shared file access functionality. Shared file access can be executed within a single host or among several hosts, with NFS being used for accessing the same file.

The iSCSI initiator is the host-based hardware or software component. Deployed on the server, this component enables data transmission to and from the storage array. The source array is also capable of serving as a data migration initiator among the storage arrays. The storage network can be created using standard Ethernet components for the software initiator. iSCSI initiators manage several parallel communication links to several targets at once.

The iSCSI target is the component deployed on the storage side. It essentially plays the role of a server that hosts storage resources and allows storage access. iSCSI targets are basically the storage resources within an iSCSI server. They generally represent hard disk storage and are usually accessed via Ethernet.

Targets are data providers and include tape libraries and disk arrays. They expose one or more SCSI logical unit numbers (LUNs) to specific iSCSI initiators. However, iSCSI targets are the logical entities within the context of enterprise storage. iSCSI targets manage several parallel communication links to several initiators.

Next comes the iSCSI HBA, similar to a fiber channel. It offloads computing responsibilities from the system processor. iSCSI HBA helps enhance server network and storage performance but can cost more than a standard Ethernet NIC.

Finally, the iSCSI iSOE can be a good alternative for iSCSI HBA, as it provides similar functionality at a lower cost.

NFS operations leverage three main components, which, logically speaking, reside at the three OSI model layers corresponding to the TCP/IP application layer.

• Remote procedure call (RPC) is a generic session layer service to deploy client/server internetworking. It enables programs that call local services on the host computer to call the service on a remote device over a network.

• External data representation (XDR) is a descriptive language to define data types consistently. Conceptually speaking, XDR is located at the presentation layer, and its universal representations enable NFS data exchange among computers with different internal data storage methods.

• NFS procedures and operations are the implementation of the actual functionality of NFS. Conceptually, these procedures function at layer seven of the OSI model, specifying particular tasks on files via a network link. Here, RPC executes the commands across the internetwork, while XDR is used to represent data.

The above three key components or “subprotocols” represent most of the NFS protocol. Apart from them, the protocol includes numerous other functions. Of these, the key ones are highlighted below.

• Mount protocol: NFS creators specifically decided not to have the protocol engage within the specifics of the file opening and closing process. Instead, the Mount protocol is a separate protocol established for this function. Before accessing a file or other resource via a network, it must be mounted using this protocol. While the Mount protocol is architecturally separate, it is related to NFS, so much so that it is defined in the appendix of the NFS standard.

• NFS file system model: NFS uses a specific model to implement the file structure and directory of the systems that rely on it. While this model is based on the UNIX file system model, it is not specific to that operating system.

• Security: Finally, version 2 and version 3 of NFS come with limited security functionality and leverage UNIX-style authentication to check permissions for different operations. However, version 4 greatly enhances security for NFS implementations, including both encryption and authentication.

A key advantage of iSCSI is its use of TCP/IP, which allows for long-distance IP routing without external gateway hardware. It also provides a large storage network environment and increased flexibility.

Standard Ethernet

Using Standard Ethernet by iSCSI means that the protocol does not require expensive components to be built and deployed.

Storage array

A large storage array for iSCSI targets can be either open-source software or commercial. Unique iSCSI targets are provided for numerous clients.

Security

Internet security protocol is leveraged to secure IP network traffic by encrypting and authenticating each data packet received.

RPC is available for servers as well as clients. It replaces the transport device interface for enhanced scalability and support.

Multiple port extensions support RPC ports that are easy to use at the client level and compatible with firewalls.

Firewall compatibility is a key advantage of NFS version 4 and uses TCP Port -2049 for service execution. This simplifies protocol usage across firewalls.

Finally, NFS is a kerberized file system interface and features additional Kerberos privacy, such as Krb5p, to support krb5, krb5i, and other existing Kerberos options.

iSCSI is primarily designed for Microsoft Windows.

This protocol facilitates block-level sharing, allowing connected devices to access and utilize storage resources at the block level, similar to a local hard drive.

In an iSCSI setup, the responsibility of managing the file system lies with the guest operating system. This means that the guest OS handles tasks related to the file system, such as organizing and managing files and directories.

With iSCSI, each volume on the block level can be accessed by a single client, ensuring dedicated access and control over the storage resources.

In iSCSI, the file system is implemented at the client level. This enables both data and metadata to be read and managed within the client file system.

Implementing iSCSI can be slightly challenging as it requires configuring hosts, storage options, virtual local area networks (VLAN), and other related settings to ensure proper functionality and integration with the system.

NFS can be used for Microsoft Windows, Linux, and UNIX operating systems, making it a versatile choice for cross-platform environments.

It facilitates file-based sharing, enabling clients to access and share individual files or directories rather than accessing storage at the block level.

In an NFS setup, the responsibility of managing the file system (such as organizing and managing files and directories on behalf of the clients) rests with the NFS server.

NFS allows files to be shared among multiple servers, providing a means for collaborative access and data sharing across server environments.

In NFS, the file system is implemented at the server level. This means the server maintains the file system, and clients access files within that shared file system.

NFS is a protocol known for its efficiency and streamlined design. It is considered a user-friendly choice as it is a shared protocol, making it easier for clients or users to implement and utilize it.

iSCSI is cost-effective in implementation, providing an economical network at the block level. The need for additional network devices is reduced as the protocol need not always use HBAs, distinct cabling, or specific storage devices.

iSCSI is also flexible as it runs on an internet protocol that does not limit the distance between the initiator and the target. This protocol fully leverages the interoperability advantages of Ethernet and TCP/IP. Plus, existing servers can be used several times for configuring iSCSI implementation.

iSCSI is known for swift data transfer even for larger volumes, as the protocol is normally configured for 10 gigabits per second Ethernet (GbE) infrastructure.

iSCSI is easy to deploy and manage, with the users who maintain it not requiring in-depth technical knowledge. The protocol is, therefore, conducive for development and disaster recovery too.

Finally, iSCSI features enhanced network security through identity authentication, physical and logical network isolation, confidentiality, and integrity.

NFS is secure as it uses strong authentication for protection against unauthorized access.

Users can share large files without breaking them down into smaller parts, and enterprises can collaborate across teams via NFS, thus enhancing productivity.

High scalability via data integration is a key benefit of NFS. The protocol can integrate local data with data from remote locations. Enterprises can, therefore, optimize their data centers and minimize costs by consolidating storage.

NFS provides speedy access to data by minimizing latency across wide area networks (WANs).

Like iSCSI, NFS is also suitable for disaster recovery and is used by organizations during disaster recovery planning. In case of a disaster, personnel can leverage NFS to create a virtualized remote copy of sensitive data.

Finally, NFS is secure and suitable for thwarting unauthorized access to data. It is also conducive for auditing and monitoring network activity remotely.