IT organizations are beginning to replace their storage area networks (SANs), the predominant storage model for many years. Newer alternatives, such as software-defined storage (SDS), offer several attractive benefits. While SAN can also be software-defined, it still requires investment in proprietary hardware. Each IT organization should assess SAN’s suitability for cost, performance, capacity needs, and scalability requirements, and decide whether to keep it or replace it. This blog explores the issues and drivers for considering a SAN replacement.
What is a Storage Area Network (SAN)?
A SAN is a network of storage devices that offers a pool of shared block storage for multiple computing devices. SANs first appeared in the mid-1990s as a viable alternative to Direct-Attached Storage (DAS) and Network Attached Storage (NAS) for high-speed, mission-critical transactional workloads, such as databases, that require scalable storage delivering high IOPS and low latency. They are also well-suited for virtualized environments, which accelerated their adoption.
A SAN consolidates storage in a single block-level storage area, allowing users to access and manage data from a central location. To maintain high storage traffic and network performance, the SAN is typically implemented via a separate network infrastructure from the local area network (LAN).
Compute servers connected to the SAN can access storage devices on SAN controllers, such as tape libraries, local storage, and disk arrays. This design also offers the advantage of centralized storage management. SANs can also help improve storage security. With data in a centralized, shared SAN storage architecture, an organization can apply consistent policies for security, data protection, and disaster recovery (DR). A SAN can support multiple data backups, and its block-level access improves application availability.
The SAN design also supports dynamic failover, which improves availability and business continuity. The SAN’s network fabric, comprising interconnected storage devices and computers, further improves availability. If one network path is disrupted, the SAN enables an alternate path. This makes it less likely that a single device failure will render storage inaccessible.
The Two Main Technologies and Interfaces for SAN
SANs usually use one of two main technologies to move data in and out of storage: Fibre Channel and Internet Small Computer Systems Interface (iSCSI). Fibre Channel is a high-speed data transfer protocol. It provides lossless delivery of raw block data on an “in-order” basis. Fibre Channel is typically based on optical fiber cables, but it can also use copper cables. (The word “fibre” was adopted by the industry rather than “fiber” to avoid confusion over whether the protocol can only run on fiber optic equipment.)
Fibre Channels support data rates that include 1, 2, 4, 8, 16, 32, 64, and 128 gigabits per second. Architecturally, the switches in a Fibre Channel network operate in unison, effectively as one big switch that comes together to form a switched fabric.
iSCSI is a transport layer protocol operating atop of Transport Control Protocol/Internet Protocol (TCP/IP). With this design, iSCSI makes possible block-level SCSI data transport between two components, the iSCSI initiator and the storage target, over TCP/IP networks. SCSI is a block-based command set connecting devices to networked storage. The iSCSI target can be a SAN controller, which exposes remote volumes that appear as local drives to host systems.
iSCSI is generally less costly than Fibre Channel because it connects servers to storage without requiring expensive Fibre Channel Host Bus Adapters (HBAs), switches, or cabling. Fibre Channel SANs also require admins who have specialized skills. In contrast, administering an iSCSI SAN, which runs on standard, existing Ethernet, is simpler. An IT generalist can easily learn to install and manage an iSCSI SAN.
Disadvantages of SAN
Despite their widespread use, SANs have several disadvantages. For one thing, SANs are expensive; they can cost hundreds of thousands of dollars and require proprietary hardware. Because they require proprietary hardware, provisioning cycles can be long, especially when supply chains are disrupted (e.g., during the COVID pandemic). Setting up and maintaining the infrastructure can be expensive. It can take time to see a return on investment. This makes SANs better suited to larger organizations that can afford the capital and management costs.
They can be complex and thus difficult to manage. A SAN is built in layers, with connections among the underlying storage arrays, the SAN network switches, and the servers that use it. Each layer, device, and connection requires ongoing administration and maintenance, incurring additional costs.
Faults and patching can occur in multiple locations. Component upgrades and interfaces need frequent, if not constant, attention. Indeed, various SAN components are not known for “playing nicely” together. SAN vendors often use proprietary protocols and management tools, further complicating SAN management. Additionally, SANs become a bottleneck in all-flash storage environments, which are becoming the norm in most enterprises.
SAN complexity often requires specialized expertise to manage and maintain. Overseeing the SAN inevitably becomes the job of someone, or even an entire team. When an organization has more than one SAN, the complexity and administrative load grow all the more strenuous. A virtualized SAN architecture can relieve some of the pressure on admins to manage hardware, but it also introduces its own complexity.
Security is also an issue, despite the advantages of the uniform policy highlighted above. A SAN is almost always a shared environment. As a result, it is vulnerable to lateral attacks, in which a malicious actor gains access to one area of the SAN and then moves across it to breach data held elsewhere in the network.
Why should you replace SAN?
Given SAN’s disadvantages, it may be time to consider another storage approach. With SDS, IT organizations now have a viable alternative to replace their SANs. For example, you can build a storage solution using software-defined, NVMe/TCP block storage that provides the storage-pooling advantages of SANs while delivering high performance and low latency, without the complexity and high costs of SANs.
Instead, using a solution like Lightbits that leverages low-latency NVMe storage and standard TCP/IP, IT organizations can deploy high-performing, clustered storage that is cost-effective and highly scalable—but without SAN’s traditional overhead headaches.
Replacing a SAN is not a minor project, so it pays to think through the pros and cons. What’s clear, however, is that new storage approaches can do everything a SAN can, but with less complexity and at a lower cost. It may be time to look forward to the post-SAN era.
Related Blogs:
