What Makes a Storage Area Network (SAN) Essential for Modern Data Centers?

SAN or storage area network is a dedicated network/subnet that interconnects shared pools of storage to several servers.

Storage accessibility and availability is a fundamental thing for computing tasks. The standard approach of direct-attached disk deployment works perfectly fine for lots of companies because of its simplicity, but they are tied to the physical server across SAS. For those corporations that are searching for more control and flexibility, they should better orient on SAN.

SAN technology covers the needs of lots of huge corporations that want to get access to more scalable, dedicated environments where several servers can be easily connected to the storage devices. The process of organization can be done in tiers or pools. SAN is a more centralized approach where users can optimize storage capacity, resilience, and have a good level of protection.

Common use cases for storage area networks

If simplifying this term, then SAN is a network of disks that can be accessed by the servers. This approach helps to consolidate storage. For instance, when talking about the data center with lots of servers, usually there are used physical disks that are located on individual servers and it is much more effective to use storage on the dedicated subsystem, as it is much easier to protect and manage all the data.

SAN is also extremely effective for the availability of the storage system. This technology functions by connecting several devices and computers, so during any kind of disruption, it is possible to use an alternative path rather than fully relying on one storage device. During any technical problems, the system will continue to function because of the accessibility of other storage devices and paths to them. The centralized approach greatly simplifies the administration process and management tasks.

All these factors positively affect business continuity and disaster recovery, so it can be a huge benefit for lots of organizations. To understand a peculiarity of SAN, let’s try to compare it with the traditional DAS.  

DAS technology functions by connecting disks directly to the computer through SAS/SATA interfaces. Disks usually store information that is needed for a certain server. Of course, information from one DAS device can be accessed from other servers, but the connection is done via the standard IP network. Moving large volumes of information over an IP network can be extremely time-consuming, and it can directly affect the performance characteristics.

SAN functions totally in another way – all the disks are connected to a dedicated storage network. The storage can be easily accessed by any server in the network, and there is no necessity to go through the LAN. That means the resilience and performance of the system can be greatly improved when compared with the standard approach.

SAN can support as many storage devices as needed and can easily scale to have more than a thousand disks. Moreover, any server with a specific SAN interface can access the system. When talking about the major network interfaces, there are 2 of them:

• Fibre Channel is considered to be a high-speed network that is very peculiar because of its lowest possible latency and high throughput. Such an approach helps to locate storage blocks at one specific location while the server may be located across the city or in campus. Traditionally, this approach might also be related to the circumstances when servers and storage are located in the same place within 10 meters apart. This approach is used by the implementation of HBAs in every server, storage, network switches, or other devices in the network. Every HBA includes one or several ports for the data exchange, and they can be physical or virtual.
• ISCSI is one more type of network that functions by connecting shared storage with computing. FC offers access to the unique network design, whereas ISCSI unites traditional data blocks, command packets, and Ethernet network technology. Due to this, iSCSI storage can utilize the same adapters, cabling, switches, and other network components in the Ethernet network. Every server that functions in the system notices iSCSI storage as another connected disk. This approach functions on such standards as targets and initiators, where initiators might be storage arrays or operating systems whereas the target is a network-connected hard disk.

The working principles behind a SAN

SAN is an important network that is needed for the connection of the server to the storage. The major aim of this approach is to locate storage separately so it will be easier to protect and manage lots of processes. This may be achieved physically by the usage of the storage arrays or through software such as vSAN or VMware.

By connecting servers and storage via a specific network, it is possible to improve performance characteristics, because there is no necessity to compete with LAN bandwidth. SAN can be divided into 3 distinct layers, such as storage layer, host layer, and fabric layer, so let’s discuss each component individually.

1. Host layer. This layer functions on the basis of servers where all the workload is running. Usually, servers use LAN/Ethernet components to communicate with other servers and users. However, it also includes a network adapter for dedicated SAN access. Such adapters are called HBAs and use firmware and a device driver. Such a configuration is needed for the storage commands’ communication to the SAN and its storage. There are lots of SAN technologies that organizations can select, and some of them are FC, iSCSI, and InfiniBand. The choice of the necessary technology totally depends on the needs of the specific company and costs.

2. The fabric layer relates to all the network devices and cabling that connect storage and hosts. Such network devices usually are as following - gateways, switches, protocol bridges, and routers. SAN fabric devices might use optical fiber connections or copper-based cables for the local network. During the construction of the SAN fabric, several connections are used for the availability of various paths. So, in case of some troubles, there will still be a possibility to continue communication through the available paths.

3. Storage layer. This layer includes lots of storage devices that are united into tiers and pools. Storages can be of various types, including HDDs, SSDs, and CD/DVD drives and even tape drives. Most of these devices can be organized in a RAID group or in disk partitions, which have an assigned LUN that is the same as the disk drive letters. That means the host can access needed LUNs so that it is possible to control all the storage assets more effectively. There are 2 options for possible permissions – zoning and LUN masking.

SAN technology also implements the use of some protocols that are needed for the preparation of data storage and for communication. The most popular is the FCP protocol, but there is also the iSCSI protocol, ATA, FCoE, iFCP, and some others. The technology usually supports several protocols for the proper communication of all layers.

How to configure a storage area network

To begin with, the enterprise should consider the computability requirements for software and hardware.

• Storage (patches list, firmware, and more)
• Host bus adapters (river version, firmware version, and more)
• Switches

For the setup of SAN, follow the next steps:

1. Cable all the necessary hardware components and install the needed software. That means you should check the versions, set up HBA, and the storage array.

2. Modify some configuration settings if necessary.

3. Test everything relating to the operating processes.

4. Set up the performance base for each component.

5. Document the operational procedures.

Inside SAN fabric architecture and functionality

The basics of SAN fabric are related to the server-storage connectivity, high-performance, and scalability in the network. When talking about the simplest possible case scenario, FC SAN can be easily connected to the HBA ports directly with the needed ports on SAN storage arrays, often with the usage of optical cables.

SAN fabric is so special because of the ability to minimize points of failure, and that can be extremely beneficial. The main goal is to create at least 2 connections between SAN elements. So, that during any issues, at least one working path will be available.

Let’s dive deeper into some behavioral factors of the system functioning. The host server creates a request to access the storage device. The traditional SCSI commands should be encapsulated into packets, and they are structured according to the rules of the protocol. These packets are delivered to the HBA, where they are placed on the cables. Then, they are transmitted to the SAN, and the request will be at the switches. A specific switch will then send the request to the storage device. To accomplish this process, the storage processor will interact with the needed device.

A closer look at SAN switches

SAN switches are considered to be the major point of the SAN. Like any other type of switches, these ones function by receiving data packets, determining their destination, and directing packets to the destination. SAN fabric has several types of possible switches, such as edge, modular, and backbone switches. Depending on the size of the SAN, the number of switches might differ. For the smaller SAN, there might be 16-32 ports, whereas a larger one might have even 128 ports.

For better resilience of the storage, fabric is not enough there should be included some internal technologies, such as RAID. Such a system will also add some other technologies that relate to data compression, storage cloning, thin provisioning, and others. A well-designed system helps any host connect to any storage device. However, for better security, LUN masking and zoning technologies might be used.

Different approaches to SAN deployment

• Virtual SAN. Virtualization is suited perfectly to SAN and helps to add scalability and flexibility. VSAN or Virtual SAN is a form of isolation that is similar to SAN, which uses virtualization to get several segments together with a physical SAN. VSAN can be used for the security improvement and for the management of the network traffic. The process functions in such a way that VSAN can help with the prevention of some issues on one segment that can impact other parts, and the segment can be changed without touching any physical components. VMware offers vSAN technology and gives access to advanced features such as moving storage, tiering, storage pooling, and others. Additionally, VMware vSAN provides features such as data lifecycle management and automatic data movement between tiers.
• Unified SAN. Usually, objects, files, and other storage types need a separate storage system such as NAS. Unified storage deals with various types of storage because of several used protocols such as NFS and SMB, and others. With the usage of a unified SAN system, users have access to the powerful features, for instance, data replication, data encryption, storage tiering, and data deduplication.
• Converged SAN. A huge disadvantage of the traditional FC SAN is its price; to overcome this drawback, it is possible to use Ethernet networking. FCoE supports converged SAN that can function on Ethernet components and helps to create a low-cost network.
• Hyper-converged infrastructure. With the help of HCI, it is possible to combine storage and compute resources into nodes as needed and manage all the processes via a simple utility. This technology uses virtualization and its major aim is to create simplified hardware deployment with fast scalability. HCI can be used instead of SAN or used together with an enterprise SAN.  

Advantages of using a SAN

• Exceptional scalability. It is possible to support thousands of SAN hosts and devices for storage, so any organization can satisfy its requirements.
• High performance. For the storage tasks, a separate network is used, and to get the best possible performance, FC is recommended. Also, there are converged and iSCSI networks available as well.
• Availability of more management features. That relates to such options as data deduplication, data encryption, self-healing variants, and more. The features are centralized and can be easily applied to all the resources.
• High availability. The points of failure are extremely minimized or almost excluded because of the diversity of alternative paths.  

Potential drawbacks of SAN systems

• Management. Such features as zoning and LUN mapping are not the easiest ones for organizations with lots of other non-technical tasks. To get the needed level of security, some time-consuming processes could not be excluded.
• Complexity. Traditional SAN is considered to be really complex, and organizations should think about professional IT experts and higher budgets for dealing with this drawback.
• Scale. SAN is extremely effective for huge environments with lots of storage and servers. Sure, it is possible to use it on a smaller scale, but that requires more investment. That’s why smaller companies should better consider the usage of converged SAN, HCI deployment, and other alternatives.

Comparing SAN and NAS solutions

NAS or network-attached storage is an alternative variant that is based on a file-based protocol. This technology is based on a dedicated file server that is located between the storage and the servers.

Both these approaches are needed for the data storage. SAN is suitable for block-based storage which relates to more structured information. Whereas NAS is better for the unstructured data – images, videos, docs, emails, and others.

NAS requires lower costs and utilizes a common network. When talking about SAN, it is way more scalable and has much better performance characteristics.

Leading SAN vendors and solutions

When planning SAN, it is important to consider the network, hosts, and storage systems.

Let’s start with the hosts. Almost any host can function on SAN, the necessity is to choose the interface for fabric access. The most suitable option is to order servers with a multi-port FC HBA. If not, HBA can be added on the server, but there should be a PCIe slot. HBA cards are manufactured by QLogic, ATTO, Agilent, Brocade, and Broadcom. These cards are sold by lots of vendors.

The next point is network cabling, switches, and other important components. Here are some options to choose from:

• QLogic SANbox 5xxx-series switches
• Cisco MDS-series switches
• ATTO Technology 8308, 8316, and 8324
• The storage vendors are as follows:
• NetApp
• Huawei
• Hewlett-Packard Enterprise
• Dell
• Hitachi Data Systems

Key standards in SAN technology

The development of the standards relates to some industry groups. The first one to mention is SMI-S. This standard is responsible for the management of the storage devices. Another stand in SAN is Channel Physical Interface.

Best practices in SAN management

For the proper performance, it is highly important to follow the following practices. Start with the proper reporting and monitoring. Here are some important indicators and metrics to consider:

• KPIs connected with write/read throughput for each array;
• LUN/SAN capacity – check shortage and common trends;
• Check the working performance of the host server I/O
• Metrics relating to SAN fabric.

By tracking the major metrics, it is possible to take proactive measures for most of the possible issues.