Storage Connectivity Options in the Data Center

The data is stored on different media, which can be electromagnetic, optical, or silicon-based. Examples are the hard disk drives (HDDs), tapes, optical disks such as compact discs (CDs), digital video (versatile) discs (DVDs), Blu-Ray disks, and solid state drives (SSDs). The storage media has evolved during the history of computing, and there are different usages for the different types of media. When it comes to a single computer, such as a laptop or a desktop for personal or office use, usually the applications that run on it are limited and include the operating system and its applications, Internet browsers, and some specialized applications for video editing, word processing, and so on. The storage needs are not big, and usually the internal storage (an HDD or SSD) is more than enough. The real challenge is in the data centers. As already mentioned, the data center servers run the applications that provide the services for companies, organizations, and so on. Some applications process requests; others, such as databases, are responsible for organizing and structuring the data, but all of the applications need a place to store data. As there are multiple applications, the generated amounts of data are usually very large and require storage solutions that can scale up to meet the needs.

Storage solutions always start with two main components: the servers, which need to store or read data and are the active part of such a communication (and are therefore called initiators), and the storage systems, which receive and store data from the servers, or read and send data to the servers in response to their requests. The storage systems do not start communication with the servers; they cannot initiate it. They can only wait for communication to start, and because it is intended for them, they are called targets.

The storage systems hold multiple storage elements: HDDs or SSDs (we’ll call them disks for simplicity). This is the physical storage in a storage system. There are also the controllers that manage these disks and that manage the communication of the storage system. Also, there is the software that manages, abstracts, and virtualizes the underlying hardware. In general, the physical disks are abstracted and virtualized on the storage system by combining the physical drives into a logical unit. From these logical units are created volumes. The volumes are assigned identifiers, which are called Logical Unit Numbers (LUNs). A LUN is the address the initiator will need to access the correct storage resource.

It is important to be in control of what resources are accessed by which server. Also, you do not want to expose the real LUN addressing to the SAN infrastructure; the information is stored in blocks on the storage systems, and if inappropriate or unauthorized access is allowed, the data can be destroyed!

To avoid such situations, two techniques are used at the targets:

• LUN masking: The real LUNs are not exposed, visible, to the SAN and the initiators. The exposed LUNs are different, usually a 0 or a 1, and an internal table is used, which maps the correct LUN to the initiator’s request, based on the initiator’s identity.
• Initiator groups: These are similar as in concept to access lists and define which initiators are allowed to access which volume.

Although this book focuses on Cisco data center solutions, the information for the storage systems is very important, because you need to know it when you design a data center storage solution as well as when you need to perform troubleshooting. When troubleshooting the communication between the initiator and the target, you have to follow the whole path of the communication to understand the protocol used, to know which components are configured, and to know where they operate in the storage network system.

The storage systems and the servers can be directly connected or can go through a communication infrastructure. Communication between the initiators and the targets can use different transport protocols. Examples are the Fibre Channel Protocol (FCP) and the Internet Protocol (IP). What transport protocol will be chosen depends on the type of network storage. There are two types: block-based and file-based. The decision of which network storage system type will be used depends on the specific needs of the data center, the requirements, and the infrastructure.