Download the authoritative guide: Enterprise Data Storage 2018: Optimizing Your Storage Infrastructure
Planning and evaluating storage options is an ongoing challenge. A common IT paradigm is that, no matter how much space is available, data will always grow to fill it. Technologies like e-commerce are placing increasing demands on the availability, capacity and speed of storage technologies. In the past, the solution for increasing storage capacity was simply to add more storage devices. However, just adding more disk space using traditional technology can create bottlenecks that slow performance. In the 21st century, there are other options available.
Even though today's savvy IT managers can speak in acronyms, storage technologies can present a challenge. Competing technologies, such as DAS, NAS, and SAN using SCSI in a JBOD or RAID configuration provide not only technical solutions, but also a range of implementation and management considerations. Acronyms aside, all of these technologies are used to address a common goal; the need for more and always-faster data storage facilities. Beyond the technologies, "standards" evolve as each manufacturer provides extensions that may or may not lock IT managers into a single vendor. For those responsible for making storage technology purchasing decisions, it can be very confusing, and a bad decision can cost money and time and impact the performance of the entire network.
Detailed planning of storage enhancements gives IT management a chance to study the technologies and determine what provides the right scalability and flexibility for their network. However, such planning cannot take place without an understanding of the advantages and disadvantages of the underlying technologies. The planning process is made more difficult because even though standards do exist, they change constantly. In the network storage arena, for example, the National Committee for Information Technology Standards (NCITS), complemented by the Internet Engineering Task Force (IETF), the Storage Networking Industry Association (SNIA), and the Fibre Channel Industry Association (FCIA) all contribute to the standards. In addition, several vendors now offer certification and accreditation programs that let buyers identify compatible products and so circumvent compatibility issues.
These accreditation programs provide the information on which decisions can be made. Unfortunately, the storage technology market is changing so quickly that it can be difficult to keep abreast of the competing claims, certifications and accreditations. This article is the first in a series which will look at how accreditation programs like those available from Brocade, IBM, and EMC work, and at how they can be valuable in helping IT managers make the right decisions when it comes to network storage purchasing. Before we start looking at the specific programs, however, a historical view of the storage arena and the related technologies will make it easier to understand how these certification programs came into being.
Storage System Standards
Back in the days of mainframe computing, the disk subsystems required proprietary interfaces. Some vendors offered replacement disks and disk controllers, but the mainframe vendor set the standards. With the advent of microcomputers, however, the standards, which define the interface between the PC and the hard drive, have became more open. Today there are only two choices for storage system standards, AT Attachment (ATA), also known as Integrated Drive Electronics (IDE), and the Small Computer Systems Interface (SCSI). Of the two, for network storage applications, there is really only one choice, which is SCSI. However, to find out why SCSI is the preferred of the two, a look at ATA is warranted.
The AT attachment standard was first ratified in 1994 with the ATA-1 standard, and there since been five additional standards. Of these standards, only the two most recent are likely to be seen in a modern computer system. Unlike the previous storage technologies that required a separate controller, IDE devices have the controller built right onto the drive. The use of ATA standards with ATA/IDE drives all but eliminates cross vendor compatibility issues between drives.
Originally, the IDE/ATA standards specified a way to connect only disk drives to the microcomputer bus, but with the popularity of CD-ROM and the introduction of tape devices for PCs, the standard added packet interface and data block handling to become the IDE/ATAPI standard. This functionality was introduced with the fourth ATA standard, ATA-33.
In spite of the compatibility advantages of IDE devices, and the relatively low cost, IDE is not the preferred technology in network storage solutions, and for good reasons. The first of these reasons is speed - each of the ATA standards has operated at lower speeds than alternatives that use the SCSI standard. Next on the negative list is the number of devices that can be supported. Unlike some SCSI standards that can support 31 storage devices, current IDE standards can only accept four. Lastly, we have the lack of support for external devices, which means that IDE is only suitable for DAS solutions.
SCSI offers distinct advantages over IDE in terms of both speed and flexibility. The SCSI standards, of which there are many, define driver software, commands, management functions, and physical connections to allow a series of peripherals to work together. Although these devices must still be located fairly close to the host device (no more than 25 meters), there is support for 15 or 31 storage devices depending on the standard, and for both internal and external devices. Adding more devices can easily expand a SCSI subsystem. For these reasons and others, SCSI is the most popular storage standard.
Like other technologies, SCSI has evolved to support higher volume disks and higher transfer rates. Unfortunately, this creates multiple levels of specifications. In the case of SCSI, the result is an almost bewildering array of standards including:
In addition, a SCSI specification that uses IP addressing to transfer data is generating a great deal of interest. Although this shows promise and has been highly touted in the technical press, it remains a new implementation. As such, it will undergo several revisions before the "standard" becomes consistent.