Storage Area Network (SAN) - Comparative
A SAN can be considered an extension of Direct Attached Storage (DAS).
DAS Where there is a point to point link between the server and storage, a SAN allows multiple servers to access multiple storage devices on a network share.
Both SAN and DAS applications and programs users make their data requests to the filesystem directly. The difference lies in the way that the file system obtains the required data storage.
In DAS, the storage is local to the filesystem, whereas in SAN, storage is remote.
SAN uses different access protocols such as Fibre Channel and Gigabit Ethernet. On the opposite side is the Network-Attached Storage (NAS), where applications make data requests to the file systems remotely using Server Message Block protocols (CIFS) and Network File System (NFS).
SANs provide connectivity I/O via the host computers and storage devices combining the benefits of Fibre Channel technologies and network architectures thus providing a more robust, flexible and sophisticated approach that overcomes the limitations of DAS using the SCSI same logic interface to access the storage.
SANs are composed of three layers:
Host Layer: This layer consists mainly of servers, devices or components (HBA, GBIC, GLM) and software (operating systems).
Fiber Layer: This layer conform cables (fiber optic) and the SAN SAN hubs and switches as a central connection point for the SAN.
Storage layer: This layer is composed formations disk (Disk Arrays, cache, RAID) and tapes used to store data.
The storage network can be of two types:
Fibre Channel network: the network fiber channel is the physical network of devices that uses Fibre Channel Fibre Channel switches and directors and protocol Fibre Channel Protocol (FCP) for transport (on serial SCSI-3 Fibre Channel).
IP network: LAN infrastructure used with standard hubs and/or Ethernet switches interconnected. IP SAN uses an iSCSI transport (SCSI-3 serial over IP).
Fibre Channel (Fibre Channel) is a standard that carries gigabit, is optimized for storage and other high speed applications. Currently the rate is handled is around 1 gigabit (200 MBps full-duplex). Fibre Channel will support full duplex data transfer rates up to 400 Mbps in the near future.
There are three Fibre Channel topologies:
Point to point (Point to Point).
Arbitrated loop (arbitrated loop).
Switched fabric (Switched Fabric).
The “Fibre Channel Fabric” was designed as a generic interface between each node and interconnection with the physical layer of the node. With the accession of this interface, any node “Fibre Channel”, can communicate over the fabric without specific knowledge required interconnection scheme between nodes.
Fibre Channel Arbitrated Loop
This topology refers to sharing architectures, which support full-duplex speeds of 100 MBps or even up to 200 MBps.
Analogously to the Token Ring topology, multiple servers and storage devices, can be added to the same segment of the loop.
Up to 126 devices can be added to an FC-AL (Fibre Channel Arbitrated Loop).
Since the loop is shared transportation, devices must be arbitrated, that is, should be monitored for access to transport loop before sending data.
Services provided by a fabric
When a device is attached to a fabric its information is recorded in a database, which is used for access to other devices on the fabric, also it keeps track of the physical topology changes. These are the basic services are presented within a Fabric.
Login Service: This service is used for each of the nodes when they perform a session to the fabric (FLOGI). For each of the communications established between nodes and the factory identifier origin (S_ID) and connection service returns sent one D_ID with the domain and port information where the connection is established.
Name services: all information equipment “logged” in the fabric are registered in a name server that performs PLOGIN. This in order to have all recorded in a database of local residents inputs.
Fabric Controller: is responsible for providing all state change notifications to all nodes that are discharged into the fabric using RSCNs (log change status notification).
Management Server: the role of this service is to provide a single point of access to the previous three services, based in “containers” called zones. A zone is a collection of nodes defined to reside in a confined space.
Although the need for storage is clear, it is not always clear what the appropriate solution in a given organization.
Choosing the right solution can be a decision with major implications, but there is no single correct answer, you need to focus on the specific needs and objectives of each end user or organization.
For example, in the case of companies, the company size is a parameter to consider. For large volumes of information, a SAN solution would be more accurate. However, small companies use a NAS solution.
However, both technologies are not exclusive and can coexist in the same solution.
As shown in the graph, there are a number of possible outcomes involving the use of technologies DAS, NAS and SAN in a single solution.
Latency: One of the main differences and characteristics of the SAN is that they are built to minimize the response time of the transmission medium.
Connectivity: allows multiple servers to be connected to the same set of disks or tape libraries, allowing the use of storage and backups are optimal.
Distance: the SAN to be constructed with fiber optic inherit the benefits of this, for example, SAN devices can have a separation of up to 10 km without repeaters.
Speed: the performance of any computer system depends on the speed of its subsystems, it is why the SAN have increased their information transfer rate, ranging from 1 Gigabit until now 4 and 8 Gigabits per second.
Availability: One of the advantages of SAN is that having greater connectivity, allow servers and storage devices more than once connected to the SAN, thus can have redundant “routes” which in turn increase fault tolerance.
Security: security in the SAN has been from the beginning a key factor, since its creation the possibility that a system agreed to a device that does not correspond to him or interfere with the flow of information is noticed, which is why we have implemented zoning technology, which consists of a group of elements are isolated from the rest to avoid these problems, zoning may be performed by hardware, software or both, being able to group by port or WWN (World Wide Name) An additional technique is implemented at the level of storage device that is the Presentation, is to make a LUN (Logical Unit Number) is accessible only by a predefined list of servers or nodes (deployed with the WWN).
Components: the primary components of a SAN are: switches, directors, HBAs, servers, routers, gateways, disk arrays and tape libraries.
Topology: each topology provides different capabilities and benefits SAN topologies are:
ISL (Inter Switch Link, link between switches) currently connections between switches in SAN ports are made through “E” and can be grouped to form a trunk (trunk) that allows greater flow of information and fault tolerance.
Architecture: the current fiber channel operate under two basic architectures, FC-AL (Fibre Channel Arbitrated Loop) and Switched Fabric, both schemes can coexist and expand the possibilities of the SAN. The FC-AL architecture can connect up to 127 devices, while switched fabric up to 16 million theoretically.
Share storage simplifies administration and adds flexibility since cables and storage devices do not need to move from one server to another. Realize that unless the model of SAN file system and clustering, SAN storage has a ratio of one to one with the server. Each device (or Logical Unit Number, LUN) of the SAN is “owned” by a single computer or server. As a counter example, NAS enables multiple servers to share the same set of files on the network.
SAN tends to maximize storage utilization, since multiple servers can use the same reserved space for growth.
Storage paths are many, a server can access one or “n” disks and a disk can be accessed by more than one server, which increases the profit or return on investment, ie, ROI (Return On Investment), for its acronym in English.
The “storage area network” has the ability to support in physically distant locations. His goal is to lose as little time as possible or better yet, do not waste time, so both backup and recovery are online.
One of the great advantages is that it also provides high data availability.
A primary advantage of the SAN is its compatibility with existing SCSI devices, leveraging existing investments and enabling growth from existing hardware. Through the use of modular devices such as hubs, switches, bridges and routers, you can create fully flexible and scalable topologies, securing investment from day one and, more importantly, taking advantage of SCSI devices as considerable cost-SCSI RAID subsystems SCSI, tape libraries or CD-ROM towers, as through a Fibre Channel SCSI bridge can connect directly to the SAN. Since they are on their own network, they are accessible by all users immediately.
SAN performance is directly related to the type of network used. In the case of a Fibre Channel network, the bandwidth is approximately 100 megabytes / second (1,000 megabits / second) and can be extended by increasing the amount of access connections.
SAN capacity can be extended almost limitlessly and can reach hundreds and even thousands of terabytes. A SAN can share data among multiple computers on the network without affecting performance because SAN traffic is completely separate from user traffic. They are application servers that act as an interface between the data network (usually fiber channel) and the user network (usually Ethernet).
Moreover, a SAN is much more expensive than a NAS because the first is a complete architecture that uses technology that is still very expensive. Normally, when a company estimates the TCO (Total Cost of Ownership) to the cost per byte, the cost can be justified more easily.
There are basic protocols used in a storage area network:
Fibre Channel arbitrated loop protocol, used in hubs, in the SAN hub this protocol is used by excellence, the protocol control who can communicate, only one at a time.
Switched Fibre Channel protocol, used in switches, in this case several communications can occur simultaneously. The protocol is responsible for connecting communications between devices and collision avoidance.
Used by applications, it is a protocol used for an application of a computer to communicate with the storage device.
In the SAN, the SCSI encapsulated over FC-AL and FC-SW.
SCSI works differently in a SAN within a server, SCSI was originally designed to communicate within the same server disks, using copper wires.
Within a server, data travels parallel SCSI and SAN traveling serialized.
It is a computer network technology that encapsulates Fibre Channel frames over Ethernet networks. This allows Fibre Channel networks use 10 Gigabit Ethernet (or higher speeds) while preserving the Fibre Channel protocol. The specification was part of the International Committee for Information Technology Standards T11 FC-BB-5 standard published in 2009.
An essential part of the safety of storage area networks is the physical location of each and every one of the network components. The construction of a data center is only half the challenge is the fact decide where to put the network components (software and hardware) the other half and the hardest. Critical network components, such as switches, storage arrays or hosts which should be in the same data center. By implementing physical security, only authorized users can have the ability to perform both physical and logical topology changes, changes such as: changing port cable, reconfigure access to a computer, add or remove devices, among others.
Planning should also take into account environmental issues such as cooling, power distribution and requirements for disaster recovery. At the same time should ensure that IP networks are used to manage the various components of the SAN they are safe and not accessible for the entire company. It also makes sense to change the default passwords with network devices in order to prevent unauthorized use.