Comparison and analysis of Fc_san and Ip_san

1. Technical background

From the perspective of storage development, the emergence of external isolated storage is to solve the performance problem of large-capacity data storage, FC (Fibre Channel protocol) becomes the perfect choice at this time. Since then, storage has evolved from the DAS Direct-attached array to the storage network stage, where storage becomes a separate data storage network (for the data itself and its security) after the communication network (for the application business), which is referred to as the "second network" or "Network Network". Since the storage network is hosting the customer's most important business data, its dual controller reliability architecture, multiple raid mechanisms, and the data backup mechanism become the cornerstone and pillar of the customer information system in the backyard.

The advent of IP sans is primarily a solution to the massive data storage in applications where data protection levels are not stringent and data access speeds are not demanding. Due to the wide application of IP technology in the interconnection network, the IP San is easy to realize remote data access. However, the efficiency, performance, and security of the IP San are its short planks relative to the FC San.

2. The technical principle of FC and IP

Fibre Channel layered According to the protocol layer, each layer of technology is independent of each other, leaving room for growth, and by the approved standardization of the development of institutions, layered structure of 5 layers.

650) this.width=650; "src=" http://s3.51cto.com/wyfs02/M02/54/35/wKiom1R8Ag3SplFNAAB-Y0F9mo8282.jpg "/>

⑴c-0 (Physical layer bottom):

The FC-0 layer defines the physical port characteristics of the connection, including the physical characteristics of the media and connectors (drives, receivers, transmitters, etc.), electrical characteristics and light characteristics, transmission rates, and some other connection port characteristics.

⑵fc-1 (Transport Protocol):

Specifies the 8B/10B encoding and transmission protocol. Includes serial encoding, decoding rules, special characters, and error control.

⑶c-2 (Frame protocol):

Specify the specific transmission mechanism, including the frame format, the information exchange between nodes.

⑷c-3 (Public Service):

Public services that provide advanced features, the fabric protocol and flow control between ports, define three services: Striping, search Group (Hunt Group), and multi-channel playback (broadcast multicast).

⑸fc-4 (ULP mapping):

Defines the interface between the Fibre Channel and the ip,scsi-3 and other upper layer protocols (ULP).

Fibre Channel also defines the flow control mechanism using Bb-credit (buffer-to-buffer credit) and relies on the Receive-ready (R-rdy) ordered set to complement the credit. A terminal contact attached to the switch will establish its bb-credit in the process of logging in to the switch. The party that participates in the communication at the end of the switch will establish the bb-credit of its own switch at logon. Bb-credit does not have an end-to-end component. The sender will bb-credit minus 1 when it emits a frame, until the number of Bb-credit is zero. No more frames can be sent at this time. The Bb-credit is added 1 when the R-rdy is received. The initial value of the Bb-credit must be non-zero. If zero, the description can no longer receive or send frames.

IP SANthe mainstream underlying protocol isISCSI, the agreement defines theTCP/IPnetwork send, receiveBlock(block)-level rules and methods for storing data. The sending side willSCSIcommands and data are encapsulated intoTCP/IPpackets are then forwarded through the network and received by the receiving side.TCP/IPpackage, revert it toSCSIThe command and data are executed and will be returned after completion.SCSIcommands and data are then encapsulated intoTCP/IPthe packet is re-routed back to the sender.

650) this.width=650; "src=" http://s3.51cto.com/wyfs02/M00/54/33/wKioL1R8ApaCDwFwAAB5XaqpBtA765.jpg "/>

The entire process, in the view of the user, uses a remote storage device as if it were accessing a local SCSI device as simple. iSCSI Technology -enabled servers and storage devices can be connected directly to existing IP switches and routers, so iSCSI Technology has easy-to-install, Low cost, no geographical restrictions, good interoperability, easy management and other advantages.

3, FC and IP comparison and analysis

First, theIP San and FC San Comparisons can be analyzed from several levels, such as speed, reliability, architecture, security, scenario, distance, cost, and more.

IP SAN up to today, the fastest transfer speed is about 100mb/s, and FC has already reached 200mb/s, then, for applications that require ultra-large bandwidth, such as video-on-demand,FC is more advantageous, but for general networking, backup, database, and so on, both can be satisfied.

FC SAN in front-end host io channel, back-end disk channel, disk interface can all reach ip SAN mainstream devices are FC SAN can use FC disk, speed can be 1 thousand Thousand, and ip SAN Basically can only choose sata disk.

In the reliability design of key components such as controllers,FC sans are usually supported for dual controller and host Multipath management, and the IP San is basically a single controller design.

In terms of architecture , theIP San is more open and can be used wherever there is Ethernet, and theFC San needs to build its own San network, and the switch also requires a dedicated FC switch. The overall architecture of the Huawei T8000 server +s31000 Storage Array overcomes this limitation by eliminating the need to purchase FC switches and host HBA adapter cards separately.

However,FC due to the characteristics of the protocol design makes it has the inherent security characteristics, theIP SAN uses the ISCSI protocol, because it will be in the public network flow, so the security of a lot of trouble, need to join the Additional protocols such as IPSec can further safeguard their security.

in the Program Application on, IP SAN focus on providing a simple and easy-to-use "super-large hard disk", the main solution is the large-capacity, low-cost basic storage needs problems;

The FC SAN is mature and has a lot of dedicated backup, disaster recovery, resource management solutions.

distance is also an interesting topic. Usually ip SAN is considered to be the strongest network adaptability as long as IP can achieve remote disaster recovery, and FC because it is not an Internet universal protocol, there is a limitation of transmission distance. but , that's not true! The disaster tolerant link is different from the general communication link, it is necessary to carry on the continuous data transmission with the long time Big data quantity, the bandwidth and the delay are certain requirements. Common Internet ip link or operator ip the metropolitan area network belongs to the bandwidth sharing, qos is not guaranteed, the general delay is high, can not meet the synchronous or asynchronous remote disaster recovery requirements; FC over SDH ,

cost , FC SAN To compete, must all upgrade to reach the next generation of 10g speed is possible with FC SAN . But at present, 10g Ethernet technology is mainly used in large metropolitan area network backbone, 10g 4g FC SAN .

Therefore, the comparison between the1G IP San and the 4G FC San is not really significant, and both performance differences and price differences are obvious, The key is to see what level of data security needs to be achieved, and what level of system construction input is.

This article is from the "Divinity New Space" blog, please make sure to keep this source http://abool.blog.51cto.com/8355508/1585000

Comparison and analysis of Fc_san and Ip_san