The two remote transmission technologies have different advantages and disadvantages: fiber network topology is undoubtedly the first choice in terms of performance, but it is more economical to use IP addresses or the existing network topology, to make full use of the high-speed IP network and existing communication links, you must consider both fish and bear's paw.
The communication link between the local and remote disaster recovery ends is an important factor to consider in the Remote Disaster Recovery Solution. The biggest impact is the TCO, because communication links are usually the most expensive overhead in Remote Disaster Recovery solutions.
RONG> remote transmission requirements
The distance between local data centers and Disaster Tolerance data centers is dozens to hundreds of kilometers from natural disasters such as flood prevention, earthquakes, and terrorist attacks. Generally, the synchronization center requires about 100 kilometers, and the asynchronous center can reach several hundred kilometers or even thousands of kilometers.
The following are two basic measures to measure the performance of any communication technology:
Bandwidth-the total amount of data that can be moved through the link within a given period of time. According to SLAService-Level Agreement, SLA) requirements, if the disaster recovery center only needs to meet the remote tape backup function, the bandwidth requirement is <50 Mbps; if you want to implement the data center integration function, the bandwidth must be 50 Mbps ~ 1 Gbps; the bandwidth required for remote disk images is more than 1 Gbps.
Latency: the total amount of time required to move data from one end of the link to the other. The latency of data transmission between local and remote disaster tolerance is mainly divided into two parts: one is the latency caused by the distance between two locations, and the other is the latency caused by the link bandwidth between two locations.
Topology of Remote Disaster Recovery
From the perspective of topology structure, there are three main network storage technologies: DAS, NAS and SAN. The Remote Disaster Recovery Solution generally uses the last two.
NAS figure 1) is a special dedicated data storage server, embedded file service management tool that provides cross-platform file sharing. It is based on LAN), uses TCP/IP protocol for communication, and transmits data in file-Level I/O mode.
Figure 1
NAS (Network Attached Storage) topology
SAN figure 2) is actually a dedicated network, independent from the traditional LAN, connected through the gateway device and LAN. SAN adopts a network-oriented storage structure, which separates data processing from data storage, it is a new storage architecture featuring flexible addressing, data transmission over long distances, high I/O performance, high storage device utilization, and high user sharing.
Figure 2
SAN (Storage Area Network) topology
Remote Transmission Technology Design
The Protocol Remote Disaster Tolerance solution is equally important to use communication links and links. The Protocol has the ability to adjust bandwidth utilization, which is a significant advantage of any Remote Disaster Tolerance solution. This feature allows users to specify how much bandwidth each application can use, that is, to allow a single link to share with a variety of applications, each application selects the bandwidth according to their own needs. In addition, the Protocol can be automatically adjusted according to the current status.
The data transmission protocols used in the existing Remote Disaster Tolerance solutions include FCPFiber Channel Protocol, optical fiber Channel Protocol), FCIPFiber Channel over IP, and IP-based optical fiber Channel Protocol) iFCPInternet Fiber Channel Protocol, Internet Optical Fiber Channel Protocol) and iSCSIInternet SCSI.
FCP Optical Fiber Channel Protocol)
Optical Fiber Channel standard (FCS) defines a high-speed data transmission mechanism for connecting hosts and storage devices. The optical fiber channel protocol is the SCSI interface protocol on the Optical Fiber Channel. Fiber Channel Network Topology supports three architectures: point-to-point, arbitration ring, and exchange architecture. The transfer rate can reach 10 Gbps, supports optical and electrical media, can transmit a variety of interface command set, including IP, SCSI, IPI, HIPPI-FP and audio/video. If the optical fiber is used as the transmission medium, the distance between the local device and the remote device can reach 10 kilometers.
Fcip ip-based Optical Fiber Channel Protocol)
The TCP/IP-based FCIP protocol uses TCP/IP data packets to encapsulate Fiber Channel commands and data, and transmits Fiber Channel commands and data over an IP network. FCIP can connect isolated fiber-channel SAN through an IP network to form a unified network for storage areas. Therefore, FCIP can be used to overcome the limitations of Fiber-channel distance, connects to the SAN Island from a distance greater than the fiber channel support.
IFCPInternet Optical Fiber Channel Protocol)
Internet Optical Fiber Channel Protocol (iFCP) is a gateway-to-gateway protocol that provides Optical Fiber Channel Communication Services for optical fiber devices on TCP/IP networks. IFCP can use the congestion control, error monitoring and recovery functions provided by TCP. The main objective of iFCP is to enable the existing Optical Fiber Channel devices to connect and networking at a wire speed on the IP network. IP components and technologies replace the Optical Fiber Channel switching and routing selection structure.
ISCSIInternet SCSI protocol)
ISCSI Internet small computer system interface) is a standard for data block transmission based on IP protocol. It was initiated by Cisco and IBM and has received strong support from IP Storage Technology advocates. The main function of iSCSI is to encapsulate and reliably transmit large amounts of data between the host system initiator on the TCP/IP network and the target of the storage device. Compared with the previous network access storage, iSCSI solves problems such as openness, capacity, transmission speed, compatibility, and security, its superior performance and low cost have attracted the attention and favor of the market since the date of release.
Link failure handling
How to deal with the temporary failure of the communication link in a remote disaster recovery solution is another factor to evaluate whether the solution meets commercial requirements. There are two common solutions for Link failure handling: I/O logs and pointer records. For the I/O log solution, if the link fails, each I/O will copy a log file when it arrives from the host. When the link becomes available, log files are transmitted to the remote site. For the pointer record scheme, the source LUN is divided into blocks. If the link fails, a pointer table is formed, showing which blocks have been modified. After the link is restored, the modified block is re-transmitted to the remote location.
How to have both fish and bear's paw
Remote Disaster recovery systems generally require a distance of more than 100 kilometers. This distance requirement makes transmission the most critical factor in the disaster recovery system.
SAN is currently recognized as the best data transmission method. It can be divided into fc san and ip san Based on different basic networks. Fc san uses full-fiber channels to meet the high performance requirements of remote data transmission. However, the cost of deploying a Remote Disaster Recovery System Based on fc san is extremely expensive, which is not affordable for small and medium-sized enterprises. In addition, the fiber channel technology has its fatal weakness: The effective transmission range of optical fiber can be guaranteed to be only 10 km, which is far from enough for remote disaster recovery systems. Although some people have proposed ways to increase SFP signal strength physically, use the SAN over SONET (SDH) method to enhance Remote Disaster Tolerance over 100 kilometers) Data Transmission Performance, however, this further increases the deployment cost of the disaster recovery solution. In a remote disaster recovery system, data transmission over the IP protocol is the cheapest data transmission method. This technology can achieve Remote Disaster Recovery to some extent, however, due to the limitations of the bandwidth and reliability of the IP network, the data transmission performance is low and the data recovery time is long.
The proposal of protocols such as iSCSI, FCIP, and iFCP can overcome the shortcomings of the Optical Fiber Channel Technology and IP technology, and combine their advantages, it provides a feasible solution for obtaining the optimal cost-effectiveness of the remote disaster recovery system. Using the ip san implemented by iSCSI and integrating the existing fc san and NAS to build a remote disaster recovery system does not have performance problems. It can use the existing disaster recovery system facilities, you can also deploy a new disaster tolerance system using low-cost iSCSI SAN, which is acceptable to various users.
With the popularization and popularization of 10G Ethernet, we believe that the era of Remote Disaster Tolerance is far from us.