please read
http://studentbank.in/report-internet-scsi-iscsi for getting technical report and presentation of iSCSI-Future of Network Storage
INTRODUCTION
The SCSI protocols widely used to access storage devices. The iSCSI protocol is a transport for SCSI over TCP/IP. SM-2 defines an architecture model for SCSI transports, and iSCSI defines such a transport on top of TCP/IP. Other SCSI transports include SCSI Serial and Fibre Channel Protocol (FCP). Until recently standard networking hardware (Ethernet) and IP-based protocols could not provide the necessary high bandwidth and low latency needed for storage access. With the recent advances in Ethernet technology, it is now practical from a performance perspective to access storage devices over an IP network. 1Gb Ethernet is now widely available and is competitive with current 1 and 2 Gb Fibre Channel technology. 10Gb Ethernet will soon also be widely available. Similar to FCP, iSCSI allows storage to be accessed over a storage area network (SAN), allowing shared access to storage. A major advantage of iSCSI over FCP is that iSCSI can run over standard off-the-shelf network components, such as Ethernet. Furthermore, iSCSI can exploit IP-based protocols such as IPSec, for security and Service Location Protocol (SLP) for discovery. A network that incorporates iSCSI SANs need use only a single kind of network infrastructure (Ethernet) for both data and storage traffic, whereas use of FCP requires a separate kind of infrastructure (Fibre Channel) for storage (Fig.2). IP based SANs using iSCSI can be managed using existing and familiar IP-based tools such as Simple Network Management Protocol (SNMP) whereas FCI SANs require specialized management infrastructure. Furthermore, iSCSI-based SANs can extend over arbitrary distances, just like TCP, and are not subject to distance limitations that currently limit FCP.
Fig(1)
Fig.(2) Classic SAN vs iSCSI
In addition to iSCSI, several other protocols have been defined to transport storage over an IP network. FCIP is used to connect separate islands of Fibre Channel SANs over an IP network to form a single unified SAN. iFCP is a gateway-to-gateway protocol for the implementation of Fibre Channel fabric functionality on a network in which TCP/IP switching and routing elements replace Fibre Channel components. Whereas, FCIP and iFCP are used to allow the connection of existing Fibre Channel infrastructures to each other and to IP network. iSCSI enables the creation of SANs complete independent of Fibre Channel.
FACTORS ENABLING IP STORAGE
SAN deployments have been driven by an insatiable demand for storage and the user benefits delivered by networked storage. These benefits include efficient storage. These benefits include efficient storage utilization through storage consolidation, the ability to manage more storage capacity, rapid deployment of new storage, higher availability, and faster backup and restore operations.
With the recent development of the iSCSI protocol and silicon-based TCP/IP offload engines, SANs based TCP/IP networks are now possible. The IP networking infrastructure includes multi-gigabit networks, sophisticated bandwidth allocation and network management tools, and the ubiquitous reach of IP and Ethernet. These factors together enable new IP Storage solutions, using iSCSI protocol.
Fig.3 Factors enabling iSCSI storage
LEVERAGING THE BEST FROM STORAGE AND NETWORKING
iSCSI builds on the two most widely used protocols from the storage and the networking worlds. From the storage side, the iSCSI used the SCSI command set, the core storage commands used throughout all storage configurations.
On the networking side, iSCSI uses IP and Ethernet, which are the basis for the most corporate networks, and are increasingly being used for metropolitan and wide area networking as well. With almost eighty years of research, development and integration, IP networks provide the almost in manageability, interoperability and cost-effectiveness.
Fig.4 Leveraging SCSI and IP for Storage Fig.5 iSCSI Protocol
Network Stack
Fig.4 shows how SCSI is mapped to TCP/IP through the discs layer, freeing SCSI from its parallel bus structure.
Fig.5 shows a simplified version of a protocol stack including iSCSI. Use of the standard SCSI command set facilities interoperability with existing operating system and applications that lay above that layer. Additionally, use of a standard TCP/IP network provides universal reach to the global IP infrastructure.
ADVANTAGES OF IP STORAGE
*Familiar network technology and management
Reduces training and staff cost
*Proven transport infrastructure
Increases reliability
*Transition from 1Gigabit Ethernet to 10 Gigabit Ethernet and beyond
Protects investment with simplified performance upgrades
*Scalability over long distances
Enables remote data replication and disaster recovery
*Brings Ethernet economics to storage
Enables lower total cost of ownership
ELEMENTS OF IP STORAGE NETWORKING
IP Storage Networking refers to computer systems and storage elements that are connected via IP. It also refers to the IP infrastructure transporting storage traffic among these targets. Fig.6 shows the various elements of an IP storage network.
The first element, Device I/O, refers to computer systems and storage resources that have native IP interfaces. These could include servers, disk arrays or tape libraries with an iSCSI adapter or iSCSI controller. The interfaces typically are Ethernet and include protocol processing, such as TCP/IP offload engines to reduce processing loads on the host devices.
The second element of an iSCSI storage area network is the fabric of switches. The advantage of an IP based fabric is that users can create a SAN using standard Ethernet switches and routers to transport data over the SAN fabric. The fabric also may include storage routers and switches that have a combination of iSCSI interfaces and other storage interfaces such as SCSI or Fiber Channel. Storage switches and routers provide multi-protocol connectivity not available in conventional IP and Ethernet switches. They also provide storage specific functionality, such as peer-to-peer copy commands.
The third element of IP storage networking is the SAN interconnection. Since iSCSI is a native IP based protocol, SAN interconnection does not require storage-specific functionality and can use a shared or dedicated IP/Ethernet network.
Fig.6 Segments of Storage Networking
iSCSI DEFINED
Internet SCSI (iSCSI) is a draft standard protocol for encapsulating SCSI command into TCP/IP packets and enabling I/O block data transport over IP networks. iSCSI can be used to build IP-based SANs. The simple, yet powerful technology can help provide high-speed, low-cost, long-distance storage solution for Web sites, service providers, enterprises and other organizations. An iSCSI HBA, or storage NIC, connects storage resources over Ethernet. As a result, core transport layers can be managed using existing network management applications. High-level management activities of the iSCSI protocol “ such as permissions, device information and configuration “ can easily be layered over or built into these applications. For this reason, the deployment of interoperable, robust enterprise management solutions for iSCSI devices is expected to occur quickly.
First-generation iSCSI HBA performance is expected to be well suited for the workgroup or departmental storage requirements of medium- and large-sized businesses. The expected availability of TCP/IP Offload Engines in 2002 will significantly improve the performance of iSCSI products. Performance comparable to Fibre Channel is expected when vendors begin shipping 10 Gigabit Ethernet iSCSI products in 2003.
Fig 7. Layered View
BENEFITS OF ISCSI
By combining SCSI, Ethernet and TCP/IP, Gigabit iSCSI delivers these key advantages:
¢ Builds on stable and familiar standards “ many IT staffs are familiar with the technologies
¢ Creates a SAN with a reduced TCO “installation and maintenance costs are low since the TCP/IP suite reduces the need for hiring specialized personnel
¢ Provides a high degree of interoperability “ reduces disparate networks and cabling, and uses regular Ethernet switches instead of special Fibre Channel switches
¢ Ethernet transmissions can travel over the Global IP Network and therefore have no practical distance limitations
¢ Scales to 10 Gigabit “ comparable toOC-192 SONET (Synchronous Optical Network) rates in Metropolitan Area Networks (MANs) and Wide Area Networks (WANs)
WHO CAN USE IT
iSCSI SANs are most suitable for organizations with a need for streaming data and/or large amounts of data to store and transmit over the network.
This includes:
¢ Internet Service Providers (ISPs)
¢ Storage Service Providers (SSPs)
¢ Organizations that need remote data replication and disaster recovery. For example, a high-technology company in San Jose remains susceptible to disaster if it uses a Fibre Channel SAN. Original and backup data copies could be lost in the same earthquake due to distance limitations.
¢ Geographically distributed organizations that require access to the same data on a real-time basis. For example, work team members who need the latest project data without waiting 24 hours for traditional replication/backup/ reconciliation procedures.
¢ Businesses and institutions with limited IT resources, infrastructure and budget. These organizations should look for iSCSI equipment those functions over standard Gigabit Ethernet Cat-5 copper cabling already in place in most buildings today.
ISCSI FEATURES
TCP
TCP was chosen as the transport for iSCSI. TCP has many features that are utilized by iSCSI :
1. TCP provides reliable in-order delivery of data.
2. TCP provides automatic retransmission of data that was not acknowledged.
3. TCP is a friendly network citizen in that it provides the necessary flow control and congestion control to avoid overloading a congested network.
4. TCP works over a wide variety o physical media and interconnect topologies.
While other protocols such as SCTP also provide many of these features, TCP has the advantage of having been deployed for decade and is therefore better understood and more widely available.
SESSIONS
SCSI commands are typically issued by a storage initiator (client) to a storage target (server). The relationship between SCSI entities is referred as Ëœnexusâ„¢. The iSCSI entity corresponding to an I_T_NEXUS (initiator-target nexus) is an iSCSI session. An iSCSI session is a collection of TCP connections between an iSCSI initiator and an iSCSI target used to pass SCSI commands are data between the initiator and the target.
Even though a single TCP connection is sufficient to establish communication between an initiator and a target, it is often advantageous use multiple connections:
1. It is often not possible to achieve the maximum bandwidth of the underlying physical interconnect using only a single TCP connection.
2. When working on a multiprocessor machine, it may be advantageous to allow separate threads running on the different processors to simultaneously utilize different TCP connections.
3. If there are more than one physical interconnect between the initiator and the target, spreading multiple TCP connections over all the possible physical interconnects can aggregate their bandwidth.
ISCSI PROTOCOL DATA UNITS
iSCSI defines its own packets, referred to as iSCSI protocol data units (PDUs). iSCSI PDUs consist of a header and possible data, where the data length is specified within the iSCSI PDU header. An iSCSI PDU is sent as the contents of one or more TCP packets.
The most commonly used iSCSI PDU types are :
(i) SCSI Command/Response.
(ii) Data In/Out
(iii) Ready to Transfer (R2T).
(iv) Login Request/Response.
The SCSI Command PDU is used to transfer a SCSI command from the initiator to the target. If the SCSI commands requests to read data from the target, the target will send the data to the initiator in one or more Data In PDUs. If the SCSI command requests to write data to the target, the initiator will send data to the target in one or more Data Out PDUs. The target may specify to the initiator which part of the data to send by sending to the initiator an R2T PDU. Upon completion of the entire data transfer, the target sends a SCSI Response PDU to the initiator indication either successful completion of the command or any error condition detected. For each SCSI Command PDU there is a corresponding single SCSI Command PDU there is a corresponding single SCSI Response PDU, but possibly multiple (or no) Data PDUs. SCSI Data and Response PDUs must be sent over the same TCP connection on which their corresponding SCSI Command PDU was issued.
LOGIN
Immediately upon establishing a TCP connection between an iSCSI initiator and iSCSI target, a login procedure must be performed. The initiator sends a Login Request PDU to the target. The initiator and target may authenticate each other and may negotiate operational parameters. A default authentication method, Challenge-Handshake Authentication Protocol (CHAP), must be supported by all compliant iSCSI implementations. Some of the operational parameters that may be negotiated are the maximum number of connections to be used in the session, the amount of unsolicited data that may be sent by the initiator, the level of error recovery supported, and whether or not digests will be used for error detection. After both sides are satisfied with the authentication and the setting of the operational settings, the target sends a Login Responds PDU with an indication that the login procedure has completed. Only then may the connection be used to pass SCSI commands and data.
NAMING
Borrowing from other Internet protocols, iSCSI use an URL-like scheme to name targets. ISCSI names are meant to be global, similar to World Wide Names used by Fibre Channel. An iSCSI entity might have its IP address changed while retaining its name. An iSCSI entity is therefore identified by its name and not its address. This allows for easier handling of iSCSI names by proxies, gateways, network address translation boxes, firewalls, and so on. iSCSI names should be unique worldwide. Typical iSCSI names might look like this:
Iqn.2001-04.com.acme
torage.disk2.sys1.xyz
DISCOVERY
When using storage devices over a network, one has to deal with the ability of an initiator to discover the devices it may use. One approach is for an administrator to statically configure the initiator, providing the initiator with a list of names and addresses of the iSCSI devices to which the initiator may connect. If additional iSCSI devices were later added to the network, the statically configured initiator would not be able to access the new devices without being reconfigured. An alternative more dynamic method is to use SLP, which already exists in the IP family of protocols. iSCSI targets can be added to the network, and the topology can change over time, but initiators can easily find new targets without having to be reconfigured. A similar mechanism is provided by the recently defined iSNS protocol.
An additional discovery mechanism, Send Targets, is provided in the iSCSI protocol itself, especially useful for gateway devices. In this method, an initiator is statically configured to connect to specific iSCSI gateway devices. The initiator establishes a discovery session with the iSCSI gateway device then responds with a list of attached iSCSI targets that are available to the initiator. The initiator may then proceed to connect to the specified iSCSI target devices.
DATA INTEGRITY
TCP has a checksum facility to help detect errors that occur during transmission. While the probability of the TCP checksum failing to detect an error is quite small, it is not good enough for some storage environments. The TCP checksum also does not provide protection for corruptions that occur while a message is in the memory of some router. iSCSI therefore defines its own cyclic redundancy check (CRC) checksum to ensure end-to-end integrity of its packet headers and data. Initiators and targets may negotiate whether or not to use this CRC checksum.
SECURITY
When storage devices were directly attached to host machines, the data on the storage devices was considered secure by its being inaccessible to the outside world. With iSCSI attached storage devices, this is no longer the case. A serious security problem may arise if sensitive storage data is accessed over a general data network. One possible solution is to use a physically separate network for the storage data, similar to what is done with Fibre Channel (Fig.2. This solution requires a second physical IP network which is still cheaper than having a second physical Fibre Channel network. Alternatively, a single physical IP network can be used to get with encryption of the storage data. Encryption of data on an IP network can be provided with IPSec. iSCSI simply uses the existing IP-faulty security protocol to protect sensitive storage data from possible security attacks such as spoofing.
ANTICIPATED USE OF iSCSI
Some of the design decisions of iSCSI were strongly influenced by the perception of how iSCSI would eventually be used. ISCSI was designed to allow efficient hardware and software implementations to access I/O devices attached over any IP network. iSCSI was designed for a wide variety of environments and applications including local and remote storage access, local and remote mirroring, local and remote backup/restore. It was assumed that TCP/IP acceleration adapters and even iSCSI host bus adapters (HBAs) would become prevalent, and would be strongly desirable to defend he protocol to allow high-performance adapter implementations Mechanisms were therefore included to overcome various anticipated problems, such as maintaining high bandwidth frequently dropped packets. Care was taken to not limit the application of iSCSI to disks; mechanisms were provided for various types of SCSI devices, especially tapes, for which it is inconvenient and perhaps prohibitive to cancel a restart commands.
DIRECT DATA PLACEMENT
In typical TCP implementations, data that arrive on a TCP connection is first copied into temporary buffers. The TCP driver then examines the connection identification information (source and destination address and port numbers) determine the intended receiver of the data. The data is then copied into the receiverâ„¢s buffer. For SCSI data, there might be many pending SCSI commands at any given instant, and the received data typically mist be copied into the specific buffer provided by the SCSI layer for the particular command. This entire procedure might require the receiving host to copy the data a number of times before the data ends up in the final destination buffer. Such copies require significant amount of CPU and memory buffer usage that would adversely affect the system performance. It is therefore most desirable to the able to place the data in its final destination with a minimum number of copies.
iSCSI Data PDU headers contain sufficient information allow an iSCSI adapter (HBA) perform direct data placement. The information provided in an iSCSI Data PDU header include a transfer tag to identify the SCSI command and its corresponding buffer, a byte offset relative the beginning of the corresponding buffer, and the data length parameter indicating the number of bytes being transferred in the current data packet. This information is sufficient to enable direct placement of the arriving data into pre registered SCSI-provided buffers. An iSCSI adapter then performs both TCP and iSCSI processing on the adapter will have sufficient information in the TCP and iSCSI headers to place arriving iSCSI data directly into the appropriate SCSI buffers without having to copy the data into additional temporary buffers on the host machine.
RECOVERY
The iSCSI protocol defines several levels of recovery to provide resilience in the face of a wide range of possible errors and failures. iSCSI handling and recovery is expected a rare occurrence, and may involve a significant amount of overhead. It is anticipated that most computing environments will not need all the levels of recovery defined in the iSCSI specification.
The most basic recovery class is session failure recovery. All iSCSI specification complaint implementations must implement session failure recovery. Session recovery involves the closing of all of the sessions TCP connections aborting all outstanding SCSI commands on that session, terminating all such aborted SCSI commands with an appropriate SCSI service response at the initiator and the restarting a new set of TCP connections for the particular session. Implementations may perform session failure recovery for any iSCSI error detected.
A less drastic kind of recovery implementations may perform its digest failure recovery. If a CRC checksum error is detected on iSCSI data, the data packet must be discarded. Instead of performing session recovery, implementations may use the digest failure recovery mechanism to ask the connecting peer to resent only the missing data. Similarly, if a sequence reception timeout occurs, a similar mechanism can be used to ask the connecting peer to resent the missing commands, responses or other number packets that are expected.
If a CRC checksum error is detected on an iSCSI packet header, the packet must be discarded since it was corrupted. As a result, synchronization between the initiator and target may be lost. The iSCSI protocol allows for a new TCP connection to be established within the session, and defines mechanisms for the initiator and target to synchronize with one another to continue to smoothly interact. A new TCP connection may be designated to take over from an old TCP connection that seems to have become defective. This level of recovery is called connection recovery. Processing of commands that were started on the defective TCP connection can be continued on the new TCP connection.
UNDERSTANDING END SYSTEMS
An iSCSI environment can leverage existing IP networks; a significant portion of the overall solution involves iSCSI end systems. Fig.8 outlines the significant functions of traditional IP Network Interface Cards, traditional Storage Adapters, and new IP Storage Adapters.
Fig.8 Network Interface Cards and Host Bus Adapters
Network Interface Cards (NICs)
Traditional NICs (Ethernet adapters in servers and PCs) are designed to transfer packetized file level data among PCs, servers and storage devices, such as NAS appliances. However, NICs do not traditionally transfer block level data, which is handled by a storage host bus adapter, such as Fibre Channel or parallel SCSI. In order for a NIC to process block level data the data needs to be placed into a TCP/IP packet before being send over the IP network. Through the use of iSCSI drivers on the host or server, a NIC can transmit packets of block level data over an IP network. When using an NIC, the server handles the packet creation of block level data and performs all of the TCP/IP processing. This is extremely CPU intensive and lowers the overall server performance. The TCP/IP processing performance bottleneck has been the driving force behind the development of TCP/IP offload engines (TOE) on adapter cards. A TOE removes TCP/IP processing from the host CPU and completes TCP/IP processing and packet creation on the HBA. Thus a TCP/IP offload storage NIC operates more like a storage HBA rather than a standard NIC.
Storage Host Bus Adapters (HBAs)
Unlike NICs, Storage HBAs are designed to transmit block level data to and from storage applications. A reference to the entire block is transferred from the application to the adapter, bypassing the need to bread the bloke into smaller frames; a process that takes place through a set of specialized chips that enable the HBA to relieve the computer resources of the CPU for this process. When it completes the task, the HBA forwards the frames.
ISCSI ADAPTERS
iSCSI Adapters combine the functions of NICs with the function of storage HBA. These adapters take the data in block form, handle the segmentation and processing on the adapter card with TCP/IP processing engines, and then send the IP packets across an IP network. The implementation of these functions enables users to create an IP based SAN without lowering the performance of the server. In advance of the introduction of iSCSI adapters, some vendors released software versions of these adapters. These software-enabled adapters accept block level data from applications, but still require CPU cycles for the TCP/IP processing. The advantage of such adapters is that they can work on existing Ethernet NICs. The main disadvantage is that they require heavy CPU utilization for TCP/IP processing.
DEPLOYMENT SCENARIOS FOR ISCSI
BUILDING NEW NATIVE ISCSI STORAGE NETWORKS
DATA CENTER APPROACH
An iSCSI SAN is a perfect choice for a user interested in moving to networked storage. Using the same block level SCSI commands as direct attach storage, iSCSI provides compatibility with user applications such as file systems, databases, and web serving. Similarly, since iSCSI runs on ubiquitous and familiar IP networks, there is no need to learn a new networking infrastructure to realize SAN benefits. To build an iSCSI storage network in a data center, iSCSI host bus adapters can be used in servers, along with iSCSI storage devices and a combination of IP and Ethernet switches. IP Storage switches and routers can be used if required. Fig.8 shows aprimary data center iSCSI storage network connecting to remote sites.
Fig.9 Data Center iSCSI Storage Networks Offer Seamless MAN/WAN Access
EXPANDING ISCSI STORAGE NETWORKS TO METRO AND WIDE AREA NETWORKS
The rapid adoption and expansion of IP data on the Internet has proves the viability of using IP across long distance wide area networks. Although it is expected that initial deployments of iSCSI will use private networks, the use of IPâ„¢s security infrastructure (such as IPSes and SSL to provide authentication and privacy) it will be possible to use public networks for wide area iSCSI storage traffic as well.
APPLICATIONS FOR NATIVE ISCSI STORAGE NETWORKS
With native iSCSI storage networks customers can benefit from the following applications in the data center:
(1) Server and storage consolidation
With a networked storage infrastructure, customers can page link multiple storage devices to multiple servers allowing for better resource utilization, ease of storage management and simpler expansion of the storage infrastructure.
(2) Accelerated Backup Operations
Backup operations previously restricted to operating across traditional IP LANs at the file level can now operate across IP Storage networks at the block level. This shift facilitates faster backup times, and provides customers the flexibility to use shared or dedicated IP networks for storage operations.
(3) Seamless Remote Site Access and Storage Outsourcing
With the storage network based on IP, customers can easily enable remote access to secondary sites across metropolitan or wide area IP networks. The remote sites can be used for off-state backup, clustering or mirroring/replication. Additionally, customers can choose to page link to storage service providers for storage outsourcing applications such as storage-on-demand
LINKING SANs WITH iSCSI
The iSCSI protocol is ideal for linking SANs over MAN and WAN environments wince it uses TCP/IP as the transport. This includes extending native iSCSI SANs and Fibre Channel SANs across a wide area network. iSCSI SANs can be connected over a wide area network with standard Ethernet equipment. When connecting to Fiber Channel SANs, an IP Storage Switch is needed to convert the FC protocol to iSCSI.
IP Storage routers can bridge FC to iSCSI, while IP storage switches bridge FC to iSCSI and provide added switching functionality. That functionality, when provided in an IP Storage switch, allows users to perform FC-to-FC switching, FC-to-iSCSI switching, or FC-to-Gigabit Ethernet switching and Gigabit Ethernet to Gigabit Ethernet switching. Both IP storage routers and switches allow users to extend the reach of the FC SAN and bridge FC SANs to iSCSI SANs. Fig.10 shows a sample configuration based on IP storage routers and switches.
Fig.10 Linking iSCSI to Fibre Channel
Fig.(11)
APPLICATIONS FOR LINKING ISCSI TO FIBRE CHANNEL
IP Access to Storage/Storage Consolidation
Using iSCSI Adapters in servers, customers can now provide access to FC storage resources across an IP network. This greatly expands storage access flexibility, allowing customers to consolidate FC storage without limiting access to servers. Database information, for example, can be directly accessed from servers across an IP network.
Remote Backup for Enterprise Customers
With the ability to use IP networks, the combination of iSCSI and IP Storage routers or switches enables remote backup for enterprise customers. Remote sites can operate independently, yet still benefit from enterprise storage resources for iSCSI server to FC storage backup and recovery. This application allows data center managers to centralize corporate data resources in one location while providing sophisticated enterprise storage management to several remote customers at different sites.
DEPLOYMENT EXAMPLES
Network Storage Services via iSCSI
Two iSCSI HBAs can be used in conjunction with standard Ethernet NICs through a Gigabit-capable switch connected to an iSCSI-capable RAID Array (Figure12). This configuration is appropriate as either the next step in transitioning to an iSCSI-exclusive SAN or as an initial iSCSI SAN configuration.
Fig.12 Network Storage Services via iSCSI
Multiple Cards to Single iSCSI Router
Multiple HBAs in separate servers can be used in conjunction with a Gigabit capable switch connected to an iSCSI capable router with Fibre Channel ports. This is then connected directly to a native Fibre Channel RAID Array (Figure 13). This configuration is appropriate as the next step in transitioning to an iSCSI front-end SAN with Fibre Channel storage devices.
Fig.13 Multiple Cards to Single iSCSI Router
iSCSI HBA and Fibre Channel Tape Backup
An iSCSI HBA can be used in conjunction with a Gigabit-capable switch connected to an iSCSI-capable router with Fibre Channel ports connected to a Fibre Channel tape drive (Figure 14). This configuration can be used as a means to perform backup and recovery using existing Ethernet infrastructure.
Fig.14 iSCSI HBA and Fibre Channel Backup
CONCLUSION
The iSCSI protocol enables access to storage devices over an IP network. One of the main objectives in defining the iSCSI protocol was to make use of existing IP infrastructure whenever possible. Thus, TCP is used as the underlying transport, IPSec is exploited to provide network security, SLP can be used to provide discovery, and so on. Since iSCSI runs over standard off-the-shelf network compoens, the cost of setin up an iSCSI SAN is significantly lower than that of a Fibre Channel SAN. This will make iSCSI more affordable and manageable than Fibre Channel, and enable it to become the protocol of choice for SANs. iSCSI will expand the market for networked storage by giving IT managers another alternative to direct attached storage that delivers the advantages of networked storage. IP storage networks take advantage of IP networking knowledge in IT departments and use existing network management and tools for LANs, MANs and WANs today. Riding the IP wave of technology develoment and enhancements like the introduction of 10 Gigabit Ethernet, iSCSI provides a logical unified infrastructure development path for corporations and service providers alike.
REFERENCES
¢ Features of iSCSI Protocol, Kalman. Z. Meth and Julian Satran, IBM Haifa Research Lab, IEEE Communication Magazine “ August 2003.
¢ Performance Study of iSCSI-Based Storage Subsystems, Yingping Lu and David. H. C. Du, University of Minnesota, IEEE Communication Magazine “ August 2003.
¢ intel.com
¢ entecollege.com
¢ ibm.com
¢ google.com
¢ delpowersolutions
ACKNOWLEDGEMENT
I extend my sincere thanks to Prof. , Head of the Department for providing me with the guidance and facilities for the Seminar.
I express my sincere gratitude to Seminar coordinator Mr. , Staff in charge, for his cooperation and guidance for preparing and presenting this seminars.
I also extend my sincere thanks to all other faculty members of Electronics and Communication Department and my friends for their support and encouragement.
Entecollege.com
CONTENTS
¢ INTRODUCTION
¢ ELEMENTS OF IP STORAGE
¢ FACTORS ENABLING IP STORAGE
¢ LEVERAGING THE BEST FROM STORAGE AND NETWORKING
¢ ADVANTAGES OF IP STORAGE
¢ ELEMENTS OF IP STORAGE NETWORKING
¢ ISCSI DEFINED
¢ BEBEFITS OF ISCSI
¢ WHO CAN USE IT
¢ ISCSI FEATURES
TCP
SESSIONS
ISCSI PROTOCOL DATA UNITS
LOGIN
NAMING
DISCOVERY
DATA INTEGRITY
SECURITY
ANTICIPATED USE OF ISCSI
DIRECT DATA PLACEMENT
RECOVERY
¢ UNDERSTANDING END SYSTEMS
¢ DEPLOYMENT SCENARIOS FOR ISCSI
¢ APPLICATIONS FOR LINKING ISCSI TO FIBRE CHANNEL
¢ DEPLOYMENT EXAMPLES
¢ CONCLUSION
¢ REFERENCES
ABSTRACT
In a world where Internet Protocol (IP) dominates local and wide area networks, the data storage requirements grow unabated; it seems inevitable that these two forces converge. The Internet Small Computer Systems Interface (iSCSI) protocol unites storage and IP networking. iSCSI enables the transport of block-level storage traffic over IP networks. It builds on two widely used technologies - SCSI commands for storage traffic over IP networks. It builds on two widely used technologies - SCSI commands for storage and IP protocols for networking. iSCSI is an end-to-end protocol for transporting storage I/O block data over an IP network. The protocol is used on servers (initiators), storage devices (targets), and protocol transfer gateway devices. iSCSI uses standard Ethernet switches and routers to move the data from server to storage. It also enables IP and Ethernet infrastructure to be used for expanding access to SAN storage and extending SAN storage and extending SAN connectivity across any distance. This paper presents general overview of iSCSI, outlining its introduction, details of the protocol, its features, user benefits, and several typical deployment scenarios and applications.
.