Storage Fundamentals 1

---------------------------------------------------------------------------------------------------

1. Network Storage main Technology

2. Major protocols and related technologies

3. File system

4.RAID Technology

5. Data replication and disaster recovery

6. Backup Technology

7.windows related

8.linux related

9. Storage Network Knowledge

10. Storage I/O

11. Cloud Storage

12. Other Storage

13. Information

--------------------------------------------------------------------------------------------------------------- -----------------

1. Network Storage main Technology 1.1 NAS

About NAS

In the early 1980s, Professor Brian Randell of the University of Newcastle, UK, and colleagues successfully demonstrated and developed remote file access on the entire UNIX machine through Newcastle Connect. Following the Newcastle connection, 1984 Sun released an NFS protocol that allowed Web servers to share their storage space with network customers. In the early 90 Auspex engineers created an integrated NetApp file manager that supports the Windows CIFS and UNIX NFS protocol, with excellent scalability and ease of deployment, with dedicated NAS devices on the market. In just a few years, Nas has become one of the flagship devices for networked data storage solutions with the central idea of simple and efficient applications. EMC company Celerra Products are now leading the global NAS market with outstanding performance and versatility.

NAS Concepts

NAS (network-attached Storage, Network attached storage) is a file-level computer data store connected to a computer network that can provide data access to different clients.

NAS is defined as a special dedicated data storage server that includes memory pieces (network devices for one or more hard drives, usually arranged as logical, redundant storage containers or RAID arrays) and embedded system software that provide cross-platform file sharing capabilities. Nas usually occupies its own node on a LAN, without application server intervention, allowing users to access data on the network, in which the NAS centrally manages and processes all data on the network, offloading the load from the application or enterprise server, effectively reducing the total cost of ownership and protecting the user's investment.

The NAS itself can support a variety of protocols (such as NFS, CIFS, FTP, HTTP, etc.) and can support various operating systems. NAS is a truly plug-and-play product with a flexible physical location that can be placed in a workgroup or in a hybrid environment, such as a hybrid unix/windows LAN environment without any modifications to the network environment. NAS products are directly connected to the network via a network interface, which can be shared by users on the network simply by configuring the IP address.

NAS Features

Compared with the scenario using the Storage Area Network (san-storage) , the scheme with the Network attached storage (nas-network-attached Storage) architecture has the following characteristics:

1). Network-centric, open standard protocol support

Different from the design of Storage Area network (SAN), Network access Storage (NAS) mode is network-centric. It protects users ' investment in Ethernet by leveraging existing Ethernet network resources to access dedicated networked storage devices instead of deploying expensive fibre switch networks to connect traditional storage devices.

In recent years, the transmission bandwidth of Gigabit Ethernet (1000Mbps, 125MB/S) has gained popularity and is expected to evolve towards Gigabit Ethernet. By then, the transmission bandwidth of the Ethernet will be 10 times times the transmission bandwidth of the various SCSI and Fiber channel protocols that the San relies on to survive. EMC company Celerra Products support the most popular TCP/IP network protocols, while the NFS and CIFS file service protocols used are industry standard protocols that are fully device compatible.

2). Stand-alone operating system

Celerra's Dart operating system has independent intellectual property rights, focusing on the transmission of file systems. The operating system is powerful and superior in performance, guaranteeing the high speed and reliable transmission of the file system. The Celerra backend connects back-end storage devices through a SAN network with multiple link redundancy to avoid a single point of failure and ensure data security. As long as the user's data is saved in one copy, it can be used by various types of hosts on the front end, so it has host independence. Celerra's Dart operating system also guarantees data sharing for UNIX and Windows on different operating systems, and the respective access rights can be guaranteed accordingly.

3). Easy Installation and Management

NAS does not need the server direct Internet, but the user-oriented, dedicated to the data storage of the simplified operating system, built-in with the network to connect the required protocols, the overall system management and setup is simpler. Celerra provides access to different types of hosts on the front-end without having to shut down without having to install any hardware or software on the host as long as the existing network has an idle network port.

4). NAS Underlying protocol

The NAS uses the NFS (Sun) communication Unix Camp and the CIFS Communication NT Camp, which also reflects the NAS's "file-level" read and write operations based on the operating system, and access requests are based on "file handle + offset".

1.2 SAN

the SAN (short for Storage area Network) is a memory-region net that uses Fibre Channel (Fibre) technology to connect storage arrays and server hosts via Fibre Channel switches. Establish a regional network dedicated to data storage . San Networked storage is a high-speed network or sub-network that provides data transfer between the computer and the storage system. A SAN network consists of a communication structure that is responsible for network connectivity, the management, storage parts, and computer systems that are responsible for the organization of the connection, so that SAN technology guarantees the security and strength of the data transfer. The SAN has several advantages:

The scalability of 1.SAN means that you have a small number of disks that are not limited by the connection to the system. Sans can grow to hundreds of disks, but there are only more than 10 limits for a normal physical server.

2.SAN performance is not constrained by Ethernet traffic or local disk access. Data is routed through a SAN from its own private network, separating user traffic, backup traffic, and other SAN traffic.

3. In the right configuration environment, SAN data is partitioned by region. The partition where the user holds the data is isolated from the same San.san zone as the other person, as if the UNIX server and the Windows Server are connected to the same SAN, but the data access is different on both servers, in fact, the Windows system cannot "see" the UNIX data, Vice versa.

The 4.SAN system does not require a reboot to add new disks, replace disks, or configure RAID groups. The data flow completely bypasses the server system, and the SAN also increases data backup and recovery performance.

5. Partitions can also separate your workloads on the SAN. Not only is your data separated from protection, but it is also shielded from unrelated workloads that affect application performance. Applying San Sharing in the right areas is not a performance problem.

The 6.SAN has an unparalleled advantage, that is, the storage connection distance is 10 km distance (about 6 miles). Not that you will use this advantage, but it will come out when you need it. With a distance advantage, data can be stored in a separate location from the system services.

7. The efficient use of automated thin provisioning on storage networks such as Sans is higher than local storage. When a system requires more storage resources, the San dynamically allocates resources. This means that the physical system can enjoy the auto-thin configuration, just like virtualization.

1.3 The difference between SAN and NAS

1.4 SCSI Access Control principle

SCSI-2 Reserve (Reserved)/release (release)/reset (reset)

The client access LUN process in the SCSI-2 protocol is as follows:

1. The client initiates a reservation operation to the LUN

2, the client obtains the LUN operation permission after the reservation operation succeeds, the reservation fails, prompts the reservation conflict, continues the attempt until the reservation succeeds.

3. After the client operation is completed, the release operation is performed and other clients can reserve.

The main disadvantages of SCSI-2 access control are:

1, the reservation operation is based on the path. Reservation and release must be done by the same client, one host cannot release the reservation of another host, and the same host HBA card cannot cancel the reservation of another HBA of the same host.

2, reservations can not be retained for long. The host reboot will lose the reservation information.

3. If the LUN is already reserved, no additional hosts can be reserved. If other hosts want to obtain LUN operation permissions, the LUN must be reset and the reset operation may result in data loss. Resets the existing reservation of the LUN after reset, the reset operation is initiated by the LUN, the original reserved host is not known.

SCSI-3 Persistent Reserve (PR)/ PREEMPT (preemption)

The SCSI-3 protocol introduces PGR (persistent group reservation) functionality. Before accessing the LUN, the client first registers (registration) a reserved key (reservation key) with the LUN, and the client can attempt to make a permanent reservation (reserve) After successful registration, and the LUN operation permission can be obtained after the permanent reservation succeeds. The reserved key is a string of 16 ASCII codes, up to 8 bytes in length. There are 6 types of permanent reservations, represented by 1, 3, 5, 6, 7, 8 digits. Includes two operation types and three customer types, including write-through and all access exclusive, customer types include all clients, registered clients, and owning clients. The corresponding relationship between the number and the permanent reservation type is as follows:

1-> Write Exclusive

3-> Exclusive access

5-> Write exclusive-registrants only

6-> Exclusive Access-registrants only

7-> Write Exclusive-all Registrants

8-> Exclusive Access-all registrants.

Different registration types correspond to different access rights. Unlike SCSI-2, the SCSI-3 release operation is based on the reserved key. Different clients can use the same key or different keys for reservations, depending on the type of permanent reservation. Clients can obtain access to LUNs that have been permanently reserved by preemption. SCSI-3 preemption and SCSI-2 reset are not the same, preemption does not cause data loss.

SCSI-3 about PGR related operations commands are divided into two main categories: Prin and Prout respectively. Prin is primarily used for queries, Prout for modification. During the execution of the SCSI command, it is necessary to specify which type of the command is.

Common usage scenarios

1. Cluster I/O Fencing

In order to prevent the cluster fault "brain crack" phenomenon, 2-node cluster can be triggered by SCSI-2 reseve/release I/O fencing to ensure that the entire cluster operation, is SCSI-2 not applicable to multi-node cluster, multi-node cluster can use SCSI-3 PGR. Mainstream vendor cluster kits have already supported SCSI-3 PGR, such as VCS, Hacamp, RHCs, and so on.

2. Cluster file system

The cluster file system needs to ensure data consistency when multiple nodes access the storage at the same time, scsi-2/scsi-3 can be satisfied, when a node tries to access an already reserved storage will generate access rights violation. SCSI-3 PGR can reduce access conflicts more than SCSI-2 reserve/release.

Summary:

SCSI-2 has specific basic access control capabilities, but it cannot meet the requirements of active/active multi-Path environment and cluster multi-node access storage. SCSI-3 by introducing the concept of client registration and operation Rights Classification, we strengthen the control of concurrent access rights and make up the SCSI-2 of the problem.

Storage Fundamentals 1