Failover at multiple level in storage networks

Storage networks can be broadly classified into 3 levels and the fail over or Redundancy feature is very important aspect of IT storage business and it is implemented in every interconnecting components in the networks.
Host or Server(s) Level
Network Level
Storage sub-system(s) Level

Fail-Over:
If the data passage path of any of the component in the storage networks fails, the auto-mechanism that provides an alternate path instantly to avoid the data loss is called fail-over.
Example: If a component such as a RAID controller module or a cable fails, or an error occurs on the data path to the preferred RAID controller module, virtual disk ownership is moved to the alternate non preferred RAID controller module for processing. This failure or error is called failover.

Host or Server Level:
Hosts or servers are associated with few interconnect peripherals such as Host Bus Adapter [HBA], Ethernet adapters et al. some common fail-over feature implementations that you see is at the HBA device in the form of Fail-over driver such as MPP or MPIO [Multi Path driver].
Another stage of fail-over implementation could be to the entire host or server in the form of Clustering, Microsoft Clustering or Linux or Solaris or Unix clustering.

Network Level:
At Switches or Routers, we can provide fail-over by connecting one port or path of the servers & storage sub-systems to one switch port & another port or path connected through another switch port. thus we need at the least two switches to handle the Preferred path & Alternative path of data passage.
usually such setups are know as High Availability Fail-over setup.

Sub-system Level:
Within the SAN storage sub-systems, there are multiple components such as RAID Processors or controllers, Hard disks, Drive trays, power supplies et al.
At Hard disk level , RAID & Hot sparing is used in co-operation with RAID processors as fail-over.
At Drive tray or Enclosure level, there will be dual Service Modules that act as fail-over.
At the RAID Processor or Controller level, there will be Dual RAID processors or Controllers that provide fail-over mechanism.

Fail-over is very vast concept and there are plenty of ways to implement it at different components of the storage networks. for more information you can always buzz me on Email.

Host Bus Adapter [HBA]

* HBA is an important component of Storage networks, especially in SAN architecture.
* HBA is usually host-side component & can be connected to PCI or PCI-X or PCI-E slots on the host or servers. it provides the crucial connectivity between Hosts-StorageSubsystems in Direct attached setup, provides connectivity between Host-FabricSwitches-StorageSubsystems in SAN environment.
* HBA vendors are very few, some vendors provide HBA management software & drivers along with HBA (ex: Qlogic, Emulex, Brocade etc).
* HBAs available in market may be of Fibre Channel (FC), SCSI, SAS and iSCSI interfaces.
* HBA comes in different priceband and the difference in price band is largely driven by the number of ports. HBAs are available in the market with single port, dual(2) Ports, Quad(4) ports, octa(8) ports etc. Depending on the number of Ports & the type of interfaces, the HBA cost varies from 500$ to 5000$ or even more.

Host or Server-side Implimentation in SAN

FC SAN:
* Host or server in SAN architecture consists of HBAs, FC protocol stack, Operating system & Software Applications as shown in the Picture.
* FC Protocol Stack consists of 5 layers (FC0, FC1, FC2, FC3, FC4). each layer corresponds to: physical (FC-0), data link (FC-1, FC-2), network (FC-2, FC-3), and transport (FC-4).
* The SAN storage/disk system is managed by the Host/Server operating system.
Software applications access data via file I/O system calls into the Operating System. The file I/O system calls are handled by the File
System, which manages the directory data structure and mapping from files to disk blocks in an abstract logical disk space.
* The Volume Manager manages the block resources that are located in one or more physical disks in the Disk System and maps the accesses to the logical disk block
space to the physical volume/cylinder/sector address.
* The Fibre Channel protocol stack, ties the Operating System to the Host Bus Adapter(FC HBA) hardware that provides the transport function for block I/O commands to the remote
disk systems across the SAN network.
* The file level I/O initiated by the Software application (SAN management S/W) is mapped into block level I/O by the FC protocol stack & FC interface between the Host/Server and the disk system (Storage Subsystems).
* Fibre Channel is the first network architecture to enable block level storage networking applications.

iSCSI SAN:
* The Host/Server-side implementation is essentially the same as the Fibre Channel SAN. The key difference here is that the Fibre Channel protocol stack & FC Interfaces (FC-HBA, Cables, connectors, Network components) is
replaced by the iSCSI/TCP/IP stack & Ethernet (TCP/IP) Interfaces (iscsi-HBA, Cables, connectors, Network components).
* In FC based SAN, FC protocol stack maps the Block level (I/O) SCSI commands into Fibre Channel Frames & carries SCSI data over FC to remote storage subsystems in SAN envirnoment.
* In iSCSI based SAN, iSCSI/TCP/IP stack maps the Block level (I/O) SCSI commands into TCP/IP Data packets & carries SCSI data over Ethernet to remote storage subsystems in SAN envirnoment.

NOTE: the Picture posted with this topic shows interesting comparison of Host or Server-side Implementation in DAS, SAN (FC & iSCSI) and NAS.