Data Storage Systems
To improve control efficiency and minimize expenses the enterprises and organizations strive to optimize existing business processes. The most important asset is information availability and relevance of which are the key factors of success in competitive environment. This situation stimulates large-scale development of data storage technologies.
The state-of-the-art data storage system ensures reliable storage of information resources and access to them. The system includes disc arrays, access infrastructure and data storage control software systems.
Network Data Storage
The infrastructure of access to data storage devices is responsible for information availability for applications and final users that is why this component of data storage systems shows the most dynamic development. Instead of conventional direct connection of disc arrays to servers (DAS - Direct Attached Storage) technologies appeared and are now widely used based on use of network as infrastructure to access the data. SAN (Storage Area Network) technology uses dedicated network based on Fibre Channel protocol and exchanges data on the unit level. NAS (Network Attached Storage) devices are, in fact, the specialized file servers. NAS uses existing IP networks as transport and operates at the file level. The best option is to use SAN and NAS simultaneously, and the data access method will be defined depending on certain requirements.
The main advantages of network storage:
- high scaling. While the data volumes grow the solution scaling becomes of great importance cutting data storage cost;
- high performance. Application of Fibre Channel SAN (Gigabit Ethernet for NAS) ensures high data exchange speed.
- high availability and reliability. The network architecture allows storage system component backup and elimination of single point of failure;
- effective implementation of data migration procedures (backup, data replication, etc.); reduced loads of the main network (SAN) and computing complexes (SAN, NAS);
- long distance of information transmission and possibility to create distributed storage systems;
- possibility to control memory distribution and data storage device consolidation;
- possibility to access heterogeneous environment data.
Nowadays the main topology in construction of SAN is FC_SF – switch fabric allowing extension of addressability up to 16 mln. devices, increase of data accessibility area dimensions and readiness of the whole storage network.
With growth of data volume use of SAN becomes the most economically feasible solution. Today the cost of storage area network introduction differs slightly from cost of direct attached storages. Further high SAN scaling will allow cutting information storage cost.
Open Technologies considers storage area network introduction as one of the most prospect trends of corporate information system development. Our partners include leaders of SAN construction - Brocade Communication Systems, EMC, Hitachi Data Systems, VERITAS and BMC Software.
Data replication ensures implementation of a large number of tasks the most relevant of which are:
- data backup - storage of information relevant copy at the alternative storage devices;
- data separation - possibility of several servers at once to change distributed data set;
- data consolidation - data copying from the remote nodes to central node;
- Follow-the-sun - control of local data copies to ensure information support of the organization.
Data replication can be ensures by integrated software complex facilities but in this case the centralized replication process will be impossible. Moreover, data replication at the application level leads to additional loads of server computing capacities. To eliminate these disadvantages special program replicators or data replication hardware facilities are used.
Irrespective of replication method one of two modes can be used - synchronous or asynchronous. The synchronous replication mode ensures absolute data relevance but requires high performance communication infrastructure, for example, SAN.
One of the efficient means to prevent information losses occurring as a result of failures is mirroring. But the mirrored data cannot eliminate the initial cause of losses - user errors. Data backup copying allows preventing losses both in case of software or hardware failures and in case of operators' errors.
When constructing backup copying and data recovery systems two tasks shall be solved:
- backup copying "gap" reduction (zero downtime);
- reduction of backup data copying traffic and computing resource cost.
The application downtime can be minimized with special methods such as snapshots. Snapshot is a shot of state and record of data location. Using this information the data can be restored in case of loss or damage. Snapshot creation requires application stop but the downtime amounts to seconds.
With application of SAN architecture the processes of backup copying and data recovery can be organized without local area network resources (LAN-free backup) and server resources (server-free backup). Moreover, centralized data backup scheme reduces the number of devices (tape storage, etc.) participating in the process.
In LAN-free backup the data are transported from disc arrays inside the SAN to tape storages connected directly to storage network.
In server-free backup the data are transported from storage devices to tape storages not involving servers directly - the data exchange occurs between storage devices inside the SAN at the data unit level but, if necessary, file-by-file data recovery is possible. The main advantage of server-free backup is release of server computing resources.