【 Weak current College 】 data archive disk storage selection

Most enterprise database data flow in growth, but the lack of control, even cannot control, money, power and data center space are limited resources, the database on the utilization of these resources must be increased.

Supplier of many ways to improve the efficiency of data archiving solutions. Data reduction techniques such as compression, single instance storage (singleinstancestorage) and most powerful sub-file de-duplication technology, this enables suppliers to 30lbs data loading 10lb data archiving. When you are not using Spinningdisk, shut it down, regardless of vendor called MAID (Copan, Nexsan) or HDS, its the 1TB drive energy consumption from 7.5W reduced to less than 1W.

But even if these drives are stopped, the MAID system still consumes is greater than or equal to its normal run-time 30% more power, because in the low-load, low-power supply efficiency. The processor is in idle state still has the power consumption, but idling memory and an up-to-date level of idling server processor EnergyStar (consumption electricity to 55W), or not.

Data archiving storage of the simplest system, I call it locked NAS (lockedNAS). Vendors including NetApp and Sun company has expanded its file system and NAS operating system will remain in the implementation of the guarantor force add in. They are modelled on the NetApp SnapLock naming patterns, the latest improvement technology of date to the end of life equipment, playing on the "read-only" flag. Such a system would be in the life of the end-to-read-only portion of the reset.

LockedNAS general system, but lack the last time I mentioned a few other features.

While NetApp file compile stalls Member verification each have many cluttered information integrity of the data block, it does not operate in the background, in case any problems are found or remotely obtain a corrupt local data block of the full version. Scalability and long-term extension is a problem, because increasingthe drive as well as every 5-7 years data transfer (because your enterprise vendors maybe will no longer support your business system) and is not a good way to solve the problem.

Disk-based data storage archiving this market sector, another competitor is ContentAddressableStorage, using each storage destination (file, email information, etc.) of the hash as a storage target (instead of the file position) of important identifier — with NAS system. And usually thinking different, CAS systems do not use the full-text index as its location, but only a hash of the selected storage object. In fact, most CAS systems, including EMC Centera, Nexsan Corporation Assurion and Caringo CAStor, the company did not give its storage content indexing.

Their internal implementation of single instance storage (for example copy the same file multiple times will cause the same redundant data), or check the filehash, to consolidate data in the background. In addition to most of the file system supports name, owner, and time stamp, most CAS system can store extended metadata. As a result, the most complex application programming interface is used for file storage and recovery, data archiving software vendors to write and test the interface. SNIA standard XMLAPI--known as XAM, a year after the first appear in the CAS, and other fixed-content storage system.

CAS suppliers attach great importance to extend metadata. ILM (currently just a concept, not products) for data classification features, functionality, and similar e-discovery capabilities are required in addition to the name and date of other data to make decisions. I have doubts that need to store a special file system API. Data archiving software or a content management system is only able to add metadata and all important full text index into independent of the file system of the database.

Some vendors have built similar NAS device, using the hash to ensure that the integrity and identify the unique identity of the data, not as the main address of the target data. Like DataDomain company equipment and NEC's Hydrastor, have all been as a backup target, but its features, and data archiving device level. DataDomain unit for data preservation and removal, and the functionality the same. Permabit company EnterpriseArchive adopt similar hash to assist in the management of NAS data.

Many such systems with RAIN (redundant array of independent nodes) structure, so a 2U server into a 1u and built-in storage groups or the distribution grid support and management of data in the array. Some systems use absorption/fix nodes, these nodes can manage the hash data to and receive data while storage support data nodes. Some systems also have both features.

If you were able to complete implementation, RAIN patterns will make 100 nodes for scalability, new nodes are available to scalability, processor speed, disk capacity, and will increase a array, once placed in the old slow node or there is a problem with the data on the nodes will be resettled, then the old node will be replaced by a small number of hits or command. However, most of the RAIN system has an associated high speed processor, this can lead to an increase in power consumption, while at the same time might cause access times fewer large data archiving to consume too much.