How to improve the efficiency of enterprise data archiving solutions

How to improve the efficiency of enterprise data archiving solutions2011-02-01 09:27:47 Source: IBM Share | Summary: Talk about all storage features, in addition to archiving the data storage System. Now let's look at how vendors present these features as solutions to IT managers and place them in the data requirements list.

Keywords: Data archive storage

Talking about all the storage features, in addition to archiving the data storage System. Now let's look at how vendors present these features as solutions to IT managers and place them in the data requirements list.

One key difference between systems that we didn't talk about last time is efficiency. The data traffic in most enterprise databases is growing, but it lacks control or even control, and the availability of funds, electricity and data center space is limited, so the utilization of these resources by the database must be increased.

Vendors use a variety of ways to improve the efficiency of their data archiving solutions. Data reduction technologies such as compression, singleton storage (single instance storage), and most powerful sub-file de-duplication technologies allow vendors to load lbs data into a LB data archive. When spinning disk is not used, it is turned off, regardless of whether the vendor calls it Maid (Copan, Nexsan) or HDS, which can reduce the energy consumption of 1TB drives from 7.5W to less than 1 W.

But even if these drives stop running, the maid system will still consume more than 30% of its normal operating time, because the power supply becomes less efficient when the load is low. The processor still has power consumption while idling, and idling memory is no better than a idling server processor (55W) that uses the latest level of EnergyStar.

The simplest system of data archive storage, which I call the locked NAS (locked NAS). Vendors, including NetApp and Sun, have expanded their file systems and NAS operating systems to add strength to their execution. They modeled on NetApp's naming of Snaplock, adding the date of the latest improvements to the end of the device's life, marking the "read Only" banner. This system will reset the read-only portion at the end of the life cycle.

Locked NAS is a general system, but lacks some of the other features I mentioned last time.

The NetApp file-teller checks the integrity of each block of data that is cluttered with clutter, not running in the background and unable to obtain a full version of a corrupted local block of data when a problem is discovered. Scalability and long-term scalability are the problems because adding drives and data transfers every 5-7 years (because your enterprise's suppliers may no longer support your enterprise system) is not a good way to solve the problem.

disk-based Data storage Archiving This market area, another competitor is content addressable Storage, using the hash of each storage target (file, email information, etc.) as an important identifier for the storage target (rather than the file location)-with the NAS system. Unlike the usual thinking, the CAS system does not use a complete text index as its location scheme, but only uses the hash of the storage target. In fact, most CAS systems, including EMC's Centera, Nexsan Company's assurion, and Caringo's Castor, have not indexed their storage content.

They implement a single instance store within the enterprise (for example, multiple copies of the same file will result in the same redundant data), and also check the file hash to consolidate background data. In addition to the names, owners, and timestamps supported by most file systems, most CAS systems can also store extended meta data. As a result, most of the complex application programming interfaces are used for file storage and recovery, which requires the data archiving software vendor to write and test the interface. SNIa has a standard XML api--called Xam that will first appear on CAs and other fixed content storage systems after a year.

CAS vendors attach great importance to extending meta data. The data classification, E-discovery, and similar features of ILM (currently just a concept, not yet a product) require other data, other than the name and date, to make decisions. What I have in doubt is the need to store a special file system with APIs. Data archiving software or content management systems can only integrate metadata and all important text indexes into a database that is independent of the file system.

Some vendors have built NAS-like devices that use hashing to ensure integrity and uniquely identify data, rather than being the primary address of the target data. Devices similar to data domain and NEC Hydrastor are used as backup targets, but from their very nature, they are about the same level as those of the archiving device. Data domain devices can be saved and removed by the same functionality as CAs. The Permabit company's Enterprise archive uses a similar hash to help manage NAS data.

Many of these systems employ rain (redundant array of independent nodes) structures, such that a group or grid with 1u to 2u servers and built-in storage supports and manages the data distributed in the array. Some systems use absorption/repair nodes, which manage hash data and receive data while storing supporting data nodes. Some systems have both of these functions at the same time.

If fully implemented, rain mode will allow 100 nodes to be measurable, new nodes can also be measurable, the processor faster, the disk capacity is larger, and all will add an array, once placed in the old slow node or the problem of the node data will be relocated, The old nodes are then replaced by a small number of clicks or commands. However, most rain systems have a related high speed processor, which can lead to increased power consumption and may lead to excessive consumption of large data archives with very little access.

(Zebian: Chuiwen)

Statement: Where the CIO Time Network (www.ciotimes.com) original works (text, pictures, charts), reproduced please be sure to indicate the source, violators of the network will be responsible for the law.