Thin Provisioning

Thin Provisioning (TP ) ( German " thin provisioning " ), refers to a cost -saving method of providing storage in virtualized storage environments ( storage virtualization ).

Classic memory allocation

When administrators take a memory system for the first time in operation, they must

  • The disk for one or more operating systems partition,
  • Logical drives ( C :) about Creating
  • And then format it with the current operating system tools.

In this classical method, which is also called hardware provisioning, Fat Provisioning or Exact Provisioning, users and application programs is allocated storage capacity. Another term that dedicate -on -allocation ( in the assignment book ), describes the actual operation.

Thus defined partitions and logical disks can not be changed with operating system functions. Should the allocated memory is full (which is usually already at 80 percent efficiency of the case, as is required for example in the reorganization of databases, additional space ), all data must first be copied to another medium and then deleted the LUN and recreated. With special software tools experienced users can adjust the size of a LUN, including all data but also during operation. On a backup should not be omitted for security reasons.

While it is not a problem for home users to reserve the entire storage capacity of a hard drive for their needs, storage administrators have to deal in larger establishments prudent with resources. To notify users to only the expected storage capacity requirement over the life of a project. To avoid changes to these rules and thus costly administrative work, the allocated capacities are dimensioned for safety's sake generous. Part of the data growth is attributable to that fixed reservation. The actual amount of data stored is often only a fraction of the allocated capacity.

Modern Memory Allocation ( Dedication -on-write )

Thin provisioning can be compared to the gestelltem of a power plant available power. This guarantees the customer a maximum power that this can relate. Nevertheless, the e- business can provide only part of the maximum power to all customers at the same time. But never all at the same time obtain the maximum power, there are no problems.

In the IT sector, the term more commonly known as overprovisioning ( overcommitment ) or of quotation is called.

The thin provisioning makes use of the fact that modern storage systems provide virtual hard disks. The collector ( server) is in this case more capacity made ​​available, as is maintained in the storage system for it. Once used by the server capacity exceeds a certain threshold, spare capacity is also provided for the purchaser of an existing storage pool.

There are basically two different ways of delivering drive capacity:

Block-based storage provisioning or file-based storage provisioning. The SNIA has developed the Shared Storage Model a description of these different methods. The difference lies in the type of access.

In the block-based storage system, the hard disk is divided into logical blocks - these blocks are as Logical Unit (LUN) provided. This may have different protocols such as Fibre Channel or iSCSI done. The server ( consumer) uses this LUN as a rule on the LUN, a file system, such as NTFS, ext3, VxFS to write. Thin provisioning is implemented in this case by the storage system. The server sees thereby a LUN that is x GB. In fact, the storage system has however reserved xn GB. If the server need more capacity than xn, the LUN is automatically enlarged from a free pool. However, in contrast, the LUN is expanded again with each additional letter of the server as a file system - regardless of whether the server has since deleted data on the LUN based on the file system. This is due to that the server has no means of communication with the storage system, which allows him to communicate to the storage system, that certain blocks can be released. Some storage vendors offer special software with which a reorganization can be done - but this is only possible if during the reorganization no rewriting occurs. Some storage systems also evaluate the known SSD TRIM command with which the file system driver can tell the underlying layers, which blocks can be considered as "free".

When file-based storage system, the hard drive is provided by a server or filer as a file system using protocols such as NFS or CIFS available. The providing server uses for its part, is itself block-based LUN. Here the provisioning on the filer. Almost all manufacturers of file-based storage systems offer such functionality. In contrast to block-based thin provisioning of the filer 's own master of the mounted file system from him. When data is deleted, it can automatically release the capacity again. As such, the effectiveness of thin provisioning is much higher here than in the block-based methods.

The advantage of the method is obvious: there is less storage needed - which has a direct impact on price, footprint, demand for air conditioning, power consumption, etc.. The risk is that a very detailed monitoring must be implemented. As with the power station must be provided if necessary new capacity quickly. The E-Werk it reaches about about purchasing abroad at overload - in the storage area physical disks must be purchased. Increases the utilization can be more provided as new capacity, this leads to the application hangs. In this respect, not only monitoring, but also a trend line analysis is an important prerequisite for the use of thin provisioning.

Footnotes

  • Computer Architecture
715021
de