There is nothing more frustrating than trying to learn a new technology and not being able to keep up with basic instruction. This is partially representative of my own journey into storage, as historically I’ve focused on servers, virtualization, and operating systems. Storage administration is a peculiar segment of IT, and there are a number of terms and acronyms that go along with it. Here are some of the terms that I have taken in during my journey:
- Thin provisioning:
- Hot spots:
- Short stroking:
- Zero Page Reclaim or (ZPR):
- Wide striping:
- Cheap and deep:
- Rotational storage:
- Storage tiering:
- Slow-Spin or No-Spin:
Disks are expensive, and many administrators make the case to provision volumes at any level as a thin provisioned solution. This effectively doesn’t count free space as used even though it is allocated. See my previous tip providing an overview of thin provisioning.
A hot spot is an area of disk that receives a large amount of I/O from the storage consumers. This can be due to a number of logical unit numbers (LUNs) residing on a single disk or simply a very busy single LUN. The objective of a storage administrator is to mitigate the hot spots across the available storage.
This is somewhat of a trick play in storage administration in that a drive can be made to perform better than a typical configuration. A short stroke is only using a very small area of a hard drive as part of a large array to make the hard drive have a short area to traverse. While this will increase the performance of the drive, there is usually a lot of wasted space.
For thin provisioning environments, this is where the storage controller goes and harvests zeroes and returns them to a master available storage pool. Depending on the architecture or use of storage products, this may be done exclusively by the storage processor or as an enhancement to new products. One example is VMware’s vSphere 4.1 which now offers support for vSphere APIs for Array Integration (VAAI) with selected storage products. This effectively lets the storage processor do the zero-handling or bulk zero I/O operations instead of the vSphere host as an optimization for disk use and storage network traffic.
This practice involves having a higher number of drives in use for a LUN to achieve greater throughput. Basically, 12 drives in an array providing a LUN will provide better aggregate throughput than two or three of those same drives.
This is typically used to describe storage that is very large in terms of GB or TB, yet is slow and inexpensive. In today’s data centers, that gravitates toward SATA drives that are 1 TB, 2 TB, or larger.
This term refers to traditional hard drives with moving platters and a seeking head to the regions on the disk. This is also known on a more casual level as “spinning rust”. The alternative is solid state (or enterprise flash) storage, which has no moving parts.
Whether automated or storage administrator-based, this is putting workloads on disk resources that perform to their requirements. In most environments, there are two levels of storage tiers: SAS and SATA drives with the SAS drives being higher performance (and price) drives. See my tip on why storage tiering is important.
Denotes a slow tier of storage that is very slow, such as 5400 or 7200 RPM; or even in a powered down state. This can include a tape storage solution.
Storage administration is filled with a number of terms that denote how disk resources are managed, provisioned and consumed. Do you have some storage jargon that you use in your daily administration? Share your terms below.
Rick Vanover is a software strategy specialist for Veeam Software, based in Columbus, Ohio. Rick has years of IT experience and focuses on virtualization, Windows-based server administration, and system hardware.