Beruflich Dokumente
Kultur Dokumente
Microcode Level:
Product/Feature:
Involved Components:
Introduction
Product description
EMC Symmetrix VMAX FAST for standard provisioned,
non-thin environments automates the identification of data
volumes for the purposes of allocating or re-allocating
application data across different performance tiers within a
single array. FAST proactively monitors workloads at the
volume (LUN) level in order to identify busy volumes
that would benefit from being moved to higher performing
drives. FAST will also identify less busy volumes that
could be relocated to higher capacity drives, without
affecting
the
existing
performance.
This
promotion/demotion activity is based on policies that
associate a storage group to multiple drive technologies, or
RAID protection schemes, based upon the performance
requirements of the application contained within the
storage group. Data movement executed during this
activity is performed non-disruptively, without affecting
business continuity and data availability.
The primary benefits of FAST include:
Elimination of manually tiering applications when
performance objectives change over time.
June 2010
Feature description
For FAST to operate on a Symmetrix VMAX, there are
three types of components that need to be configured
Storage Groups, FAST Policies, and Symmetrix Tiers.
Storage Groups are a logical collection of Symmetrix
volumes, typically associated with an application, that are
to be managed together.
FAST Policies contain a set of tier usage rules that can be
applied on one or more storage groups.
Symmetrix Tiers are a combination of a drive technology
(EFD, FC, or SATA) and a RAID protection type.
The storage group definitions are shared between FAST
and Auto-provisioning Groups. However, a Symmetrix
device may only belong to one storage group that is under
FAST control.
A FAST policy groups between one and three tiers and
assigns an upper usage limit for each Symmetrix tier. The
Page 1
FAST algorithms
FAST uses three distinct algorithms to determine the
appropriate tier a device should belong to. The algorithms,
in order of probability, are:
Capacity-based algorithm
June 2010
Page 2
Device movement
There are two methods, a swap or move, by which a device
will be relocated to another tier.
A swap occurs when there is no unconfigured space in the
target tier, and results in a corresponding device being
moved out of the target tier. In order to preserve data on
both devices involved in the swap, a single DRV is used.
Packaging
Syntax
Requirements
June 2010
Page 3