FAST VP Step by Step Module 1

Welcome to FAST VP – Step by Step.

Copyright © 2012 EMC Corporation. All rights reserved

Welcome to FAST VP – Step by Step.

Copyright © 1996, 2000, 2001, 2002, 2003, 2004, 2005, 2006, 2007, 2008, 2009, 2010, 2011, 2012 EMC Corporation. All Rights Reserved. EMC believes the information in this publication is accurate as of its publication date. The information is subject to change without notice.

THE INFORMATION IN THIS PUBLICATION IS PROVIDED “AS IS.” EMC CORPORATION MAKES NO REPRESENTATIONS OR WARRANTIES OF ANY KIND WITH RESPECT TO THE INFORMATION IN THIS PUBLICATION, AND SPECIFICALLY DISCLAIMS IMPLIED WARRANTIES OF MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE.

Use, copying, and distribution of any EMC software described in this publication requires an applicable software license.

EMC2, EMC, Data Domain, RSA, EMC Centera, EMC ControlCenter, EMC LifeLine, EMC OnCourse, EMC Proven, EMC Snap, EMC SourceOne, EMC Storage Administrator, Acartus, Access Logix, AdvantEdge, AlphaStor, ApplicationXtender, ArchiveXtender, Atmos, Authentica, Authentic Problems, Automated Resource Manager, AutoStart, AutoSwap, AVALONidm, Avamar, Captiva, Catalog Solution, C-Clip, Celerra, Celerra Replicator, Centera, CenterStage, CentraStar, ClaimPack, ClaimsEditor, CLARiiON, ClientPak, Codebook Correlation Technology, Common Information Model, Configuration Intelligence, Configuresoft, Connectrix, CopyCross, CopyPoint, Dantz, DatabaseXtender, Direct Matrix Architecture, DiskXtender, DiskXtender 2000, Document Sciences, Documentum, elnput, E-Lab, EmailXaminer, EmailXtender, Enginuity, eRoom, Event Explorer, FarPoint, FirstPass, FLARE, FormWare, Geosynchrony, Global File Virtualization, Graphic Visualization, Greenplum, HighRoad, HomeBase, InfoMover, Infoscape, Infra, InputAccel, InputAccel Express, Invista, Ionix, ISIS, Max Retriever, MediaStor, MirrorView, Navisphere, NetWorker, nLayers, OnAlert, OpenScale, PixTools, Powerlink, PowerPath, PowerSnap, QuickScan, Rainfinity, RepliCare, RepliStor, ResourcePak, Retrospect, RSA, the RSA logo, SafeLine, SAN Advisor, SAN Copy, SAN Manager, Smarts, SnapImage, SnapSure, SnapView, SRDF, StorageScope, SupportMate, SymmAPI, SymmEnabler, Symmetrix, Symmetrix DMX, Symmetrix VMAX, TimeFinder, UltraFlex, UltraPoint, UltraScale, Unisphere, VMAX, Vblock, Viewlets, Virtual Matrix, Virtual Matrix Architecture, Virtual Provisioning, VisualSAN, VisualSRM, Voyence, VPLEX, VSAM-Assist, WebXtender, xPression, xPresso, YottaYotta, the EMC logo, and where information lives, are registered trademarks or trademarks of EMC Corporation in the United States and other countries.

All other trademarks used herein are the property of their respective owners.

© Copyright 2012 EMC Corporation. All rights reserved. Published in the China.

1Module 1: FAST VP HOWTO

Copyright © 2012 EMC Corporation. All rights reserved 2Module 1: FAST VP HOWTO

Upon completion of this class, you should be able to:


Upon completion of this class, you should be able to:

•Provide an overview of FAST and FAST VP

•Explain FAST VP elements and terminology

•Describe the algorithms used by FAST VP

•Explain the use of time windows and other Symmetrix parameters to manage FAST VP

•Describe interoperability of FAST VP with other Enginuity software


Upon completion of this lesson, you should be able to:



•Provide an overview of FAST and FAST VP

•List the benefits of FAST and FAST VP


FAST moves application data at the volume (LUN) level. Entire devices are promoted or demoted



between tiers based on overall device performance.

FAST VP adds finer granularities of performance measurement and data movement. The data

from a single Thin Device under FAST control can be spread across multiple tiers. Based on

performance data gathered at the extent level, the FAST controller is free to re-locate individual

sub-extents of a Thin Device.


The chart on the left is a Device Activity Report that shows a small number of active devices that


The chart on the left is a Device Activity Report that shows a small number of active devices that

are candidates for Flash drives and a large number of inactive devices that are candidates for

SATA drives.

Moving the active data to Flash can improve application performance by two-to-eight times,

while moving inactive data to SATA can free up Fibre Channel capacity for other applications and

reduce the cost of storage. A key point here is that a 1 TB SATA drive costs 80 percent less per

megabyte than a 146 GB Fibre Channel drive.


FAST is most effective in an environment where the workload tends to be skewed and the skew


FAST is most effective in an environment where the workload tends to be skewed and the skew

patterns are dynamic.

For example, if devices that are heavily accessed this month may be under-utilized next month,

a suitable FAST policy will continually perform the necessary re-locations to locate the heavily

used devices on faster performing drives.

FAST is not beneficial if the workload is not skewed or if the I/O workload is static, since

workload skews can then be addressed using manual methods such as Symmetrix VLUN

Migration.


FAST and FAST VP automate and optimize storage tiers allowing better utilization of EFDs for


FAST and FAST VP automate and optimize storage tiers allowing better utilization of EFDs for

data in need of high performance and SATA technology for infrequently accessed data.

Combining EFD, Fibre Channel, and low-cost SATA drives provides improved performance in a

tiered storage solution at a lower operating cost than a similarly sized Fibre Channel only array.

FAST and FAST VP use defined policies to non-disruptively relocate Symmetrix devices to the

most beneficial drive technology based upon I/O profile. FAST VP can relocate smaller chunks of

data located on Thin Devices from one tier of storage to another.

FAST supports both FBA and CKD devices. FAST VP supports FBA devices only.

FAST and FAST VP can be managed via the Symmetrix Management Console or SYMCLI.


FAST not only improves performance by leveraging Flash drives, it also enables efficient use of


FAST not only improves performance by leveraging Flash drives, it also enables efficient use of

SATA drives to reduce storage costs.

Here, we see a comparison of a 100 TB configuration of all Fibre Channel compared to a

combination of Flash, Fibre Channel, and SATA. This is an example of a 100 TB configuration,

which is close to the average capacity configured in a Symmetrix VMAX.

Configuring SATA helps offset the cost of the Flash drives (on a drive basis, a 1 TB SATA drive

costs 80 percent less per megabyte than a 146 GB Fibre Channel), reducing total storage system

acquisition costs by 20 percent.

FAST reduces management overhead (assuming one FTE [full-time equivalent] without FAST and

one-half FTE with FAST)) to lower operational costs by 43 percent. This also includes power,

cooling, and maintenance.

Combining both leads to a 27 percent lower, three-year total cost of ownership.


The benefits of FAST are:


The benefits of FAST are:

• The automated movement of hot devices to higher performing drives and less used

drives to lower performing drives in a dynamically changing application environment.

• Achieves better cost / benefit ratios by improving application performance at the same

cost, or providing the same application performance at lower cost. Cost factors include

the price of drives, their energy usage and the effort to manage them.

• Management and operation of FAST is provided by Symmetrix Management Console

(SMC), as well as the Solutions Enabler Command Line Interface (SYMCLI).

EMC Ionix ControlCenter® StorageScope™ and Symmetrix Performance Analyzer can provide

visibility into the use of storage types in FAST environments and the resulting impact on

performance.

With Symmetrix Performance Analyzer, storage administrators can quickly view key

performance indicators such as IOPs and the response time of a storage group before and after

the execution of a FAST change plan to assess the impact on performance.





•Explain FAST VP elements and terminology

•Describe the algorithms used by FAST VP

•Explain the use of time windows and other Symmetrix parameters to manage FAST VP


FAST VP adds finer granularities of performance measurement and data movement. As


FAST VP adds finer granularities of performance measurement and data movement. As

performance data is gathered at the sub-LUN level and extents are moved based on activity, the

data from a single Thin Device under FAST VP control can be spread across multiple tiers. FAST

VP can put less-frequently accessed data on more cost-effective drives and only the most busy

extents on higher performing drives.

This allows the most active 20% of data to be placed on EFD for greater performance while the

80% of less active data can be placed on Fibre Channel and SATA drives at a lower cost.


There are three main elements related to the use of FAST. These are:


There are three main elements related to the use of FAST. These are:

• Symmetrix Tier — A shared resource with common technologies.

• FAST Policy — A policy that manages data placement and movement across Symmetrix

tiers to achieve service levels and for one or more storage groups.

• Storage Group — A logical grouping of devices for common management.


For FAST VP, the storage tier is called a VP tier. When defined, VP tiers contain between one and


For FAST VP, the storage tier is called a VP tier. When defined, VP tiers contain between one and

four thin storage pools. Each Thin Pool must contain Data Devices of the same RAID protection

type, and be configured on the same drive technology.

In the case of Fibre Channel and SATA drives, the rotational speed of the drives must also match.

However, Thin Pools containing Data Devices configured on rotating drives of different sizes and

speeds may be combined in a single VP tier.

The maximum number of tiers that can be defined on a Symmetrix array is 256. If creating a

new tier exceeds this limit, then an existing tier must be deleted before creating the new tier.

Symmetrix tier names cannot exceed 32 characters. Each tier name must be unique. Only

alphanumeric characters, hyphens ( - ), and underscores ( _ ) are allowed, however, the name

cannot start with a hyphen or an underscore. Tier names are case insensitive, in other words,

TierA and tiera are not unique names.


The first tier that is added to a given policy, determines the type of tier that policy will contain.


The first tier that is added to a given policy, determines the type of tier that policy will contain.

For policies that include VP tiers, the upper capacity usage limit for each storage tier is specified

as a percentage of the configured, logical capacity of the associated storage group.

FAST VP supports the association of up to 1,000 storage groups with FAST policies containing

thin storage tiers.

The usage limit for each tier must be between 1 percent and 100 percent. When combined, the

upper usage limit for all thin storage tiers in the policy must total at least 100 percent, but may

be greater than 100 percent.

Creating a policy with a total upper usage limit greater than 100 percent allows flexibility with

the configuration of a storage group. Here, data may be moved between tiers without

necessarily having to move a corresponding amount of other data within the same storage

group.

Multiple FAST policies may re-use the same tier, allowing different usage limits to be applied to

different storage groups for the same tier.

A Symmetrix VMAX storage array will support up to 256 FAST policies. Each FAST policy name

may be up to 32 alpha-numeric characters, hyphens (-), and underscores (_). Policy names are

not case-sensitive.


A storage group is a logical collection of Symmetrix devices that are managed together. Storage


A storage group is a logical collection of Symmetrix devices that are managed together. 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.

Storage groups are associated with a FAST policy that defines the maximum percentage of

devices in the storage group that can exist in a particular tier.

FAST and FAST VP support the movement of certain device types within the Symmetrix. A

storage group created for the purposes of FAST or FAST VP may not contain the following device

types: Thin (TDEV), VDEV, DLDEV, CKD, AS400, ICOS, Metadevice members, SAVEDEV, Data

Devices (TDATs), DRV, SFS and Vault.

A Symmetrix VMAX storage array will support up to 8192 storage groups associated with FAST

policies. Storage groups may contain up to 4096 devices.


A policy associates a storage group with up to three tiers. When aggregated, the percentage of


A policy associates a storage group with up to three tiers. When aggregated, the percentage of

storage specified for each tier in the policy must total at least 100 percent.

The same FAST policy may be applied to multiple storage groups; however, a storage group may

only be associated with one policy.

When a storage group is associated with a FAST policy, a priority value must be assigned to the

storage group. This priority value can be between 1 and 3, with 1 being the highest priority—the

default is 2.

When multiple storage groups share the same policy, the priority value is used when the devices

contained in the storage groups are competing for the same resources in one of the associated

tiers. Storage groups with a higher priority will be given preference when deciding which devices

need to be re-located to another tier.


Assume that a policy is configured with three tiers and that the three tiers have a maximum


Assume that a policy is configured with three tiers and that the three tiers have a maximum

allocation of i%, j% and k% respectively. When the sum of the tier allocations (i% + j% +k%) is

100%, there can be no flexibility of storage allocations between the different storage groups.

It also provides an easy algorithm for chargeback.

A 300% sum of all Tier Allocations means that at any given time each associated storage group

can be entirely on any tier. FAST VP will place the allocations completely based on performance

criteria.

In this case, all storage of the tiers is shared dynamically by all storage groups associated with

any FAST policy.

Since users will mostly be concerned with usage of premium tiers, the greatest practical

flexibility will be achieved by configuring the least performing tier with a Tier Allocation of 100%

and limit the Tier Allocation of the premium tiers.

As their needs change over time, this will enable sharing of the limited, expensive storage by

multiple applications.


The example shows two production and one development storage groups. The two production


The example shows two production and one development storage groups. The two production

storage groups have a tier allocation that aggregates to 100%. This means the the percentage of

storage associated with each tier is fixed for these storage groups.

In contrast to the production application storage groups, the development storage group can be

moved around more easily, since up to 100% of the storage group can be located on the SATA

pool if the I/O activity is low. Later, when I/O demand picks up, the more active allocation

extents can be moved to the FC pool.


Initial Analysis Period:


Initial Analysis Period:

The Initial Analysis Period defines the minimum amount of time a Thin Device should be under

FAST VP management before any performance related data movements should be applied. This

period only accounts for time passed while the performance time window is open.

The initial analysis period can be configured to be between 2 hours and 4 weeks, however, it

cannot exceed that of the workload analysis period. The default is 8 hours.

Workload Analysis Period:

The Workload Analysis Period determines the degree to which FAST VP metrics are influenced

by recent host activity, and also less recent host activity, that takes place while the performance

time window is considered open.

The longer the time defined in the workload analysis period, the greater the amount of weight

assigned to less recent host activity.

The workload analysis period can be configured to be between 2 hours and 4 weeks. The default

is 1 week (7 days).

These FAST VP performance metrics provide a measure of activity that assigns greater weight to

more recent I/O requests, but are also influenced by less recent activity. By default, based on a

Workload Analysis Period of 24 hours, an I/O that has just been received is weighted two times

more heavily than an I/O received 24 hours previously.


The performance time windows are used to identify the business cycle for the Symmetrix array.


The performance time windows are used to identify the business cycle for the Symmetrix array.

They specify date and time ranges (past or future) when samples will be included in, or excluded

from, the FAST performance data analysis.

The intent of defining performance time windows, is to distinguish periods of time when the

Symmetrix is idle from periods when the Symmetrix is active, and to only collect performance

data during the active periods.

Performance windows are common for Optimizer, FAST, and FAST VP.

A default performance time window collects all performance data samples, 24 hours a day, 7

days a week, 365 days a year.


Device movement time windows are used to specify date and time ranges when moves are


Device movement time windows are used to specify date and time ranges when moves are

allowed, or not allowed, to be performed.

In order to prevent FAST VP from changing the current tiering allocation of a Thin Device, a

feature called device pinning may be used. Pinning a device will lock all current extent

allocations for the device in their current locations, and will prevent FAST VP from relocating

them.

Any new allocations performed for a pinned device, will come from the Thin Pool the device is

bound to. These allocations will also not be moved by FAST VP.


In the example below – An exclusive performance window has been added to exclude performance data collections for a specific weekend. We will covert the specifics of setting up


In the example below – An exclusive performance window has been added to exclude performance data collections for a specific weekend. We will covert the specifics of setting up time windows in the next module.

sun200 /VMAXe symtw list -sid 95 -type perf

Symmetrix ID: 000195900495

Performance Time Windows

Sunday : 00:00 - 24:00

Monday : 00:00 - 24:00

Tuesday : 00:00 - 24:00

Wednesday : 00:00 - 24:00

Thursday : 00:00 - 24:00

Friday : 00:00 - 24:00

Saturday : 00:00 - 24:00

Exclusive Time Windows (1)

{

Fri Dec 02 22:00:00 2011 - Mon Dec 05 06:00:00 2011

}


The Data Movement Mode for FAST VP can be either Off (no movement) or Automatic.


The Data Movement Mode for FAST VP can be either Off (no movement) or Automatic.

The Pool Reserved Capacity (PRC) reserves a percentage of each pool included in a VP tier for

non-FAST VP activities. The purpose of this, is to ensure that FAST VP data movements do not fill

a Thin Pool, and subsequently cause a new extent allocation, a result of a host write, to fail.

When the percentage of unallocated space in a Thin Pool is equal to the PRC, FAST VP will no

longer perform data movements into that pool. However, data movements may continue to

occur out of the pool to other pools.

When the percentage of unallocated space becomes greater than the PRC, FAST VP can begin

performing data movements into that pool again.

The PRC can be set on a per pool basis or system-wide.

The PRC can be configured to be between 1% and 80%. The default is 10%.

The Relocation Rate is a quality of service (QoS) setting for FAST VP and affects the

aggressiveness of data movement requests generated by FAST VP. This aggressiveness is

measured as the amount of data that will be requested to be moved at any given time, and the

priority given to moving the data between pools.

The relocation rate can be configured to be between 1 and 10, with 1 being the most

aggressive. The default is 5.


The SYMCLI listing of the FAST control parameters highlights those parameters that are shared


The SYMCLI listing of the FAST control parameters highlights those parameters that are shared

between FAST DP, FAST VP, and optimizer as well as those that are unique to FAST DP and FAST

VP.

Note that only the FAST VP control parameters are relevant for VMAXe arrays.

On VMAXe arrays all the FAST DP parameters show a value of N/A.


The FAST & FAST VP control parameters can be set via SMC as well. This is where we check FAST


The FAST & FAST VP control parameters can be set via SMC as well. This is where we check FAST

and Optimizer control parameters. The first half of the screen refers to parameters that are

shared between FAST, FAST VP, and Optimizer.

The second half contains parameters that are specific to FAST VP.

Please note that on VMAXe Arrays, SMC will only show the settings relevant FAST VP. VMAXe

arrays only support virtual provisioning.


There are two components of FAST VP – Symmetrix microcode and the FAST controller.


There are two components of FAST VP – Symmetrix microcode and the FAST controller.

The Symmetrix microcode is a part of the Enginuity storage operating environment that controls

components within the array. The FAST controller is a service that runs on the service processor.

When FAST VP is active, both components participate in the execution of two algorithms – the

intelligent tiering algorithm and the allocation compliance algorithm – to determine appropriate

data placement.

The intelligent tiering algorithm uses performance data collected by the microcode, as well as

supporting calculations performed by the FAST controller, to issue data movement requests to

the VLUN VP data movement engine.

The allocation compliance algorithm enforces the upper limits of storage capacity that can be

used in each tier by a given storage group by also issuing data movement requests to the VLUN

VP data movement engine.

Data movements performed by the microcode are achieved by moving allocated extents

between tiers. The size of data movement can be as small as 768 KB, representing a single

allocated Thin Device extent, but will more typically be an entire extent group, which is 7,680 KB

in size.


The read miss metric accounts for each DA read operation that is performed. Reads to areas of a


The read miss metric accounts for each DA read operation that is performed. Reads to areas of a

Thin Device that have not had space allocated in a Thin Pool are not counted. Also, read hits,

which are serviced from cache, are not considered.

Write operations are counted in terms of the number of distinct DA operations that are

performed. The metric accounts for when a write is de-staged – write hits to cache are not

considered.

Writes related to specific RAID protection schemes will also not be counted. In the case of RAID

1 protected devices, the write I/O is only counted for one of the mirrors. In the case of RAID 5

and RAID 6 protected devices, parity writes are not counted.

Pre-fetch operations are accounted for in terms of the number of distinct DA operations

performed to pre-fetch data spanning a FAST VP extent. This metric considers each DA read

operation performed as a pre-fetch operation.

Workload related to internal copy operations, such as drive rebuilds, clone operations, VLUN

migrations, or even FAST VP data movements, is not included in the FAST VP metrics.

The intelligent tiering algorithm considers the performance metrics of all Thin Devices under

FAST VP control, and determines the appropriate tier for each extent group.

The allocation compliance algorithm is used to enforce the per-tier storage capacity usage

limits.


The intelligent tiering algorithm is structured into two components; a main component that


The intelligent tiering algorithm is structured into two components; a main component that

executes within Symmetrix microcode and a secondary, supporting, component that executes

within the FAST controller on the service processor.

The main component assesses whether extent groups need to be moved in order to optimize

the use of the FAST VP storage tiers. If so, the required data movement requests are issued to

the VLUN VP data movement engine.

When determining the appropriate tier for each extent group, the main component makes use

of both the FAST VP metrics, previously discussed, and supporting calculations performed by the

secondary component on the service processor.

The intelligent tiering algorithm runs during open data movement windows, when FAST is

enabled and the FAST VP operating mode is Automatic.


The goal of the allocation compliance algorithm is to detect and correct situations where the


The goal of the allocation compliance algorithm is to detect and correct situations where the

allocated capacity for a particular storage group within a thin storage tier exceeds the maximum

capacity allowed by the associated FAST policy.

A storage group is considered to be in compliance with its associated FAST policy when the

configured capacity of the Thin Devices in the storage group is located on tiers defined in the

policy and when the usage of each tier is within the upper limits of the tier usage limits

specified in the policy.

When a compliance violation exists, the algorithm will generate a data movement request to

return the allocations within the required limits. This request will explicitly indicate which Thin

Device extents should be moved, and the specific Thin Pools they should be moved to.

The size of the data movement request depends on the amount of capacity that is currently out

of compliance, but also on the user-defined relocation rate. The maximum size of request that

can be generated is 10 GB worth of data movements.

When the relocation rate is set to anything other than 1, the FAST controller divides 10 GB by

the relocation rate to determine the new maximum. For example, if the relocation rate is set to

2, the maximum request size will be 5 GB; if it is 10, the maximum size will be 1 GB.

The compliance algorithm runs every 10 minutes during open data movement windows, when

FAST is enabled and the FAST VP operating mode is Automatic.


Data movements executed by FAST VP are performed by the VLUN VP data movement engine,


Data movements executed by FAST VP are performed by the VLUN VP data movement engine,

and involve moving Thin Device extents between Thin Pools within the array.

Extents are moved via a move process only; extents are not swapped between pools.

The movement of extents, or extent groups, does not change the Thin Device binding

information. That is, the Thin Device will still remain bound to the pool it was originally bound

to. New allocations for the Thin device, as the result of host writes, will continue to come from

the bound pool.

Only extents that are allocated will be moved. No back-end configuration changes are

performed during a FAST VP data movement, and no configuration locks are held during the

process.

As swaps are not performed, there is no requirement for any swap space, such as DRVs, to

facilitate data movement.

In order to prevent FAST VP from changing the current tiering allocation of a Thin Device, a

feature called device pinning may be used. Pinning a device will lock all current extent

allocations for the device in their current locations, and will prevent FAST VP from relocating

them.





•Compare FAST (DP) and FAST VP

•Discuss FAST Controller states and activities

•Describe interoperability of FAST and FAST VP with other Enginuity software





between tiers based on overall device performance.

FAST VP adds finer granularities of performance measurement and data movement. The data

from a single Thin Device under FAST control can be spread across multiple tiers. The FAST

controller is free to relocate individual sub-extents of a Thin Device, based on performance data

gathered at the extent level.

Note: Only FAST VP is relevant for VMAXe arrays.


There are five possible states that the FAST controller can be reported in. These are:


There are five possible states that the FAST controller can be reported in. These are:

1. Enabled — The FAST controller will perform all of its functions: performance data collection,

performance data analysis, configuration change plan generation, and configuration change plan

execution.

2. Disabled — The FAST controller will only perform one of its functions: performance data

collection. Data analysis and configuration change plan generation or execution will not be

performed.

3. Disabling — The FAST controller is transitioning from Enabled to Disabled.

4. DisabledwithError — The FAST controller has stopped operation due to an internal error.

None of the FAST Controller operations will be performed.

5. Degraded — The FAST controller can perform some or all of its functions. However, it cannot

perform each function fully.

For more details on each of the states consult the FAST VP Technical note.


When the state of the FAST controller is queried, and the state is Enabled, the current activity


When the state of the FAST controller is queried, and the state is Enabled, the current activity

being performed by the controller will also be displayed. Valid activities include:

• Idle — The FAST controller is currently idle.

• FetchingStats — The FAST controller is collecting Symmetrix device performance

statistics.

• AnalyzingStats — The FAST controller is performing analysis on the collected device

performance statistics.

• PendingPlan — A configuration change plan has been approved and its execution

scheduled.

• RunningPlan — A configuration change plan is currently being executed.


FAST and FAST VP operates alongside Symmetrix features such as Symmetrix Optimizer, Dynamic


FAST and FAST VP operates alongside Symmetrix features such as Symmetrix Optimizer, Dynamic

Cache Partitioning, and Auto-provisioning groups.

While both FAST and Symmetrix Optimizer can be operated independently of each other on a

Symmetrix VMAX array, they do share several configuration settings between them. Refer to

the page entitled: “Control Parameters related to FAST and FAST VP” to identify the shared and

unique parameters.


FAST and FAST VP are fully interoperable with all Symmetrix replication technologies—SRDF,


FAST and FAST VP are fully interoperable with all Symmetrix replication technologies—SRDF,

TimeFinder/Clone, TimeFinder/Snap, and Open Replicator.

Any active replication on a Symmetrix device remains intact while the device is being moved or

swapped.

Similarly, all incremental relationships are maintained for the moved or swapped devices.

However, what must be kept in mind, is that FAST device movements will consume array

resources. This means that planning must be performed in determining when FAST device

movements should occur to minimize impact on other replication processes.

Remember, that all relocations performed by FAST DP occur on standard devices. All relocations

performed by FAST VP occur on Thin Device extents.

The source device of the TimeFinder/Clone session, or Clone emulation session, can be moved

or swapped by FAST and FAST VP. A device that is the target device of a TimeFinder/Clone

session, or a Clone emulation session, can also be moved or swapped by FAST or FAST VP,

provided that the target has been split or activated.

The source device in a TimeFinder/Snap session can be moved or swapped by FAST or FAST VP.

The control device in an Open Replicator session, push or pull, can be moved or swapped by

FAST or FAST VP.


An RDF1 volume, with local protection, can be moved or swapped by FAST while also actively


An RDF1 volume, with local protection, can be moved or swapped by FAST while also actively

replicating to an RDF2 volume, in either synchronous or asynchronous mode.

Similarly, an RDF2 volume can be moved or swapped while being replicated to by an RDF1

volume.

Thin SRDF devices, R1 or R2, can be associated with a FAST policy. Extents of SRDF devices can

be moved between tiers while the devices are being actively replicated, in either synchronous or

asynchronous mode.

While there are no restrictions on the ability to move or swap SRDF devices with FAST and FAST

VP, device movements are restricted to the array upon which the FAST Controller is operating. If

an RDF1 device is moved between two tiers, FAST or FAST VP will not automatically perform a

corresponding move of the respective RDF2 device on a remote array. This means that, in a

SRDF failover scenario, the remote Symmetrix array will have different performance

characteristics than the failed local production array.

Based on FAST DP movements, while it is possible to move R2 devices manually on the R1 side,

it is not possible to move Thin Device extents away from the pool that the Thin Device is bound

to using VLUN migration. So, it is not possible to replicate the changes made to the R1 side on

the R2 side when Thin Pools are being managed by FAST VP.


Storage groups created for the purposes of FAST or FAST VP may also be used for Auto-


Storage groups created for the purposes of FAST or FAST VP may also be used for Auto-

provisioning, and the other way round. However, while a device may be contained in multiple

storage groups for the purposes of Auto-provisioning, it may only be contained in one storage

group that is associated with a FAST or FAST VP policy.

While Auto-provisioning storage groups can share devices, FAST or FAST VP managed storage

groups cannot share devices. As a result of this, it may not be possible to use already configured

storage groups for the purposes of FAST or FAST VP.


Key points covered in this class:


Key points covered in this class:

•Overview of FAST and FAST VP

•FAST and FAST VP elements and terminology

•Algorithms used by FAST and FAST VP

•Time windows and other Symmetrix parameters to manage FAST and FAST VP

•Interoperability of FAST and FAST VP with other Enginuity software


Please feel free to contact me if you have any questions.


Please feel free to contact me if you have any questions.

Kevin Wang ([email protected])


Documents

FAST VP Step by Step Module 1