VNX FAST Suite Fundamentals

Welcome to the EMC VNX FAST Suite Fundamentals class.
Copyright 1996, 2000, 2001, 2002, 2003, 2004, 2005, 2006, 2007, 2008, 2009, 2010, 2011, 2012, 2013 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 2013 EMC Corporation. All rights reserved. Published in the USA.
Revision Date: September 2013

Revision Number: MR-1WP-VNXFASTFD
Copyright 2013 EMC Corporation. Do not copy - All Rights Reserved.
VNX FAST Suite Fundamentals
This course covers the VNX FAST Suite and its components. Those components VNX FAST
Cache, and VNX FAST VP are introduced, and their features and operation are discussed.
This module introduces the VNX FAST Suite, and also introduces the concept of data
workload skew.
These are the software suites available for the VNX storage systems. These suites each
contain a unique set of solutions to improve efficiency by simplifying and automating many
storage tasks.
This training will focus on the VNX FAST Suite. The VNX FAST Suite is used for optimizing
performance and increasing storage efficiency.
The VNX FAST Suite offers solutions that can significantly enhance a customers business and
technical operations. It does this by providing an additional layer of caching to aid
performance, and by using automated tiering of disks to improve both performance and
efficiency.
The VNX FAST Suite contains two products: VNX FAST Cache, and VNX FAST VP.
VNX FAST Cache uses Flash drives to create an additional tier of cache, thereby improving
access speed to data referenced frequently. This disk-based cache work in conjunction with
SP RAM-based cache.
VNX FAST VP allows Pools to consist of up to 3 disk tiers, with different performance and
capacity characteristics on each tier. FAST VP automatically moves data which is accessed
more frequently, often referred to as hot data, to faster drives. At the same time, data
accessed infrequently, or cold data, is moved to higher capacity, slower drives. Keeping data
on the appropriate tier improves system performance and efficiency.
This course will cover the products offered in the VNX FAST Suite in separate modules, and
describe and compare their functionality and benefits.
The VNX FAST Suite consists of 2 components. Those components, VNX FAST Cache (simply
FAST Cache hereafter) and VNX FAST VP (simply FAST VP hereafter), are shown here.
Each component will be discussed in more detail in subsequent slides.
VNX FAST Cache and VNX FAST VP work together to allow the most heavily used data to be
hosted on Flash drives, while data that is less frequently accessed is kept on lower cost,
slower drives. This process is automatic, and requires no user intervention after initial
configuration. This leads to more efficient use of system resources, lowered management
costs, improved system performance, and increased productivity.
The components of the FAST Suite can be managed from the Unisphere GUI as well as from
the command line, with Navisphere Secure CLI. The CLI sometimes allows more detail to be
reported than the GUI useful for troubleshooting, though less so for day-to-day
management and also allows scripting of commands and procedures.
Unisphere is browser-based, and is therefore supported on a number of host platforms. It
requires a supported browser and a supported version of Java. Users are required to log in,
and their username and password tie them to a specific role, allocated at user creation time,
that allow certain operations to be performed and others to be disallowed. Multiple VNX
systems can be managed from the same instance of Unisphere, and do not all have to be
managed at the same permission level.
Navisphere Secure CLI uses the same authentication mechanism, and role-based
permissions, as Unisphere. Commands are sent to an SP. As with Unisphere, support is
available for a number of host platforms.
Note that, while Unisphere allows management of block-based as well as file-based features
on VNX systems, Navisphere Secure CLI addresses only the block-based features.
Workload skew is a concept that is important to understand it forms the basis for deciding
whether or not products such as FAST Cache or FAST VP would be beneficial in a given
environment. Skew refers to workload asymmetry over the data area - some LUNs, or some
parts of LUNs, are more heavily utilized than others.
Skew charts are plotted by dividing the total data capacity into slices of the same size when
working with portions of a LUN, or by using individual LUNs as the slices when dealing with
the complete system. Determine the percentage of I/O associated with each slice - the
leftmost chart - and plot the cumulative values, with the largest I/O contributor first the
rightmost chart. The example on the slide uses 10 slices for the data. The relative
contribution of each slice is determined as a percentage of the total I/O, and the cumulative
values are plotted by stacking the bars as per the chart on the right. Note the shape of the
curve a fairly rapid increase at first, with more gradual increases later on. In published
documentation this is usually a smooth curve because so many data slices are used, and
because a line chart, rather than a bar chart, is common. A line chart is shown on the next
slide.
This chart illustrates how workload skew is usually represented graphically. The x-axis is
cumulative data, expressed as a percentage, and the y-axis is cumulative I/O, expressed as a
percentage. Note that were dealing with percentages an important point when we need to
specify the skew.
Skew is the ratio of total workload to total data capacity, at the point where the workload
and capacity percentages add up to 100%. In the example shown above, we see that 90% of
the I/O activity (total workload) is performed on 10% of the data (total data capacity). The
sum of those values is 100%: 90% + 10%. This environment will then be described as having a
skew of 90%.
Skew may be calculated at the LUN level, and the results will then show which LUNs are the
most active. Those LUNs would typically be relocated to a higher tier. The most accurate
results are achieved when skew can be calculated at the sub-LUN level - 256 MB slices for
FAST VP, or 64 kB extents for FAST Cache.
VNX FAST Suite Fundamentals 10
Here are some representations of skew values, ranging from heavy skew to no skew. Note
that no skew is a skew of 50%, meaning that I/O is distributed evenly across the data
surface. 50% of the I/O is therefore performed on 50% of the data capacity, and the line
showing this is a straight line from point (0,0) to point (100,100). Workloads that exhibit
heavy skew are ideal candidates for the use of Flash drives, possibly as part of a FAST Cache
or FAST VP implementation.
This module introduced the VNX FAST Suite and workload skew.
This module covers an overview of VNX FAST Cache, introduces FAST Cache operating
principles, and discusses the business use of FAST Cache.
VNX systems use SP cache to speed up both reads and writes. That cache consists of DRAM,
is very expensive, and is limited in size to a few GB. Inserting another tier of cache larger,
cheaper, though slower between SP cache and the disk drives allows faster access to data.
It is worth noting that DRAM access times are typically measured in nanoseconds, Flash
access times in microseconds, and disk access times in milliseconds around 3 orders of
magnitude difference between adjacent tiers.
The arrows show the possible data flow in a system that implements FAST Cache.
FAST Cache, though Flash-based, behaves in many ways like a RAM-based write-back cache,
and the terminology reflects that. FAST Cache is empty when created, and will take a while to
fill; the process of filling the cache is called FAST Cache warming. If the I/O access pattern
changes drastically, old data in FAST Cache may be replaced by new data this process will
also be referred to as FAST Cache warming. Note that the presence of FAST Cache does not
change the operation of SP cache it will be used to satisfy requests where possible. If a
request is satisfied from FAST Cache, a FAST Cache hit is logged. If the request can not be
satisfied from FAST Cache, a FAST Cache miss is logged. Data residing in FAST Cache which is
infrequently accessed can be returned to the original LUN in a write-back operation.
FAST Cache will improve the performance of most applications that demonstrate skew. Two
examples are shown here database environments and virtual desktop infrastructure
environments.
Databases often use small I/Os that are clustered together, and that is a good fit for FAST
Cache. More I/Os can be performed, and those that are performed will complete more
quickly. In the case of VDI (Virtual Desktop Infrastructure) environments, large numbers of
users boot from a common image (or common images) which can be served from FAST
Cache. When desktops are refreshed, data requests can be serviced from FAST Cache in the
same way.
Because both reads and writes are cached, FAST Cache is configured as RAID 1 to protect
data in the event of a Flash drive failure. Note that in some cases (an example is large,
sequential I/O), I/O will bypass FAST Cache and be serviced by the disk drives and/or SP
cache.
When FAST Cache is enabled, eligible LUNs are viewed as collections of 64 kB extents, and
those 64 kB extents are tracked in a list. The position of an extent in the list determines
whether or not it is eligible for promotion into FAST Cache. Active extents that are promoted
may displace less active extents already in FAST Cache; the latter are then written back to
their original LUNs.
The FAST Cache feature consists of a number of components: the driver loaded into the LUN
I/O stack, the policy engine that determines whether or not an extent should be promoted,
the map that tracks the extents in FAST Cache, and the Flash drives themselves. Any I/O
request from a host will be satisfied by SP cache if the data is cache-resident; if it is not, FAST
Cache will be checked. If the data is not in FAST Cache, the request will be satisfied from disk.
Once an extent is promoted, subsequent requests will be satisfied from FAST Cache rather
than disk. If an extent in FAST Cache becomes less active and the space is needed for data
which is more active, the extent will be written back to the original location on the owning
LUN.
FAST Cache can be enabled at the LUN level (for Classic LUNs) or at the Pool level if using
Thick or Thin Pool LUNs, in which case all LUNs in the Pool will be enabled for FAST Cache. It
is not recommended to enable FAST Cache for all LUNs in the system as some workloads do
not experience sufficient benefit from FAST Cache.
The maximum FAST Cache capacity varies by VNX model, but can be up to 4.2 TB. These
capacities also depend upon the drive type (100 GB / 200 GB drives), with 200 GB drives
allowing for larger FAST Cache sizes. 100 GB and 200 GB drives cannot be mixed in a FAST
Cache configuration.
FAST Cache benefits most workloads as well as supporting internal functionality such as
Snapshots, Thin LUNs and compressed LUNs.
FAST Cache configuration is simple: choose the drives to be used, create FAST Cache, and
enable the feature for the LUNs and Pools that can benefit from it.
This module covered VNX FAST Cache, the first component of the VNX FAST Suite. The
benefits, features, and operation of FAST Cache were discussed.
This module covers the features and functionality of VNX FAST VP, and looks at the difference
between FAST VP and FAST Cache.
VNX FAST VP, or Fully Automated Storage Tiering for Virtual Pools, is the second component
of the FAST Suite. FAST VP tracks data in a Pool at a granularity of 256 MB a slice and
ranks slices according to their level of activity and how recently that activity took place. Slices
that are heavily and frequently accessed will be moved to the highest tier of storage,
typically Flash drives, while the data that is accessed least will be moved to lower
performing, but higher capacity storage typically NL-SAS drives. This sub-LUN granularity
makes the process more efficient, and enhances the benefit achieved from the addition of
Flash drives.
The ranking process is automatic, and requires no user intervention. Relocation of slices
occurs according to a schedule which is user-configurable, but which defaults to a daily
relocation. Users can also start a manual relocation if desired.
FAST VP depends for its operation on tiers of disks up to 3 are allowed, and a minimum of 2
are needed for meaningful FAST VP operation. The tiers relate to the disk type in use. Note
that no distinction is made between 10 krpm and 15 krpm SAS disks, and it is therefore
recommended that disk speeds not be mixed in a tier.
NOTE: VNX systems running MCx code use 356 MB slices. Other VNX models, such as the
VNX5700 and VNX7500, use 1 GB slices.
Applications that exhibit skew, and have workloads that are fairly stable over time will
benefit from the addition of FAST VP. EMC has white papers describing several such
environments look on Powerlink or support.emc.com for more details.
OLTP (OnLine Transaction Processing) environments typically exhibit enough skew to make
FAST Cache or FAST VP useful. DSS environments, because of the sequential nature of the
data access, usually do not show much improvement with FAST Cache, though they may with
FAST VP because of the larger working data set.
Virtualized environments will respond to the addition of FAST VP to the storage system in
much the same way as environments with physical hosts. The factor that typically plays the
largest role is the application or applications used in the environment.
File systems, as implemented on VNX File or VNX Unified storage systems, will typically show
a performance improvement when FAST VP is implemented, and the environment will use
storage system resources more efficiently.
FAST VP uses a number of mechanisms to optimize performance and efficiency.

Processes which can be performed are:
Data [workload] optimization across tiers: the most frequently accessed data slices will be
moved up in the tier structure, and the least frequently accessed data will be moved down.
Data [workload] optimization within a tier: data slices will be moved within the tier structure
to achieve an even distribution of the workload across the drives in the tier.
Data balancing within a tier: when a tier is expanded by the addition of drives, data slices will
be moved within the tier structure so as to balance the data across all drives in the tier.
The drives used in a Pool can be configured in many ways supported RAID types are RAID
1/0, RAID 5, and RAID 6. For each of those RAID types, there are recommended
configurations. These recommended configurations balance performance, protection, and
data efficiency. The configurations shown on the slide are those recommended for the
supported RAID types. Note that, though each tier may have a different RAID type, any tier
may have only 1 RAID type associated with it, and that type cannot be changed once
configured.
FAST VP collects statistics about slice activity on an ongoing basis, and analyzes those
statistics each hour. At the time the analysis is performed, the relocation status of the Pool,
as visible from Unisphere or Navisphere Secure CLI, is updated. Relocation occurs once daily
by default, though the user can configure which days and at what time the process will run
by using the scheduler. The user may also force a manual relocation to start at any time for a
specific Pool.
As its name implies, FAST VP is a completely automated feature and implements a set of user
defined policies to ensure it is working to meet the data service levels required for the
business. FAST VP will move data between Flash, SAS and NL-SAS media as it ages and
becomes less active.
FAST VP policies control how FAST VP should apply to individual LUNs in a storage Pool via
the following options:
Start High, then Auto-tier - This is the default setting for all Pool LUNs upon their
creation. FAST VP will initially place LUN data on the highest available tier, and will
thereafter relocate slices of these LUNs based on their activity level.
Auto-tier - Auto-tier will initially place LUN data across all tiers in an optimal
manner, and will then relocate slices of these LUNs based on their activity level.
Some slices may be initially placed on an inappropriate tier, so this policy is not
recommended.
Highest available tier - Highest available tier should be selected for those LUNs
which require the highest level of performance. FAST VP will prioritize slices of a
LUN with highest available tier selected above all other settings.
Lowest available tier - Lowest available tier should be selected for LUNs that are
not performance- or response-time-sensitive. FAST VP will maintain slices of these
LUNs on the lowest storage tier available regardless of activity level.
No data movement This policy may only be selected after a LUN has been
created. FAST VP will not move slices from their current positions once this policy
has been selected.
FAST Cache is a system-wide feature, which can accelerate the performance of Classic LUNs
and Pool LUNs in the system, while FAST VP operation is limited to the individual Pools for
which it is enabled. FAST Cache reacts almost immediately to changes in workload, while
FAST VP relocates slices only when scheduled to do so. FAST Cache granularity, 64 kB, allows
for more efficient use of the configured Flash drives than FAST VP with its 256 MB
granularity.
The two products are complementary, though, and are often used together. Any 64 kB
extents on a FAST VP LUN can be promoted to FAST Cache, though the system will not
promote extents which are already present on Flash drives.
The combined benefit is that FAST Cache will provide immediate performance benefit to any
bursty data while FAST VP will move warmer data upwards (to Flash or SAS drives) and colder
data downward (to SAS or NL-SAS drives) over time. In addition to the performance benefit,
there is also a cost benefit in that FAST Cache with a small number of Flash drives serves the
data that is accessed most frequently, while FAST VP with Flash, SAS and NL-SAS drives can
optimize disk utilization and efficiency as well as providing for longer term data access
patterns.
This module covered VNX FAST VP, the second component of the VNX FAST Suite. The
benefits, features, and operation of FAST VP were discussed.
This course covered the VNX FAST Suite, and its components, VNX FAST Cache and VNX FAST
VP. The benefits and use of these VNX features was introduced, and their business impact
discussed.
This concludes the training.

VNX FAST Suite Fundamentals

Transféré par

Informations du document

Copyright

Formats disponibles

Partager ce document

Partager ou intégrer le document

Options de partage

Avez-vous trouvé ce document utile ?

Ce contenu est-il inapproprié ?

Droits d'auteur :

Formats disponibles

VNX FAST Suite Fundamentals

Transféré par

Droits d'auteur :

Formats disponibles

Welcome to the EMC VNX FAST Suite Fundamentals class.

Revision Date: September 2013

Copyright 2013 EMC Corporation. Do not copy - All Rights Reserved.