Emc VNX Fast VP: VNX5200, VNX5400, VNX5600, VNX5800, VNX7600, & VNX8000 A Detailed Review
Emc VNX Fast VP: VNX5200, VNX5400, VNX5600, VNX5800, VNX7600, & VNX8000 A Detailed Review
Abstract
This white paper discusses EMC Fully Automated Storage
Tiering for Virtual Pools (FAST VP) technology and describes
its features and implementation. Details on how to use the
product in Unisphere are discussed, and usage guidance and
major customer benefits are also included.
November, 2013
EMC VNX FAST VP
VNX5200, VNX5400, VNX5600, VNX5800, VNX7600, & VNX8000
A Detailed Review
2 EMC VNX FAST VP A Detailed Review
VNX5200, VNX5400, VNX5600, VNX5800, VNX7600, & VNX8000
Copyright 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.
For the most up-to-date listing of EMC product names, see EMC
Corporation Trademarks on [Link].
VMware is a registered trademarks or trademarks of VMware,
Inc. in the United States and/or other jurisdictions. All other
trademarks used herein are the property of their respective
owners.
Part Number h12102.2
3 EMC VNX FAST VP A Detailed Review
VNX5200, VNX5400, VNX5600, VNX5800, VNX7600, & VNX8000
Table of Contents
Executive summary.................................................................................................. 4
Audience ............................................................................................................................ 4
Terminology ............................................................................................................ 5
Introduction ............................................................................................................ 7
Using FAST VP for block ........................................................................................... 8
Storage pools ..................................................................................................................... 8
Homogeneous pools ...................................................................................................... 8
Heterogeneous pools ..................................................................................................... 8
Storage tiers ....................................................................................................................... 9
Per-tier RAID configuration ............................................................................................. 9
Extreme performance tier ............................................................................................. 10
Performance tier ........................................................................................................... 10
Capacity tier ................................................................................................................. 11
Considerations for using tiers ....................................................................................... 11
FAST VP LUN management ................................................................................................ 12
Tiering policies ............................................................................................................. 12
Managing FAST VP at the storage pool level ...................................................................... 15
Relocation schedule ..................................................................................................... 17
Manual relocation ........................................................................................................ 18
Using FAST VP for file ............................................................................................. 19
Management .................................................................................................................... 19
Creating a file system using FAST VP LUNs ........................................................................ 21
FAST VP operations................................................................................................ 23
FAST VP algorithm ............................................................................................................ 23
Statistics collection ...................................................................................................... 23
Analysis ....................................................................................................................... 24
Relocation .................................................................................................................... 24
General guidance and recommendations for FAST VP and multicore FAST cache ...... 27
Large scale migrations ...................................................................................................... 28
What drive mix is right for my I/O profile? ......................................................................... 29
Conclusion ............................................................................................................ 29
References ............................................................................................................ 30
4 EMC VNX FAST VP A Detailed Review
VNX5200, VNX5400, VNX5600, VNX5800, VNX7600, & VNX8000
Executive summary
Fully Automated Storage Tiering for Virtual Pools (FAST VP) can lower the Total Cost
of Ownership (TCO) and increase performance by intelligently managing data
placement. When FAST VP is implemented, the storage system measures, analyzes,
and implements a dynamic storage-tiering policy in a faster and more efficient way
than a human analyst.
Storage provisioning can be repetitive and time consuming, and estimates calculated
incorrectly can produce uncertain results. It is not always obvious how to match
capacity to the performance requirements of a workloads data. Even when a match is
achieved, requirements change, and a storage systems provisioning will require
constant adjustments.
Storage tiering allows a storage pool to use varying levels of drives. The LUNs within
the storage pool can use the necessary storage capacity for various applications
while meeting some of the required performance goals. FAST VP uses I/O statistics at
a 256MB slice granularity (known as sub-LUN tiering). It uses the relative activity level
of each slice to determine the need to promote to higher tiers of storage. For arrays
containing Flash drives, this creates a Flash-optimized array. Relocation is initiated at
your discretion through either manual initiation or an automated scheduler. FAST VP
removes the need for manual, resource intensive, LUN migrations, while still
providing the performance levels required by the most active dataset.
FAST VP is a licensed feature available on the new VNX series. FAST VP licenses are
available for both individual platforms or as part of the FAST Suite, which includes
licenses for both FAST VP and Multicore FAST Cache.
This white paper discusses the EMC FAST VP technology and describes its features,
functions, and management. It assumes that the FAST VP enabler is installed on the
array.
Audience
This white paper is intended for EMC customers, partners, and employees who are
considering using the FAST VP product. Some familiarity with EMC midrange storage
systems is assumed. You should be familiar with the material discussed in the white
papers Introduction to the New EMC VNX Series and Virtual Provisioning for the New
VNX Series.
5 EMC VNX FAST VP A Detailed Review
VNX5200, VNX5400, VNX5600, VNX5800, VNX7600, & VNX8000
Terminology
Automatic Volume Management (AVM) - Feature of VNX that creates and manages file
volumes automatically. AVM organizes volumes into storage pools for File that can be
allocated to file systems.
Command Line Interface (CLI) - Interface for typing commands through the Control
Station to perform tasks that include the management and configuration of the
database and Data Movers, and the monitoring of statistics for VNX for File cabinet
components.
Diskmark - A disk identification number on VNX for File.
File System - Method of cataloging and managing the files and directories on a
system.
Fully Automated Storage Tiering Virtual Pools (FAST VP) - Automatically assigns
different categories of data to different types of storage media within a tiered pool.
Data categories can be based on performance requirements, frequency of use, cost,
and other considerations. This is done by retaining the most frequently accessed or
important data on fast, high-performance (more expensive) drives, and moving the
less frequently accessed and less important data to lower-performance (less
expensive) drives.
Graphical User Interface (GUI) - Software that uses graphical objects such as pull-
down menus and operations such as drag-and-drop to allow the user to enter
commands and execute functions.
Load Balance - Relocation of slices across and within tiers based on relative slice
temperature.
Pool LUN - Logical unit of storage created in a pool. A pool LUN can be either a thin
LUN or a thick LUN.
Redundant Array of Independent Disks (RAID) - Method for storing information where
the data is stored on multiple disk drives to increase performance and storage
capacities and to provide redundancy and fault tolerance.
RAID Group - Group of disks that take on a RAID configuration
Rebalance - Automated data relocation to improve the data imbalance when new,
empty drives are combined with relatively full, existing drives in a pool.
Skew - Small percentage of capacity that is responsible for most of the I/O activity.
Slice - Minimum increment of capacity that can be allocated to a pool LUN. Pool LUNs
are comprised of slices.
Slice Relocation - Movement of 256MB blocks of data within or between tiers, based
on the relative activity level (temperature) of each block.
Storage Pool for Block - Group of drives for configuring pool LUNs (thick and thin).
6 EMC VNX FAST VP A Detailed Review
VNX5200, VNX5400, VNX5600, VNX5800, VNX7600, & VNX8000
Storage Pool for File - Groups of available file disk volumes, organized by AVM, that
are used to allocate available storage to file systems. They can be created
automatically when using AVM or manually by the user.
Temperature - Moving average of a slices activity level over time.
Thick LUN - Type of pool LUN in which allocated physical space is equal to the user
capacity seen by the host server.
Thin LUN - Type of pool LUN where physical space allocated can be less than the user
capacity seen by the host server.
Tier - Different categories of media in a storage pool. These categories provide various
levels of performance and capacity through several drive types. Available tiers are
Extreme Performance, Performance, and Capacity.
7 EMC VNX FAST VP A Detailed Review
VNX5200, VNX5400, VNX5600, VNX5800, VNX7600, & VNX8000
Introduction
Data has a lifecycle. As data progresses through its lifecycle, it experiences varying
levels of activity. When data is created, it is typically heavily used. As data ages, it is
accessed less often. Fully Automated Storage Tiering for Virtual Pools (FAST VP) is a
simple and elegant solution for dynamically matching storage requirements with
changes in the frequency of data access. FAST VP segregates disk drives into three
categories, called tiers:
Extreme Performance Tier Flash drives
Performance Tier Serial Attach SCSI (SAS) drives
Capacity Tier Near-Line SAS (NL-SAS) drives
FAST VP is a feature designed to aggressively reduce Total Cost of Ownership (TCO)
and maintain performance. Rather than using only one type of drive, mixing Flash,
SAS, and NL-SAS drives allows you to maintain performance requirements while
reducing the drive count. In some cases, a nearly two-thirds reduction in drive count
can be achieved, while in other cases, performance throughput can double by simply
adding less than 10 percent of a pools total capacity to Flash drives.
FAST VP has proven highly effective for a number of applications. Tests in OLTP
environments with Oracle1 and Microsoft SQL Server2 show that using FAST VP with a
heterogeneous pool containing Flash drives instead of a homogenous pool can lower
capital expenditure by 15 to 38 percent, reduce power and cooling costs by over 40
percent, and still increase performance. For more information about these benefits,
see the What drive mix is right for my I/O profile section of this paper.
The VNX series of storage systems delivers high value by providing a unified approach
to auto-tiering for file and block data. Both block and file data can use virtual pools
and FAST VP. This provides compelling value for users who want to optimize the use
of high-performance drives across their environment.
You can use FAST VP in combination with other performance optimization software,
such as Multicore FAST Cache. A common strategy is to use FAST VP to gain TCO
benefits while using Multicore FAST Cache to boost overall system performance.
There are also several scenarios in which it makes sense to use FAST VP for both
purposes. This paper discusses considerations for the best deployment of FAST VP.
1 Leveraging Fully Automated Storage Tiering (FAST) with Oracle Database Applications EMC white paper
2 EMC Tiered Storage for Microsoft SQL Server 2008 Enabled by EMC Unified Storage and EMC Fully Automated Tiering (FAST) EMC white paper
8 EMC VNX FAST VP A Detailed Review
VNX5200, VNX5400, VNX5600, VNX5800, VNX7600, & VNX8000
Using FAST VP for block
Storage pools
Storage pools are the framework that allows FAST VP to take full advantage of
different drive types. A pool is somewhat analogous to a RAID group, which is a
physical collection of drives on which logical units (LUNs) are created. Pools can
contain a few drives or hundreds of drives, whereas RAID groups are limited to 16
drives. Because of the large number of drives supported in a pool, pool-based
provisioning spreads workloads over more resources and requires minimal planning
and management efforts.
Pools can be homogeneous or heterogeneous. Homogeneous pools have a single
drive type (Flash, SAS, or NL-SAS), whereas heterogeneous pools have different drive
types.
Homogeneous pools
Homogeneous pools are recommended for applications with limited skew, such that
their access profiles can be very random across a large address range. Multiple LUNs
with similar profiles can share the same pool resources. These LUNs provide more
predictable performance based on the drive type employed. In a homogeneous pool,
only one drive type (Flash, SAS, or NL-SAS) is selected during pool creation.
Figure 1Homogeneous storage pools
As seen in Figure 1, each pool contains various sized LUNs that consist of one
particular drive type.
Heterogeneous pools
Heterogeneous pools consist of LUNs with different drive types. VNX supports Flash,
SAS, and NL-SAS drives in one pool. As shown in Figure 2 , there can be a maximum
9 EMC VNX FAST VP A Detailed Review
VNX5200, VNX5400, VNX5600, VNX5800, VNX7600, & VNX8000
of three drive types in a heterogeneous pool. Data in a particular LUN can reside on
some or all of the different drive types.
Figure 2Heterogeneous storage pools
Storage tiers
When creating storage pools on a VNX, data can use three different categories of
media in a single pool. These categories, referred to as storage tiers, provide various
levels of performance and capacity through several drive types. The available tiers
when creating a storage pool are:
Extreme Performance
Performance
Capacity
FAST VP differentiates each of these tiers by drive type, but it does not take rotational
speed into consideration. EMC strongly encourages that you avoid mixing rotational
speeds per drive type in a given pool. If multiple rotational-speed drives exist in the
array, you should implement multiple pools as well.
FAST VP can leverage one, two, or all three storage tiers in a single pool. Each tier
offers unique advantages in terms of performance and cost.
Per-tier RAID configuration
During storage pool creation, you can select RAID protection on a per-tier basis. Each
tier has a single RAID type, and once the RAID configuration is set for that tier in the
pool, it cannot be changed. Table 1 shows the RAID configurations that are supported
for each tier.
10 EMC VNX FAST VP A Detailed Review
VNX5200, VNX5400, VNX5600, VNX5800, VNX7600, & VNX8000
Table 1RAID configuration options
RAID type Preferred drive count options
RAID 1/0 4+4
RAID 5 4+1, 8+1
RAID 6 6+2, 14+2
Because RAID 5 (8+1) and RAID 6 (14+2) have high data to parity ratios, they provide
50% savings compared to RAID 5 (4+1) and RAID 6 (6+2), respectively. The tradeoff
for higher data to parity ratios are larger fault domains and potentially longer rebuild
times. This is especially true for RAID 5, which has only a single parity drive. EMC
advises you to choose carefully between RAID 5 (4+1) and RAID 5 (8+1) and decide
whether robustness or efficiency is a higher priority. Robustness is less likely to be an
issue for RAID 6, because it has two parity drives.
When creating a storage pool with any of the available tiers, EMC recommends that
you add drives in multiples of the preferred RAID configuration. If you select a drive
count that is not a multiple of the preferred RAID configuration, the system creates
non-preferred drive groups, which can affect performance.
For best practice recommendations, refer to the EMC Unified Storage Best Practices
for Performance and Availability Common Platform and Block Storage white paper
on EMC Online Support.
Extreme performance tier
Use the Extreme Performance tier when response times and performance are the most
important criteria for storage. The Extreme Performance tier uses Flash technology
solid-state drives (SSDs) that contain no moving parts. This revolutionizing
technology eliminates rotational latencies and can lead to a performance boost and
significant energy savings. The new VNX series also supports FAST VP-optimized Flash
drives. Compared to other Flash drives, FAST VP-optimized drives are more cost
effective and appropriate to use when data change rates are moderate.
Tests show that adding a small percentage of Flash capacity to storage and using
intelligent tiering products, such as FAST VP, can deliver double-digit percentage
gains in throughput and response time performance in some applications.
Flash drives have a higher per-GB cost but a lower per IO cost compared to traditional
spinning drives. To receive the best return, use Flash drives for data that requires fast
response times and high IOPS. FAST VP enables you to optimize the use of these
high-performance resources, because it automatically relocates hot data to the
Flash drives at a sub-LUN level.
Performance tier
Use the Performance tier to achieve a combination of performance and capacity. This
tier, composed of Serial Attach SCSI (SAS) drives, offers high levels of performance,
reliability, and capacity. SAS drives are based on industry-standardized, enterprise-
11 EMC VNX FAST VP A Detailed Review
VNX5200, VNX5400, VNX5600, VNX5800, VNX7600, & VNX8000
level, mechanical hard-drive technology that stores digital data on a series of rapidly
rotating magnetic platters.
The Performance tier includes both 10K and 15K RPM spinning drives, which are
available on all EMC midrange storage systems. When adding these drives to a
storage pool, EMC recommends using a single rotational speed per tier. The
Performance tier is valuable, because it offers high, all-around performance with
consistent response times, high throughput, and good bandwidth at a mid-level price
point.
Capacity tier
Use the Capacity tier to decrease the cost per GB of data. This tier, consisting of 7.2K
RPM Near-Line SAS (NL-SAS) drives, is designed for maximum capacity at a modest
performance level. Although NL-SAS drives have a slower rotational speed compared
to drives in the Performance tier, NL-SAS drives can significantly reduce energy use
and free up capacity in the more expensive and higher performing storage tiers.
In a typical system, up to 95% of application data has little I/O activity. Since NL-SAS
drives cost less than performance drives on a per-GB basis, and their cost is a small
fraction of the cost of Flash drives, they are the most appropriate type of media for
this cold data. NL-SAS drives consume 96% less power per terabyte than
performance drives and offer a compelling opportunity for TCO improvement that
considers both purchase cost and operational efficiency.
Note: For more information about the latest supported drives, refer to the Introduction
to the new EMC VNX Series VNX5200, VNX5400, VNX5600, VNX5800, VNX7600, &
VNX8000 white paper found on EMCs Online Support.
Considerations for using tiers
When all three tiers are available, there are considerations for using one tier or
another. For example, response times are lower with Flash drives compared to NL-SAS
drives, but capacity is greater in the Capacity tier as compared to the Extreme
Performance tier. Table 2 compares the three tiers.
12 EMC VNX FAST VP A Detailed Review
VNX5200, VNX5400, VNX5600, VNX5800, VNX7600, & VNX8000
Table 2. Comparison of the extreme performance, performance, & capacity tiers
Extreme Performance
(Flash)
Performance
(SAS)
Capacity
(NL-SAS)
P
e
r
f
o
r
m
a
n
c
e
User Reponse Time
1-5 ms 5 ms 7-10 ms
Multi-Access Reponse Time
< 10 ms 10-50 ms 100 ms
High IOPS/GB and Low
Latency
High bandwidth with
contending word loads
Low IOPS/GB
Leverages storage array SP
cache for sequential and
large block access
S
t
r
e
n
g
t
h
s
Provides extremely fast
access for reads
Sequential reads leverage read-ahead
Sequential writes leverage system optimizations
favoring disks
Executes multiple sequential
streams better than SAS
Read/write mixes
provide predictable
performance
Large I/O is serviced
efficiently
O
b
s
e
r
v
a
t
i
o
n
s
Writes slower than reads
Uncached writes are
slower than reads
Long response times for
heavy-write loads
Heavy concurrent writes
affect read rates
Single-threaded, large,
sequential I/O is equivalent
to SAS
FAST VP LUN management
Although FAST VP is a feature that is implemented for the entire storage pool, you can
modify settings at the LUN level. During the creation of a LUN via the GUI or
Navisphere Secure Command Line Utility (naviseccli), you can define a tiering policy.
A tiering policy specifies where the initial placement of data will be and how that data
will be relocated within the pool during scheduled and manually invoked relocation
periods. FAST VP bases decisions for how data relocation occurs on performance
statistics collected every hour. For more information, see the FAST VP Algorithm
section in this paper.
Tiering policies
FAST VP is a completely automated feature that implements a set of user-defined
tiering polices to ensure the best performance for various business needs. These
requirements are met by using FAST VP tiering policies, which determine how new
allocations and ongoing relocations should apply within a storage pool. The
13 EMC VNX FAST VP A Detailed Review
VNX5200, VNX5400, VNX5600, VNX5800, VNX7600, & VNX8000
relocations are done for each set of logically ordered blocks, called slices. FAST VP
uses an algorithm to make data relocation decisions based on the activity level of
each slice. It ranks the order of data relocation across all LUNs within each separate
pool. The system uses this information in combination with the per LUN tiering policy
to create a candidate list for data movement.
The following LUN level policies are available:
Highest Available Tier
Auto-Tier
Start High then Auto-Tier
Lowest Available Tier
No Data Movement
Users can set all LUN level policies except the No Data Movement policy both
during and after LUN creation. The No Data Movement policy is only available after
LUN creation.
Highest Available Tier
Use the Highest Available Tier policy when quick response times are a priority. This
tier is effective for LUNs which, although not always the most active, require high
levels of performance whenever they are accessed. The Highest Available Tier
policy starts with the hottest slices first and places them in the highest available
tier until the tiers capacity or performance capability limit is hit. Then it places the
slices into the next highest tier.
Auto-Tier
A small portion of a large set of data may be responsible for most of the I/O activity in
a system. FAST VP allows for moving a small percentage of the hot data to higher
tiers while maintaining the rest of the data in the lower tiers. The Auto-Tier policy
automatically relocates data to the most appropriate tier based on the activity level of
each data slice. Slices provisioned to a LUN are relocated based on the highest
performance disk drives available and the LUNs slice temperature. Although this
setting relocates data based on the LUNs performance statistics, LUNs set with
Highest available Tier take precedence.
Start High then Auto-Tier (recommended)
Start High then Auto-Tier is the default policy for each newly created pool LUN. This
is the recommended policy, because it takes advantage of the Highest Available
Tier and Auto-Tier policies. Start High then Auto-Tier sets the preferred tier for
initial data allocation to the highest performing disk drives with available space, and
then it relocates the LUNs data based on the performance statistics and the auto-
tiering algorithm. This tiering policy maximizes initial performance and takes full
advantage of the highest performing drives while improving TCO. With this tiering
14 EMC VNX FAST VP A Detailed Review
VNX5200, VNX5400, VNX5600, VNX5800, VNX7600, & VNX8000
policy, less active data is moved to lower tiers, making room for more active data in
the higher tiers. This tiering policy is especially good for LUNs that exhibit skew.
Lowest Available Tier
Use the Lowest Available Tier policy when cost effectiveness is the highest priority.
With this policy, data is initially placed on the lowest available tier with capacity.
Select this policy for LUNs that are not performance sensitive or response-time
sensitive. Regardless of their activity level, all slices of these LUNs will remain on the
lowest storage tier available in their pool. Slices with colder temperatures in the
LUN have higher priority to be relocated to the lowest tier.
No Data Movement
Select the No Data Movement policy after a LUN is created. If a LUN is configured
with this policy, no slices provisioned to the LUN are relocated across tiers. Data
remains in its current position, but can still be relocated within the tier. The system
still collects statistics on these slices after the tiering policy is changed.
Initial slice placement within a tier
The tiering policies listed above affect the initial placement of a LUNs slices within
the available tier. For example, initial placement with a pool set to Auto-Tier results
in the data being distributed across all storage tiers available in the tier. The
distribution of the data is based on the available capacity of the pool. If 70% of a
pools free capacity resides in the lowest available tier, then 70% of the new slices
are placed in that tier. LUNs set to Highest Available Tier and Lowest Available
Tier have their slices placed in the highest or lowest tiers, respectively, that have
capacity available. Since the No Data Movement policy can only be applied after a
LUN is created, initial placement is set to the policy that was used prior to the change
to no movement.
Summary of tiering policies
Table 3 summarizes the tiering policies.
15 EMC VNX FAST VP A Detailed Review
VNX5200, VNX5400, VNX5600, VNX5800, VNX7600, & VNX8000
Table 3. Tiering policies
Tiering Policy
Corresponding Initial Tier
Placement
Description
Highest Available
Tier
Highest Available Tier Sets the preferred tier for the initial data
placement and subsequent data relocation (if
applicable) to the highest-performing disk
drives with available space.
Ato-Tier Optimize for Pool
Performance
Sets the initial data placement to Optimized for
Pool and then relocates the LUN's data based
on the LUN's performance statistics such that
data is relocated among tiers according to I/O
activity.
Start High then
Auto-Tier
(Default)
Highest Available Tier First sets the preferred tier for the initial data
placement to the highest-performing disk
drives with available space, then relocates the
LUN's data based on the LUN's performance
statistics and the auto-tiering algorithm.
Lowest Available
Tier
Lowest Available Tier Sets the preferred tier for the initial data
placement and subsequent data relocation (if
applicable) to the most cost-effective disk
drives with available space.
No Data
Movement
Based on previous FAST VP
policy used
Prevents any ongoing data movement for this
LUN. The current placement of data is retained.
Only available as a LUN property after the LUN
is created.
Managing FAST VP at the storage pool level
Once you have created LUNs with the most appropriate tiering policy, you can view
and manage FAST VP properties at the storage pool level. When viewing the
properties of a specific pool, the Tiering tab provides summary information about the
amount of data targeted to be moved up, down, and within tiers. This tab also
provides the ability to choose to use manual or scheduled data relocation for the
pool.
16 EMC VNX FAST VP A Detailed Review
VNX5200, VNX5400, VNX5600, VNX5800, VNX7600, & VNX8000
Figure 3 shows these properties.
Figure 3. Storage pool tiering properties
In the storage pool tiering properties window, users can view both the tier status and
the tier details for the pool. The Tier Status section enables you to define the method
for starting data relocation for the pool. Choose Manual to manually start the data
relocation on a selected storage pool at any time. Choose Scheduled to have the
system perform data relocation based on the settings defined in the relocation
schedule. Click the Relocation Schedule button to bring up the Manage Auto Tiering
window and view this schedule.
The Tier Status section displays the following information:
Data Relocation Status, which can be:
Ready (no active data relocations)
Relocating (data relocations are in progress)
Paused (all data relocations for this system are paused)
Quantity of pool data, in GB, that is ready to move up, down, or be redistributed
within existing tiers for load balancing.
This is the amount of data to move during the next scheduled data relocation. This
number will change, since a recalculation is done right before the relocation
window starts. If the Data Relocation Status is Relocating, this quantity is the
amount of data to be moved during the current data relocation.
Estimated time for the data relocation based on the data gathered by the
software.
17 EMC VNX FAST VP A Detailed Review
VNX5200, VNX5400, VNX5600, VNX5800, VNX7600, & VNX8000
The tier details section displays the following information:
Data distribution per tier, including the quantity of free capacity, allocated
capacity per tier, and total capacity.
Amount of data that is ready to be moved up, down, or redistributed within the
tier.
RAID type for each tier.
Relocation schedule
The FAST VP feature allows for automatic data relocation based on a user-defined
relocation schedule. This schedule defines when and how frequently the array starts
data relocation on the participating storage pools. These scheduled relocations,
which take place on all pools concurrently, can be arranged to take place during off-
hours to minimize any adverse performance impact. The Manage Auto-Tiering window
accepts changes to be made to the schedule. You can launch this window from the
storage pool tiering properties or from the link in the task pane located in the Storage
section of Unisphere.
Figure 4 depicts the Manage Auto-Tiering window.
Figure 4. Manage Auto-Tiering window
In the Manage Auto-Tiering window, you can view both the current auto-tiering status
and the data relocation schedule. The Auto-Tiering Status section indicates whether
data relocation is enabled or paused. When data relocation is paused, all active
relocation tasks for all storage pools pause, meaning no future relocation requests
18 EMC VNX FAST VP A Detailed Review
VNX5200, VNX5400, VNX5600, VNX5800, VNX7600, & VNX8000
will be issued. Use the button to the right of the Relocation Status field to pause or
resume data relocations.
This section also displays the Pools with data to be moved field. This is the number
of tiered pools that have data queued for relocation. Scheduled Pools refers to the
number of tiered pools associated with the data relocation schedule. Active Pool
Relocations is the number of tiered pools currently participating in manual or
scheduled data relocation operations. The Auto-Tiering Status section also shows the
total amount of data in all tiered storage pools that is ready to move up to the next
higher tier, down to the next lower tier, and/or be redistributed on the disks for load
balancing, along with the estimated time (in hours) for completing all data relocation
operations.
The Data Relocation Schedule section enables users to define the operational rate for
the data relocation operation. Valid values are Low, Medium (default), or High, where
Low has the least impact on system performance, and High has the most impact.
This section also gives users the ability to disable/enable the schedule and change
the relocation start time, relocation duration, and days on which relocations can take
place. Multiple relocation operations can occur during this timeframe, and the
software assigns this schedule to all participating pools. In the example shown in
Figure 4, data relocations are allowed to run each day of the week for an eight hour
window, starting at 23:00 (11:00 PM) and concluding at 7:00 AM.
By default, the data relocation state for a created pool is set to Scheduled. As Figure 5
shows, users can right-click on a pool to change the data relocation state to Manual,
and can stop or start data relocation for the pool.
Figure 5. Scheduled auto-tiering
Manual relocation
Users can initiate manual data relocation for selected pools, whether or not they are
included in the auto-tiering schedule, at any time using the GUI or naviseccli. FAST VP
analyzes all gathered statistics prior to beginning the relocation to ensure that up-to-
date statistics and settings are properly accounted for prior to relocation. If the
relocation completes within the defined timeframe, FAST VP recollects statistics and
continues relocation until the timeframe closes.
19 EMC VNX FAST VP A Detailed Review
VNX5200, VNX5400, VNX5600, VNX5800, VNX7600, & VNX8000
When users select Start Data Relocation in the GUI, they can set the data relocation
rate and duration. The data relocation rate can be set to Low, Medium, or High. The
default is the relocation rate set in the Manage Auto-Tiering window. (See Figure 4.)
Users can also set the relocation duration, which specifies the time, in hours and
minutes, at which data relocations can occur.
Figure 6. Start data relocation
Although the automatic scheduler is an array-wide setting, the manual relocation
operation allows for immediate relocation to take place at the pool level. Common
situations for initiating a manual relocation include:
When reconfiguring a pool, if LUN properties, such as the tiering policy, change,
and the new priority structure must be realized immediately
As part of a script for a finer-grained relocation schedule
Using FAST VP for file
Management
The process of implementing FAST VP for file storage begins by provisioning LUNs
from a block storage pool with mixed tiers. The provisioned LUNs are then added to
the ~filestorage Storage Group. After performing a rescan of the storage systems via
the link on the Storage page in Unisphere, a diskmark operation is started that
presents the newly created storage to a new file storage pool. The rescan
automatically creates and maps the storage pool for File to the block storage pool
using the same pool name. Additionally, it creates disk volumes in a 1:1 mapping for
each LUN that was added to the ~filestorage Storage Group. At this point, you can
create file systems from the disk volumes contained in the new storage pool for File.
The FAST VP policy that was applied to the LUNs presented to the storage pool for File
operates the same as with other LUNs in the system, dynamically migrating data
between storage tiers in the pool.
Figure 7 illustrates the FAST VP for File workflow.
20 EMC VNX FAST VP A Detailed Review
VNX5200, VNX5400, VNX5600, VNX5800, VNX7600, & VNX8000
Figure 7. FAST VP for File workflow
FAST VP for File is supported for both Unisphere and the CLI. Its properties can be
seen in various areas throughout Unisphere, including on the properties page of
pools for File and the properties page for both volumes and file systems. The
Advanced Data Services section lists the tiering policy for the LUN along with other
enabled data services, including values for whether the LUN is thin and/or mirrored,
and whether its data is compressed. This section displays only if the storage pool on
which the file system is created is a tiered pool, or if the file system is created directly
from pool-based LUNs using manual volume management. If Advanced Data Services
are not enabled, the tiering policy does not appear.
For more information about other Advanced Data services, refer to the Introduction to
EMC VNX Storage Efficiency Technologies white paper found on EMCs Online
Support.
21 EMC VNX FAST VP A Detailed Review
VNX5200, VNX5400, VNX5600, VNX5800, VNX7600, & VNX8000
Creating a file system using FAST VP LUNs
As outlined in the diagram above, you can create file systems once LUNs are
presented to the ~filestorage Storage Group, and a storage pool for File is
automatically mapped to the block storage. After the new storage pool for File is
created, you can view the corresponding tiering policy in the properties window for
that particular pool, as shown in Figure 8.
Figure 8. Properties window for a storage pool for file
As seen Figure 8, the tiering policy is displayed in the Advanced Data Services
section. If multiple LUNs with different tiering policies are present in the pool, a
tiering policy of Mixed is displayed.
Similarly, once you create a file system, the tiering policy is also displayed in the
properties window of that particular file system, as shown in Figure 9.
22 EMC VNX FAST VP A Detailed Review
VNX5200, VNX5400, VNX5600, VNX5800, VNX7600, & VNX8000
Figure 9. File system properties window
As seen in Figure 9, you can view all of the properties of the file system along with the
tiering policy.
EMC highly recommends that you create file Systems using thick, non-compressed
LUNs with block deduplication disabled, as block deduplication is not supported on
File LUNs. In VNX OE for block release 33, the default LUN type for a new LUN is thin
so this can be accomplished by unchecking the Thin checkbox when creating the
LUN. Block deduplication and compression will be disabled by default. Thin,
deduplication, and compression optimization can be achieved using the thin and file
deduplication (which includes compression) attributes at the file system level.
Also, when creating a storage pool, the entire pool should be allocated to file
systems, and it is recommended to:
Create approximately one LUN for every four drives in the storage pool.
Create LUNs in even multiples of 10.
Number of LUNs = (number of drives in pool divided by four), rounded up to
nearest multiple of 10.
23 EMC VNX FAST VP A Detailed Review
VNX5200, VNX5400, VNX5600, VNX5800, VNX7600, & VNX8000
Make all LUNs the same size.
Balance LUN ownership across SPA and SPB.
Leave 5% available in the pool for relocations and recovery operations.
Have the same FAST VP tiering policy for all LUNs in a given VNX OE for file storage
pool.
Create a user-defined storage pool to separate VNX OE for File LUNs from the
same block storage pool that have different tiering policies.
Use thin enabled file systems, when using FAST VP with VNX OE for file, for
increased benefit from FAST VP multi-tiering.
When using MirrorView for file replication, it is important to remember that VNX file
configurations will not allow mixing of mirrored and non-mirrored types in a pool. If
you try to do this, the disk mark will fail.
Note: For more information about best practices for VNX for File, refer to the Applied
Best Practices Guide: EMC VNX Unified Best Practices found on EMCs Online Support.
FAST VP operations
FAST VP operates by relocating the most active data up to the highest available tier
(either the extreme performance or performance tier). To ensure sufficient space in
the higher tiers, relocations attempt to reclaim 10% free space in the tiers to allow for
new slice allocation that occurs when new LUNs are created or when thin LUNs
consume additional capacity. By reclaiming this 10% headroom, the least active
slices within each tier move to lower tiers (Performance or Capacity).
FAST VP works at a granularity of 256 MB and relocates data by moving the entire
slice to the appropriate tier, depending on the tiering policy selected for that
particular LUN.
FAST VP algorithm
FAST VP uses three different strategies to improve performance, capacity, and TCO.
These techniques help identify and relocate slices to the most appropriate tiers by
collecting statistics on each slice, analyzing the data, and relocating each slice based
on its activity level.
Statistics collection
As previously noted, a slice of data is considered hotter (more active) or colder
(less active) than another slice of data based on the relative activity level of those
slices. The activity level of a particular slice is determined by counting the number of
I/Os, reads and writes, bound for each slice. FAST VP maintains a cumulative I/O
count and weighs each I/O according to how recently it arrived. This weight
deteriorates over time, and newer I/Os are given a higher weight. After approximately
24 EMC VNX FAST VP A Detailed Review
VNX5200, VNX5400, VNX5600, VNX5800, VNX7600, & VNX8000
24 hours, the weights of new I/Os are nearly cut in half and continue to decrease.
This statistics collection occurs continuously in the background for all pool LUNs.
Analysis
FAST VP analyzes the collected data once per hour. This analysis process produces a
ranking order from hottest to coldest for each slice within the pool. Before
relocation is invoked with automatic or manual relocation, FAST VP performs a final
calculation and creates a candidate list of slices to move up, down, and within a pool.
The ranking of a LUN and its slices can be influenced by changing the tiering policy, in
which case the tiering policy takes precedence over activity level.
Relocation
During the user-defined relocation window, FAST VP promotes slices according to the
candidate list that it created in the analysis stage. During relocation, FAST VP
prioritizes relocating slices to higher tiers. Slices are only relocated to lower tiers if
the space they occupy is required for a higher priority slice. This way, FAST VP ensures
that the higher performing drives are always used.
After data is added to a pool, FAST VP attempts to move it to the higher tiers if space
is available and the tiering policy allows for it. The relocation process aims to keep
10% of the space in all tiers free. This creates space for any new slice allocations of
higher priority LUNs before the next relocation. Lower tiers are used for capacity as
the need arises. This entire relocation process is done automatically based on the
user-defined relocation schedule, or manually, if user-initiated. Figure 10 provides an
illustration of how FAST VP can improve slice placement for a pool.
25 EMC VNX FAST VP A Detailed Review
VNX5200, VNX5400, VNX5600, VNX5800, VNX7600, & VNX8000
Figure 10. FAST VP slice relocation
As seen in Figure 10, before FAST VP is installed on the array, data ranging from very
active to not active at all resides in a pool across each tier. The more expensive and
higher performing drives may contain non-active data that would be more appropriate
in the capacity tier, for which the cost per GB is less expensive. Once FAST VP is
installed, the array automatically collects and analyzes statistics and relocates data
across tiers to better utilize the more expensive drives. FAST VP moves the most
active data to a higher tier and the least active data to a lower tier.
Rebalance
Upon the expansion of a storage pool, the system recognizes the newly added space
and initiates an auto-tiering data relocation operation called rebalancing. A rebalance
relocates slices within the whole pool to achieve the best performance. If the pool is
heterogeneous, relocation happens not only within each tier, but also across tiers.
26 EMC VNX FAST VP A Detailed Review
VNX5200, VNX5400, VNX5600, VNX5800, VNX7600, & VNX8000
When a rebalance is initiated and running, Unisphere displays a percentage number
to indicate the progress. Figure 11 shows this progress report, which is displayed in
the pool properties.
Figure 11. Rebalance in progress
Figure 11 shows a rebalance operation that is 58% complete. When the pool is
expanding, the progress report also shows the operation state, which can be New,
Running, or Failed. There is no relocation window for a rebalance operation, which
runs until all slices in the relocation candidate list have been processed.
If you do not want an automatic rebalance to occur, because the newly added drives
will be used for a special purposed LUN, you can issue a rebalance skip using
naviseccli. For example:
st or agepool - expand - i d 0 - di sks 0_1_5- 0_1_10 ski pRul es
Note: For a more detailed synopsis of this command and others, refer to the VNX
Command Line Interface Reference for File documentation, located on EMC Online
Load-balance
In addition to relocating slices across tiers based on relative slice temperature, FAST
VP can also relocate slices within a tier to achieve maximum performance across
drives in that tier. Some drives within a tier may be more heavily used than others. To
improve performance, data slices are relocated within the tier to balance the load, if
the tier is not 100% full. Maintaining 5% or more of free space allows for load-
balancing to work efficiently. This is accomplished by extending auto-tiering data
relocation to also analyze and move data within a tier.
27 EMC VNX FAST VP A Detailed Review
VNX5200, VNX5400, VNX5600, VNX5800, VNX7600, & VNX8000
General guidance and recommendations for FAST VP and
multicore FAST cache
As noted previously, the FAST Suite consists of FAST VP and Multicore FAST Cache.
These two products are packaged together, work together, and complement each
other. Multicore FAST Cache allows the storage system to provide Flash drive class
performance to the most heavily accessed chunks of data across the entire system.
Multicore FAST Cache absorbs I/O bursts from applications, thereby reducing the load
on back-end hard drives. This improves the performance of the storage solution. For
more details on Multicore FAST Cache, refer to the EMC VNX Multi-Core FAST Cache
white paper found on EMC Online support. Table 4 compares FAST VP and Multicore
FAST Cache.
Table 4. Comparison between FAST VP and multicore FAST cache features
Multicore FAST Cache FAST VP
Allows Flash drives to be used to
extend the existing caching
capacity of th estorage system.
Allows a single LUN to leverage the
advantages of multiple drive types
through the use of storage pools.
Granularity is 64 KB. Granularity is 256 MB.
Data that is accesed frequently is
copied from HDDs to Flash drives.
Data is moved between different
storage tiers based on weighted
average-of-access statistics collected
over a period of time.
Used when workload changes are
unpreditable and very dynamic,
and require a quick response time.
Used when workload pattern changes
are predictable and relatively low.
Constantly promotes frequently
accessed HDD data to Multicore
FAST Cache. There are no
relocation cycles.
Data movemnet occurs in scheduled or
manually invoked relocation window.
Real-time monitoring decides
which data needs to be promoted
to Multicore FAST Cache.
Hourly analysis decides which portion of
data needs to be moved.
Multicore FAST Cache and FAST VP can be used together for sub-LUN tiering features
to yield high performance and improved TCO for the storage system. For example, in
scenarios where limited Flash drives are available, the Flash drives can be used to
create the Multicore FAST Cache, and FAST VP can be used on a two-tier pool
(Performance and Capacity). From a performance point of view, Multicore FAST Cache
dynamically provides performance benefits to bursts of data, while FAST VP moves
hotter data to higher tiers. From a TCO perspective, Multicore FAST Cache, with a
28 EMC VNX FAST VP A Detailed Review
VNX5200, VNX5400, VNX5600, VNX5800, VNX7600, & VNX8000
small number of Flash drives, serves the data that is accessed most frequently, while
FAST VP optimizes disk utilization and efficiency.
As a general rule, use Multicore FAST Cache when storage system performance needs
to be improved immediately for burst-prone data. Multicore FAST Cache complements
FAST VP by focusing on improving performance, while FAST VP focuses on improving
TCO.
The Multicore FAST Cache feature is storage-tier-aware and works with FAST VP to
make sure that storage system resources are not wasted by unnecessarily copying
data to Multicore FAST Cache, if that data is already on a Flash drive. If FAST VP moves
a chunk of data to the Extreme Performance Tier, Multicore FAST Cache will not
promote that chunk of data into Multicore FAST Cache, even if Multicore FAST Cache
promotion criteria are met.
A general recommendation for the initial deployment of Flash drives in a storage
system is to use them for Multicore FAST Cache. In almost all cases, Multicore FAST
Cache, with a 64K granularity, offers the industrys best optimization of Flash
technology.
Large scale migrations
EMC recommends that large scale migrations turn off tiering until the migration
completes. If critical data is being tiered during or immediately following the
migration, EMC recommends using the Highest Available tiering policy.
When the migration process starts, it is best to fill the highest tiers of the pool first.
This is especially important for live migrations. Using the Auto-Tier setting would
place some data in the Capacity tier. At this point, FAST VP has not yet run an analysis
on the new data, so it cannot distinguish between hot and cold data. Therefore,
with the Auto-Tier setting, some of the busiest data may be placed in the Capacity
tier.
In these cases, you should set the target pool LUNs to the highest tier. This way, all
data is initially allocated to the highest tiers in the pool. As the higher tiers fill and
capacity from the Capacity (NL-SAS) tier starts to be allocated, you can stop the
migration and start manual FAST VP relocation.
Assuming the analysis process has run for a sufficient amount of time, relocation
orders the slices by rank and moves data appropriately. In addition, since the
relocation attempts to free 10% of the highest tiers, these tiers will have more
capacity for new slice allocations.
This iterative process can be used while more data is migrated into the pool. You can
start FAST VP to begin relocation when most of the new data is allocated to the
Capacity tier. Once all of the data is migrated into the pool, you can select the best
tiering policy.
29 EMC VNX FAST VP A Detailed Review
VNX5200, VNX5400, VNX5600, VNX5800, VNX7600, & VNX8000
What drive mix is right for my I/O profile?
It is common for a small percentage of overall capacity to be responsible for most of
the I/O activity. Analysis of an I/O profile may indicate that 85 percent of the I/O
activity to a volume only involves 15 percent of the capacity. The resulting active
capacity is called the working set. Software like FAST VP and Multicore FAST Cache
keeps the working set on the highest-performing drives.
It is common for OLTP environments to yield working sets of 20 percent or less of their
total capacity. These profiles hit the sweet spot for FAST VP and Multicore FAST
Cache. Other I/O profiles, such as Decision Support Systems (DSS), may have much
larger working sets. In these cases, you can use FAST VP to deploy Flash drives,
because DSS workloads are not typically Multicore-FAST-Cache-friendly. You can use
Capacity Tier drives to lower TCO.
At a minimum, the capacity across the Performance Tier and Extreme Performance Tier
(and/or Multicore FAST Cache) should accommodate the working set. However,
capacity is not the only consideration. The spindle count of these tiers needs to be
sized to handle the I/O load of the working set.
Drives in the Performance Tier are versatile in handling a wide spectrum of I/O
profiles. Therefore, EMC highly recommends including Performance Tier drives in each
pool. The Multicore FAST Cache can be an effective tool for handling a large
percentage of activity, but inevitably, there will be I/Os that are not promoted or are
cache misses. The Performance Tier drives offer good performance for these I/Os.
The Performance Tier drives also facilitate faster promotion of data into the Multicore
FAST Cache by quickly providing promoted 64 KB chunks to the Multicore FAST Cache.
This minimizes Multicore FAST Cache warm-up time as some data gets hot and
other data goes cold. If the I/O profile has a significant component of random
writes, these are best served from Performance Tier drives as opposed to Capacity
drives.
You can use Capacity drives to optimize TCO. This often equates to comprising 60% to
80% of the pools capacity. Of course, there are also profiles with low IOPS/GB and or
sequential workloads that may result in the use of a higher percentage of Capacity
Tier drives.
You can engage EMC Professional Services and qualified partners for detailed
analysis and recommendations for future configurations. They have the tools and
expertise to make very specific recommendations for tier composition based on an
existing I/O profile.
Conclusion
With the use of FAST VP, users can remove complexity and management overhead
from their environments. FAST VP utilizes Flash, SAS, and NL-SAS drives (or any
combination thereof) within a single pool. LUNs within the pool can then leverage the
advantages of each drive type at the 256 MB slice granularity. This sub-LUN-level
30 EMC VNX FAST VP A Detailed Review
VNX5200, VNX5400, VNX5600, VNX5800, VNX7600, & VNX8000
tiering ensures that the most active dataset resides on the best-performing drive tier
available, while maintaining infrequently used data on lower-cost, high-capacity
drives.
Relocations can occur without user interaction on a predetermined schedule, making
FAST VP a truly automated offering. In the event that relocation is required on-
demand, you can invoke FAST VP relocation on an individual pool using the
Unisphere GUI or CLI.
Both FAST VP and Multicore FAST Cache work by placing data segments on the most
appropriate storage tier based on their usage pattern. These two solutions are
complementary, because they work on different granularity levels and frequency.
Implementing both FAST VP and Multicore FAST Cache can significantly improve
performance and reduce cost in the storage environment.
References
The following white papers are available on the EMC Online Support website:
EMC Introduction to the new EMC VNX Series VNX5200 VNX5400, VNX5600,
VNX5800, VNX7600, & VNX8000
EMC Tiered Storage for Microsoft SQL Server 2008Enabled by EMC Unified
Storage and EMC Fully Automated Storage Tiering (FAST)
EMC Unified Storage Best Practices for Performance and Availability Common
Platform and Block Applied Best Practices
EMC Unisphere Unified Storage Management Solution for the New VNX Series
EMC Virtual Provisioning for the New VNX Series Applied Technology
EMC VNX Multicore FAST Cache
Introduction to EMC VNX Storage Efficiency Technologies
Leveraging EMC Unified Storage System Dynamic LUNs for Data Warehouse
Deployments
Leveraging Fully Automated Storage Tiering (FAST) with Oracle Database
Applications