Você está na página 1de 51

1-1

This section discusses the hardware fundamentals of the Storage Center. At the
end of this section, the student will be able to identify the various components
that make up the Storage Center.

Section Objectives:
Storage Center hardware components include:
SC8000 controller
Controller port identification
Controller FRUs
Enclosures
Enclosure FRUs
Disk Types
IO Cards

1-2
Whats in it for you?
At the conclusion of this module, you should be able to:
Understand the basics of a Storage Center solution.
Define the core software included in a Storage Center solution.
Define the application software available in a Storage Center solution.
Discuss updates included in version 6.3 and 6.4 of the Storage Center software.


1-3
The Dell Compellent Storage Center
TM

Storage Center is a modular product offering a build-as-you-go approach that
offers a low cost of entry with unlimited growth potential. An open-systems
hardware architecture enables simultaneous mixing and matching of any
server/host interface or drive technology, providing the flexibility to use the
optimum technology for the deployment-specific requirements. Features
include:
Technology independence allows the mix and match of server interfaces and
drive technology
Flexible growth allows for growth of the capacity, connectivity and
functionality of the Storage Center
Configurable performance designed with the speed and bandwidth the
deployment requires
Enterprise level features, reliability and availability
Online migration allows adding additional capacity, connectivity or
functionality while online
Core software with capabilities that integrate functionality others price
separately
Powerful options that offer additional application software that optimizes
utilization and accelerates data recovery
A single robust interface that streamlines administration of all storage
resources

1-4
Deployment Architecture Options

Storage Center can be deployed in a variety of configurations:

Single Controller, Single Enclosure

Single Controller, Multiple Enclosures

Dual Controller, Single Enclosure

Dual Controller, Multiple Enclosures
1-5
Multi-Site Architecture

Storage Center supports the use of multiple systems in a local site / remote site
architecture in support of data replication.

Replication is used to duplicate volume data, possibly in support of a disaster
recovery plan or simply for rapid local access to a remote data volume.

Typically, data is replicated over some distance to safeguard against local or
even regional data threats as part of an overall disaster recovery plan.

The source system is referred to as the replicating system and the destination
system is referred to as the target system.

The connections between the replicating system and the target system can be
either Fibre Channel or iSCSI.

The replication process can be configured as either synchronous or
asynchronous replication.
1-6
The Storage Center
Storage Center consists of the following components:
Controllers
Performs the overall control of the Storage Center
Processes the RAID functionality
Processes all core and application software
Redundant firmware on industrial flash
Real-time embedded operating system
Redundant hot swappable power supplies
Modular PCI-based IO cards to provide connectivity
Redundant hot swappable fans

Enclosures
SAS 12 or 24 disk drive capacity
Fibre Channel 16 disk drive capacity
Redundant hot swappable power supplies

Switches
Provides server connectivity

Host Bus Adapters (in the servers)
Provide server connectivity to FC, FCoE and iSCSI

1-7
HBA (Host Bus Adapter) Server Configuration

Host bus adapters may need certain settings modified to operate with the Storage Center. The
following references a Qlogic HBA with a Windows OS. Other HBAs will require similar
changes.

Qlogic HBA Server Configuration Recommendations
Host Adapter Settings:
Host Adapter BIOS = Enable (if booting from SAN; if booting locally accept the default
of Disable)
Connection Options = 1 (for point to point connections on the Front End)
Spinup Delay = Disabled
Advanced Adapter Settings:
Luns per Target = (as appropriate for your site)
Enable LIP Reset = Yes
Login Retry Count = 60
Port Down Retry Count = 60
Link Down Timeout = 30
Execution Throttle = Number of buffers/number of connected servers

There are additional settings that may need to be configured. It is recommended that you refer
to the 6.x User Guide for a complete list of HBA settings. If additional information is needed,
check with the HBA manufacturer or Dell Compellent Copilot Services.
3
rd
party HBA vendors tools to view these values:
Qlogic - SANSurfer
Emulex HBAnywhere
To view the WWN of HBAs in a physical Windows server - fcinfo

1-8
Fibre Channel Zoning

Fibre Channel Switched Fabrics can be segmented into areas that are known as
zones through a process known as zoning. Zoning is the partitioning of a
fabric into smaller fabrics to restrict access and/or add security.

Zoning applies only to the Fibre Channel - Switched Fabric (FC-SW) topology.
Zoning is used to create dedicated paths between Initiators (devices requesting
IO) and Targets (devices IO is requested from such as a hard drive).

Zoning can be created using either ports on the switches or the WWNs of the
end-devices.
Port zoning restricts communication at the port level
This prevents ports from communicating with unauthorized ports

Name zoning restricts access by World Wide Name
This is more flexible, but WWNs can be spoofed, reducing security

Note: Dell Compellent Best Practice is to use single initiator, multiple target
zones.
1-9
Storage Center Software Overview

Storage Center - Core
The Storage Center core software provides the essential storage
functions, providing functionality previously available only in high-
end systems.

Storage Center - Applications
The Storage Center provides a rich set of licensed applications

These applications have been designed to support the dynamic
storage needs of the customer.

Storage Center - Unified Management
Storage Center provides a common management and unified
control interface to both the controller and the enclosure.

Through this interface, the administrator can configure both
controller and enclosure resources.

1-10
Storage Center - Core

Using Dell Compellents Fluid Data Architecture, the Storage Center provides
the essential storage features providing functionality previously available only in
high-end systems.

Virtualization
Efficient utilization of storage resources by managing disks as a
single pool and presenting disk resources to any server
Manages all disks (any capacity, any speed, any interface) as single
disk pool
Upgrades and modifications can be performed in a non-disruptive
matter

Cache
Enhances overall performance and availability through multi-
threaded read ahead and mirrored write cache (in a Dynamic
Controller environment)

Boot from SAN
Allows servers to boot from the SAN eliminating the need for
internal drives in the servers
1-11
Storage Center - Core (cont.)

Copy-Mirror-Migrate
Copy, mirror and migrate volumes without impacting users across
different disk types, RAID levels, or enclosures

Thin Import
Convert data from legacy systems into thin provisioned volumes on a
Dell Compellent SAN

Virtual Ports
Eliminates the requirement for reserve ports in a Cluster Controller
configuration now allowing all front end ports to act as primary ports

Charting Viewer
The Storage Center Manager comes with a complete suite of
management tools and real-time charting capabilities are provided

Phone Home
Contacts Dell Compellent technical support regularly, on-demand or
upon a system affecting event
1-12
Storage Center - Core (cont.)

Dynamic Capacity
Use significantly less storage and improve overall performance by
allocating space only when data is physically stored

Data Instant Replay
Recover from data hazards in minutes by creating any number of space-
efficient Replays that restore data instantaneously

Enterprise Manager
Manage multiple Storage Centers from one location, configure and
manage Remote Instant Replay and Live Volumes, establish trends using
multiple reports and configure user defined threshold alerts for
notification about system performance

Datapage Sizes
Optimized application performance through variable page sizes

Dynamic Controllers Add a second Storage Center controller to increase
system availability, capacity and performance

1-13
Storage Center Applications

The Storage Center provides a rich set of licensed applications that have been
designed to support the dynamic storage needs of any customer. Storage Center
applications can be added to any configuration, allowing users to scale hardware and
software from entry-level to enterprise with one common interface.

Remote Instant Replay Synchronous or asynchronous Remote Instant Replay sends
volume replays to remote systems

Data Progression Lower total storage costs by automatically tracking usage and
migrating data between storage classes based on user defined rules. Retain
frequently accessed data on high performance storage devices, and
infrequently accessed data on lower cost storage devices

FastTrack Dynamically places the most active data on the outer disk tracks

Replay Manager Allows for application consistent replays for SQL, Exchange and
Hyper-V volumes

Live Volume Volumes are replicated between primary and secondary Storage
Centers.


1-14
The page pool is the technical term that refers to all the physical disks that
reside in a Storage Center. In the Storage Center interface the page pool is
referred to as a Disk Folder. It is a collection of allocated and unallocated
disk blocks (pages).
The default page size is 2MB.
The page pool takes up a percentage of the total space. That percentage
varies based on the overall space in the Storage Center. Some pages are
ready to be used as RAID 10, some RAID 5 and some RAID 6.
The page pool provides automated, sophisticated block management to
make it much easier for the user to get his data in the optimum location on
the SAN.
The page pool grows and shrinks based on the demands the users place on
it. It is intelligent enough to automatically defragment and tune itself.
Metadata that is saved for each page (block):
Creation, access and modification time
Frequency of access
Type and tier of disk drive
RAID level
Corresponding volume
Active (RW) or Frozen (Read Only)
The blocks metadata is not stored on the block itself.


1-15
Core: Disk Virtualization

The Storage Center manages all disks as a single pool of resources from which
the administrator can dynamically select and assign disk resources to server
resources. Virtualization employs full disk parallelization across heterogeneous
servers and disks.
Overcome conventional problems:
Volumes span a fixed subset of disks
Storage restricted to single servers
Homogeneous drives and interfaces
Long-term configuration decisions must be made before usage
patterns are understood

Disk Virtualization benefits:
Manage all disks as a single pool
Present disk resources to any server
Mix and match any capacity, speed, or interface
Changes can be made in a non-disruptive matter
Reduce costs
Optimize performance
Increase availability
Maximize resource utilization

1-16
Common RAID Levels
RAID (Redundant Array of Independent Disks) is the method of using multiple
drives acting as one storage pool. There are different methods of storing the
data and each method offers different performance and redundancy levels.
RAID 0 offers fast performance but no redundancy. All drives are used to store
data but if one drive fails, the information is lost.
RAID 1 offers fast performance and high availability. All data is written to one
drive and then mirrored on a second drive. If either drive fails, the remaining
drive maintains access to the data. The trade off is in cost. While both drives are
used, the available storage space is only half the raw disk space. This makes the
cost of storage higher.
RAID 5 offers good performance and efficient storage. Data is striped across
multiple drives but a parity bit is added so if a drive fails, the missing data can be
reconstructed from the remaining drives and the parity. The parity information
consumes the equivalent of one drive so usable storage is the total drive space
minus one drive.
RAID 10 offers fast performance and high availability. Data blocks are mirrored
to other drives and then striped. If a drive fails, data is still accessible from the
mirror. Rebuild times are shorter since no parity needs to be calculated. The
trade off is again cost. Usable space is half the raw disk space.
1-17
RAID Consumption Considerations

The use of the different RAID types that the Storage Center supports has a direct impact on the
amount of storage consumed on the disks when a write occurs. Dell Compellent implements
RAID 5 in either of two methods; RAID 5-5 and RAID 5-9. RAID 5-5 includes a parity bit every
5
th
disk and has a 20% overhead while RAID 5-9 includes a parity bit every 9
th
disk and has an
11% overhead. Dell Compellent uses RAID 5-9 by default. Below is a comparison of the
different RAID types.

RAID 10
Fastest performing RAID type
Consumes 2 times the RAW disk space

RAID 5-5
Medium performance on WRITEs, little or no impact on READs
Consumes 20% of storage for overhead
Provides better performance in degraded mode because it uses half the I/Os

RAID 5-9
Provides the same performance on WRITEs as RAID 5-5, little or no impact on READs
Consumes approximately 11% of storage for overhead
Need at least 9 disks to use
Greater possibility of losing a disk in the RAID set
Provides lesser performance in degraded mode due to I/O usage
1-18
Dual Redundancy

Dual Redundancy is offered through the use of RAID 6 and RAID 10 Dual
Mirror. Each addresses the concern of losing two drives within the same stripe
set. RAID 10 DM offers protection by striping the data and then creating two
mirrors. This has a higher overhead than normal RAID 10.

RAID 6 offers similar protection by striping the data, then calculating a parity bit
twice and writing each of the parity bits to a separate disk. While the writes are
slower than RAID 10 or RAID 10 DM, RAID 6 offers good read performance
while still protecting against a two drive failure.

RAID 10 DM overhead is 66%

RAID 6 is offered in two versions; RAID 6-6 and RAID 6-10
RAID 6-6 overhead is 34% while RAID 6-10 overhead is 20%

Best Practice: If individual drive capacities in a tier are greater than 900GB, set
the tier to Dual Redundancy. Tiers may be modified at anytime, however within
a tier it may only be R5/R10 or R6/R10DM.

1-19
Core: Boot From SAN

The boot from SAN function allows servers to use an external SAN volume as
the servers boot volume. This removes the requirement that servers contain
internal DAS drives.

Benefits
Diskless servers reduce cost
Easier transition to blade servers
Improves server management and recovery
Allows multiple boot images to be centrally stored for easy
backup, management and restoration
No license-per-server is required for Dell Compellents boot
from SAN function

Considerations
HBA BIOS needs to be configured to boot from storage
The server BIOS needs to be configured to boot from the HBA
When loading the O/S, the O/S may need to be configured to
boot from a remote server

1-20
Core: Dynamic Capacity

Dynamic Capacity allows the system to allocate space only when data is
physically written to the volume, significantly reducing the initial storage needs
and improving overall performance.
In the first example, you will notice how 10 terabytes can be presented to
servers from the Storage Center but because of how Dynamic Capacity works
only 1 Terabyte of actual space is used.
The second scenario shows how a traditional SAN (without Thin Provisioning)
allocates disk space and how the Storage Center allocates space. Notice how
much space is allocated but not used on a traditional SAN. Both volume 1 and
volume 2 have lots of unused allocation.
In the Storage Center you dont need to purchase extra space until you actually
intend to use it.
Be aware - A user can possibly over allocate beyond physically available
storage.
1-21
Core: Data Instant Replay Features

Leverages Fluid Data Architecture
No pre-allocation needed
No duplication of data
Automatic coalescence

Unrestricted number of replays per volume

Multiple schedules per volume

Recovered replays are full read/write volumes
Can be mapped to any server

1-22
Core: Data Instant Replay

A replay is a point-in-time-copy (PITC) of a volume to provide fast recovery
from hazards such as viruses, power outages, hardware failure, or human error.
Restoring a replay can recover lost data or revert the volume to a previous
point in time.

Dell Compellent's replays differ from the traditional PITC concept in that the
blocks or pages are frozen and not copied. No user data is moved, making the
process efficient in both time taken to complete the replay and space used by
replays.
Benefits
Unlimited, space efficient, time-based replays at any time interval
Instantaneous recovery from data hazards by restoring a replay
Minimize system downtime
Reduce dependence on tape-based backups
Intuitive profiles and scheduling mechanism
Replays can also be used to test new applications on actual data without
risk of data loss or corruption
1-23
Replay Recovery

When a volume is recovered, it is labeled as a view. View volumes share some
of the data with the replay it was recovered from and data from the source
volume.

Volume 1 (Source Volume - Blue)
Recovery of Volume 1 is started at Time 2

Volume 2 (Recovered Volume - Red)
Data in Volume 2 consist of Volume 1 coalesced data and the
deltas at the time of the replay was taken on the Volume 1.
This becomes a new branch
The new branch is only dependant on the old branch
while read data is needed, that replay cannot be expired
until Volume 2 no longer needs that data.
DIR not enabled on Volume 2 by default
Any new writes become part of the newest data set and
considered active on Volume 2

If a replay was taken on Volume 2, only the new writes on Volume 2 would be
in that replay and would be independent of Volume 1 data
1-24
How Data Instant Replay Works:

A replay is a point-in-time-copy (PITC) of a volume to provide fast recovery from
hazards such as viruses, power outages, hardware failure, or human error. Restoring a
replay can recover lost data or revert the volume to a previous point in time.
Dell Compellent's replays differ from the traditional PITC concept in that the blocks or
pages are frozen and not copied. No user data is moved, making the process efficient
in both time taken to complete the replay and space used by replays.

Replay is taken - Once a replay has been scheduled or initiated manually, only the
data that has been changed (Deltas) between the time marks makes up the size
of the replay

Follow-on Replays - Only the Deltas will consume space in successive replays

A Replay Expires - When a DIR expires the information is coalesced
Newer written data takes precedence in a conflict
Space is then released back into the common pool

Expiration is complete - Data has finished coalescing
Space has been released back to the common pool
1-25
Data Instant Replay Data Pages

Data in various pages have different accessibility characteristics. Their
interaction describes the process of the DIR.
Active Pages
Is Data that is Read/Writable. This is the newest or most current
data set

Accessible Data Populated in R/O Pages
This data is Read/Only, but is accessible. It is from an earlier time
mark, but has not been changed

Inaccessible Data Populated R/O Pages
Data that has had changes to them. This data is Read/Only and
can only be accessed through recovering a view volume

Replay Data consists of only the data that has been changed (Deltas).
After a replay is taken on a volume, all active pages become Read/Only
1-27
Data Progression, a licensed application, leverages cost and performance differences
between storage tiers allowing the maximum use of lower cost, higher capacity SAS
(7.2K RPM) drives for stored data, while maintaining performance oriented (15K RPM)
SAS or SSD drives for frequently-accessed data. Data Progression takes advantage of
this by automatically tracking block usage and migrating data between storage classes
based on storage profiles.

Best Practice: For replay data progression to work properly, there must be at least a
daily replay and an hour overlap for replay expiration. If only a single daily replay is
being taken, the minimum expiration time should be 25 hours.
Benefits
Cost and performance are optimized when volumes use the Recommended
profile
Automate data movement to the proper storage class based on real-time
usage
All applications and operating systems require zero modification
Minimize data management administrative time

Note: If Data Progression is not licensed and a system uses RAID 10 and RAID 5, data is
migrated within a tier (drive class) but cannot be migrated between tiers.

How drives are defined into Tier 1, 2 or 3:
Tier 1 fastest drives in the Storage Center
Tier 3 slowest drives in the Storage Center
Tier 2 all other drives that are not in tier 1 or tier 3.
1-28
Storage Profiles used in Data Progression
Here is the interface that shows the four different system defined Storage
Profiles that can be used in support of Data Progression.
The profiles are defined on the previous slide and will be discussed in more
detail in a later chapter.
We wanted to show you the slide here so you can see where these profiles
come into play in the Storage Center interface.

1-29
Data Progression Volumes: Wheres My Data

The Data Progression application allows the user to create volumes which will
take advantage of using multiple RAID types and multiple classes of disks. This
will enable frequently accessed data to be on the highest performance RAID
type on the highest performance class of disk, while less frequently accessed
data may reside on a lower performance RAID type on a lower performance
class of disk.

All initial writes to the volume are written to the highest performance RAID type
and class of disk, providing optimum performance for write operations.
The System Manager groups disks by speed; 7K, 10K, 15K and Solid State drives.
Data Progression uses Fluid Data Architecture to move data to performance
appropriate and cost-effective disk tiers.

Data Progression is configurable with storage profiles, although it is strongly
suggested to use the Recommended profile for optimal performance and
storage usage.

Note: Data Progression works with VMware just as it would any other operating
system. All virtual hard disks contained in a single volume will share the same
data progression configuration


1-30
Here is an example of how Data Progression would work with an Exchange file. A 10GB
Exchange edb file is initially created and placed on Tier-1,15Krpm SAS storage to benefit
from its speed and performance. (It could also be placed on SSD drives.)
An edb file contains components such as user InBox, Sent items, Contact information and
Calendar data. Initially this information is new and current for a given user and requires
maximum performance.
Over time, the user data becomes stale and infrequently accessed. Traditionally, this older
data remains on more expensive tier-1 storage even though it is infrequently or never
accessed. Other data migration solutions granularity is either the entire volume or the entire
.edb file. In this example, the full 10GB .edb file will remain in Tier 1 taking up the most
expensive disks.
Dell Compellents Data Progression software will automate this migration process at the
block level. The Data Progression software makes the decision to migrate the stale data
blocks to a lower, more cost effective tier of storage and not affect the blocks needing
higher performance.
Because Data Progression works at the block level, the 10GB Exchange.edb file will actually
be made up of blocks from all 3 tiers of storage in the Storage Center. The result will
typically look something like this: with 500MB of active data (Inbox, calendar, etc.)in Tier 1.
500MB of data that was accessed a couple weeks ago (last months calendar, deleted items,
etc.) 9GB of data that have not been accessed in a while sitting on the less expensive drives.
If data that has been migrated becomes active again the Storage Center will migrate the
data, at the block level, back up the tiers to provide the performance needed.
All this is done without a host agent deployed.
1-31
Page Lifecycle - Concepts

When data is first written, by default, it will be striped in Tier 1 using RAID 10
(blocks A&A,B&B,C&C,D&D). When a Replay is taken, the blocks will be marked
as READ ONLY.

Then when Data Progression runs, the blocks will be restriped into RAID 5 (5-9
by default) blocks A,B,C,D with Parity (P).

When newer data is written (blocks C1&C1) they are striped as a RAID 10. Block
C is now older data.

The next time Data Progression runs, block C is moved to the lowest Tier-
R5/R6 as Inaccessible Data. This means that the only way to view this data is to
create a view volume from the Replay.

In addition block B has now aged. While there is no newer data, since the block
has aged, it is now progressed down to the next defined tier.


Note: Conceptual model. In reality, data layout is more complex

1-32
Applications: FastTrack

FastTrack is a licensed Storage Center feature. With this enabled data is
promoted and demoted from outer to inner tracks of a disk using the same
algorithms implemented for Data Progression.

FastTrack is enabled for all volumes if it is licensed
Use Online Storage report to see a system-wide view of used space on
FastTrack blocks
Use the Statistics screen to view FastTrack use on a per volume basis

FastTrack Features:
Reserves the outer tracks for the most active blocks
Dynamically places the most frequently used data on the fastest tracks
of each drive
Automated FastTrack configuration optimizes storage usage
Seek time is minimized through intelligent data placement




1-33
Core: Datapage Sizes

2 MB is Dell Compellents default datapage size. This selection is appropriate
for most applications.

Additional page sizes can be selected. These additional sizes can be made
accessible by changing Advanced User Options under Configure My Volume
Defaults.

512 KB: This datapage size is appropriate for applications with high
performance needs or environments in which replays are taken frequently
under heavy IO. Selecting this size reduces the amount of space the System
Manager can present to servers.

4 MB: This datapage size is appropriate for systems that use a large amount of
disk space with infrequent replays.

Caution: If you are considering using either the 512 KB or 4 MB datapage
setting, it is recommended you contact Copilot services so that system
resources remain balanced and the impact on performance is considered.

1-34
Command Utility (CompCU) Requirements

The Storage Center Command Utility (CU) is a java program that is used to
run scripted commands to the Storage Center. It can be run from multiple
operating systems (such as Windows, Linux, AIX, etc.) and can automate
functions such as:
Creating and/or modifying a volume
Creating replays on volumes
Mapping volumes to servers
Creating and mapping replay view volumes
Creating, modifying or deleting users and user groups
Perform alert management and copy/mirror/migrate operations

Commands can be executed from a command prompt or included in a script
file for batch execution.

Note:
To create or modify Storage Center objects, administrative rights are
required on the Storage Center
For a complete list of all CU functions refer to the Command Utility
User Guide

1-35
Command Set for Windows PowerShell

PowerShell is a command line shell and scripting technology from Microsoft
for enterprise automation. PowerShell is a powerful tool to manage Windows
servers, workstations, and applications. PowerShell increases administrative
control and productivity.

PowerShell features include:

Included with Windows Server 2008
Download PowerShell 1.0 or 2.0 for Server 2003
Uses an interactive command shell
Uses an admin focused scripting language
Easy to adopt, learn and use
Works with existing Microsoft IT infrastructures
Automates repetitive tasks
Ensures complex, error-prone routines are performed consistently

1-36
Storage Center User Interfaces

There are two interfaces that will be utilized during the configuration and
operation of the Storage Center:

Serial Command Line Interface (CLI)
The CLI is used primarily for initial serial number and IP address
configuration

The CLI may also be utilized to support engineering-level
troubleshooting

NOTE: CLI is to be used only under the direction of Copilot
Support

Graphical User Interface (GUI)
After serial number and IP address configuration, the GUI interface
can be accessed via a web browser.

This GUI connection will be used for the configuration, operation,
administration and management of the Storage Center.
1-37
Storage Center Graphical User Interface

The Storage Center Graphical User Interface is utilized after initial serial number and IP
address configuration. The GUI interface is accessed via a web browser. This GUI
connection will be used for the setup, operation, administration and management of
the Storage Center.

The Storage Center GUI requires the use of the Java plug-in. For optimum
performance, use of the latest version of the Java plug-in is recommended. The Java
plug-in can be downloaded from http://www.sun.com/java

The Graphical User Interface also uses the Secure Socket Layer (SSL) technology to
provide advanced client-server security. Any pop-up blocker applications should be
disabled when utilizing the GUI.

To login to the Storage Center, open a web browser and enter the eth0 (Single
Controller) or the management (Dual Controller) IP address.

The default login credentials are as follows:
Default User ID: Admin (case sensitive)
Default Password: mmm (case sensitive)
1-38
Storage Center Version 6.3 Features
The Storage Center Operating System (SCOS) version 6.3 adds new features
including:
Performance improvements
Active Directory and OpenLDAP support
IPv6 support
New Synchronous Replication options
High Availability Mode
High Consistency Mode
Smart Resynchronization minimizes the amount of data replicated when
the remote site becomes available after an interruption
W2012 support
Thin provisioning support (TRIM/UNMAP)
Offload Data Transfer
Support for 16Gb Fibre Channel
Volume Advisor
SNMP Enhancements
Login Security Enhancements include support for customer installed 2048-bit
SSL Certificates
Upgrading: SC040 controllers must be at v5.5.4 or above to update to v6.3.1;
SC8000 controllers must be at v6.0 or above
Requirements:
Web browser: IE 7,8, and 9. IE Desktop 10. Mozilla Firefox Java v6 or later.
1-39
Storage Center Version 6.4 Features
Storage Center version 6.4 adds additional features including:

All Flash systems
On-Demand Data Progression for flash optimization
New Flash Optimized Storage Profile
Supports the new SC280 High Density Enclosures
Enhancements to Enterprise Manager

1-40
All Flash Solution

The all Flash solution is made possible thanks to the introduction of Read
Intensive (RI) SSDs
- these Read Intensive SSDs are often referred to as MLCs.

The MLC SSDs provided the opportunity for Dell Compellent engineers to
design a new Storage Profile that will do automatic tiering across write
intensive (WI) SSDs
- often referred to as SLCs
as well as the MLCs.

That same profile can also be used on traditional spinning disks.
Lets take a look at an SC220 enclosure and what options are available for SSDs.
Customers can start with:
A 6 pack of 400GB, write intensive SSDs
and then add
A 6 pack of 1.6TB, read intensive SSDs.

The final 12 drives can be a combination of 400GB write intensive SSDs, 1.6TB
read intensive SSDs and/or traditional spinning disks (this example shows 1TB
7.2K RPM drives but they can be any type of spinning disk.)
1-41
Flash Optimized Solution

The Flash optimized solution released with version 6.4 takes full advantage of
SSD technology and allows for Storage Center solutions that include both SLC
(Single-Level Cell SSD drives which are write intensive) and MLC (Multi-Level
Cell SSD drives are read intensive).

The solution uses enterprise class SAS SSDs that are dual ported for high
performance and high availability. They also have built-in wear monitoring and
non-volatile write cache to protect data during write operations.

The 1.6TB MLC flash drives are new and are the densest 2.5 flash drives
available.

These flash drives improve performance for data-intensive apps and Online
Transaction Processing (OLTP) workloads.

Dell testing has shown >300K IOPs with low latency as well as >100K IOPs
with sub-millisecond latency running OLTP workloads.
The cost of the system is optimized because of the new Storage Profiles that
are optimized for multiple tiers of flash.

1-42
There are several types of SSDs on the market, each with different attributes designed to
deliver the best performance, cost or endurance. Most of the available flash shared storage
solutions use a single type of SSD, usually a write-intensive (WI) SLC [Single Level Cell] SSD
which has high endurance but lower capacity and a high cost. This is what Dell Compellent
has always offered in the past.
In Compellents innovative approach in which two types of flash drives are deployed in a
single enclosure, flash is tiered across the SLC SSDs and MLC [Multi Level Cell] SSDs, which
have a higher capacity and lower endurance but a considerably lower price, blending the
attributes of these SSDs to achieve a superior $/GB.
Since competitive flash solutions do not separate data reads from data writes, they are not
able to use the read-intensive MLC SSDs and rely only on a single SSD type, eMLC or SLC. As
these systems address all performance needs with a single flash tier, they waste the
expensive capacity on data that is frequently read but not frequently written to, keeping their
$/GB pricing very high.
CML uses enterprise class SSD drives, SLCs and MLCs, for reliability and performance. These
area characteristics of enterprise class SSD drives.
Built-in wear monitoring; SSD Gas Gauge
Dual port SAS
Over-provisioned for endurance and sustained performance
Non-volatile write cache on each drive
Full end-to-end data protection (IOEDC/IOECC)

1-43
Here are a few different options for configuring a flash optimized solution. The
solution on the left shows one SC220 enclosure that combines SLC and RLC
SSDs for an all flash option. For Data Progression purposes, the 12 WI drives
would be in tier 1 and the 12 RI drives would be in tier 2.
This middle option is showing a similar all SSD solution but with two SC220
enclosures. Again, the customer would purchase the drives in packs of six.
The system on the right is the hybrid option. With this option we are adding
traditional, spinning disks to complete the solution. The 1TB, 7200RPM drives
would be in tier 3.
A few things to be aware of with the flash optimized solutions:
1. The flash solution is only available with SC8000 controllers. Series 40
controllers do not support the flash solution.
2. Drives are only sold for the flash solution in bundles of six.
3. At initial release the flash solution is only available for brand new
installations. An enclosure with RLC & SLC drives cant be added to an
existing system.
4. The Storage Center must be running a minimum SCOS of 6.4.
1-44
On demand Data Progression

To take full advantage of the new read intensive MLC SSD drives, Dell
Compellent engineers developed a new approach to data progression referred
to as On-demand data progression. With on-demand data progression data
can be automatically tiered from write intensive SLC SSDs to lower cost read
intensive MLC SSDs.

On-demand data progression is effectively scheduled to run following a
Replay. Whether the replay is created by a Replay profile, by an application
replay (such as Replay Manager) or by a manual replay, pages will be moved
from SLC (Tier 1) to MLC (Tier 2) once the replay is completed for that volume.
These replays are mountable.

On-demand data progression introduces a new type of replay called a Space
Management Replay. A Space Management Replay is created when the SLC
drives (Tier 1) fills to approximately 95% of capacity. The Space Management
Replay is taken and the blocks are then moved from Tier 1 to Tier 2 SLC to
MLC. This will free up space on the SLC drives so those drives have more room
for new writes.
Space Management Replays are NOT mountable.
1-45
To support On Demand Data Progression a new default Storage Profile has
been created. Upon initialization of a new Flash Optimized Storage Center with
SCOS 6.4 there will be five options for Storage Profiles. The options are:
Low Priority (Tier 3)
Flash Optimized with Progression (Tier 1 to All Tiers)
Write Intensive (Tier 1)
Flash Only with Progression (Tier 1 to Tier 2)
Low Priority with Progression (Tier 3 to Tier 2)
These will be defined further in a later section.




1-46
Dell Compellent vSphere Plug-in (VSP)

The Dell Compellent vSphere Client Plug-in allows an administrator to manage
the Storage Center directly from vSphere.
Tasks include:
View system info and statistics
Provision volumes as VMFS datastores or RDMs
Take VM consistent replays
Extend/grow VMFS datastores
Create and manage replications
Deploy new virtual machines

The Dell Compellent vSphere Client Plug-in is available for download via
Knowledge Center http://kc.compellent.com

1-47
vStorage APIs for Array Integration (VAAI)
Uses industry standard T10 spec
Reduces:
CPU utilization
Memory utilization
IO between Host and Array
Time to complete tasks
Complexity when managing thinly provisioned storage
Four Primitives:
Block Zeroing
Copy Offload
Hardware Assisted Locking
Thin Provisioning
Unmap
Stun
1-48
vSphere APIs for Storage Awareness (VASA)

Dell Compellent VASA provider gathers information about the available storage
topologies, capabilities and statuses of the Storage Centers and passes this information
to VMware vCenter, making it accessible to vSphere clients. This information helps
VMware administrators make intelligent decisions when selecting the datastores on
which to place new virtual machines.

The VASA provider supports Profile-Driven Storage, which categorizes volumes by
performance and stores this information in storage profiles. Benefits include:
Rapid, intelligent provisioning of applications
Application service levels that match available storage
Better visibility into the available storage pool

Policy-Based Storage Management helps further provision virtual machines by
automating datastore placement decisions.

Requirements:
Enterprise Manager v6.1.1 or above
Storage Center v5.5 or above
VMware ESX/ESXi v5.0 or above
VMware vCenter Server v4.1 or above
1-49
Section Review
This section discussed the fundamentals of Storage Area Networks as well as
the Storage Center.

Section Objectives:

Storage Center functional overview

Storage Center deployment architecture

Review Storage Center communication links

Review the Storage Center user interfaces

1-51
Review Questions

1. Fluid Data Architecture what information does the Storage Center
track for each block of data?





2. Which Storage Center core function manages all disks as a single pool of
resources from which the administrator can select and assign volumes
to a server/host?





3. Which application provides for the movement of different categories of
data to different types of physical disks?






1-52
Review Questions (cont.)

4. Which core function provides for the ability to allocate space only when data is
physically stored?



5. What is the Dell Compellent licensed module that reserves the outer tracks of
the disk for most active blocks?



6. Which core function provides for the ability to create space-efficient replays
that restore data instantaneously?


1-53
Review Questions (cont.)


7. What type of SSD drive is considered read intensive SLC or MLC?



8. What is the feature in 6.3 called that allows movement of existing volumes from
one Storage Center to another?



9. When an expiration of a replay is compete, what happens to that space?




10. What solution is part of the Storage Center 6.4 release that allows multiple tiers
of SSD drives to work with HDD drives?

Você também pode gostar