Server

Virtualisation
And
Storage
Management

Optimising Storage Management to Obtain

Greater Benefits from Server Virtualisation

Alan McSweeney
 ServerVirtualisation and Storage Management

Contents

Server Virtualisation ........................................................................................3

Server Virtualisation Architecture....................................................................4
Configuration ...............................................................................................4
Virtual Server Resilience ..............................................................................4
Virtual Server Data Backup and Recovery ...................................................5
Storage Management and Server Virtualisation................................................6
Virtual Server Backup and Recovery............................................................8
IBM N Series SnapMirror .............................................................................9
IBM N Series FlexVol and FlexClone............................................................9
IBM N Series RAID Double Parity ..............................................................9
Conclusion.................................................................................................. 10

Page 2
 Server Virtualisation and Storage Management

Server Virtualisation

Implementing server virtualisation (using software such as VMware) can deliver

substantial benefits to IT and business by reducing the number of deployed
Intel-based servers by 50% or more while delivering benefits such as:

• Improved server utilisation rates - 60% to 80% (up from typical rates of less
than 15%)
• Server and application provisioning time measured in hours not days
• Reduction in downtime such as hardware maintenance Once confined to server
• Reduction in operating costs rooms, virtualization has
• Increased server and application resilience now made it into the
• Simplification of server operation and administration mainstream of computing.
• Increased levels of service And by 2010,
virtualization will be the
Implementing server virtualisation also provides a platform for: single most important
technology for IT
• Facilitating business continuity and disaster recovery departments, according to
• Improved backup and recovery research firm Gartner.

A little additional planning and investment will deliver significant additional

benefits for a virtualised server environment in terms of:

• Faster backup and recovery

• Protection against application corruption
• Improved business continuity and disaster recovery

Implementing server virtualisation requires some form of shared storage to

enable features such as HA (High Availability) and DRS (Dynamic Resource
Scheduling).

In order to implement an effective, efficient, resilient and performing virtualised

server infrastructure, the storage component needs to be designed correctly.
This can include taking advantage of features in a virtualised server
infrastructure such as snapshotting and virtual backup.

This whitepaper discusses some of the issues regarding the configuration of

storage for server virtualisation to derive a more complete set of benefits.

This whitepaper makes reference to one storage subsystem from NetApp and
also sold by IBM as the N Series.

Page 3
 ServerVirtualisation and Storage Management

Server Virtualisation Architecture

Configuration

Optimising the benefits of server virtualisation will involve implementing a

virtualisation infrastructure along the following lines:

• 80% of businesses
have no plan
o “It won’t
happen to
me”
• 68% of businesses
who experience a
disaster and don’t
have a plan go out
of business within 2
years
• One in five
organisations will
suffer a major IT
disaster in five
years
• A company
experiencing a There will be two (or more) sets of physical servers onto which servers are
computer outage virtualised. These can be configured to operate as active/active or active/passive.
lasting longer than
10 days will never The physical servers will be connected to shared storage such as a fibre-channel
fully recover or iSCSI SAN.

The I/O load that was distributed across a large number of separate physical
servers is now concentrated onto a smaller number of servers hosting the virtual
servers.

Virtual Server Resilience

A configuration such as this provides multiple points of resilience

Page 4
 Server Virtualisation and Storage Management

In the event of the failure of a physical server within the virtual infrastructure
the virtual servers can be restarted on either another server within the same
infrastructure or to a server on the other infrastructure.

Virtual Server Data Backup and Recovery

There are two options for backing-up virtual servers within a virtual
infrastructure using software-based backups:

1. The existing process for backing-up physical servers can be implemented in

the virtual environment – backup agents for your existing backup tool can
be installed on the virtual servers and these servers backed-up as normal.

2. Using VCB (Virtual Consolidated Backup) to capture server images – the

entire virtual server can be backed-up to a separate Proxy Server that can
then backup the images using existing backup software. A backup agent for
your existing backup software is installed on the VCB server. This reduces
backup software costs but decreases the granularity of what can be
recovered.

Standard Server Backup Virtual Consolidated Backup

Page 5
 ServerVirtualisation and Storage Management

Storage Management and Server Virtualisation

In a virtualised server environment, storage and its management assumes a

greater importance that for normal server configurations:

• Shared storage in the form of a SAN is required to take advantage of

advanced virtualisation features such as HA (High Availability) and DRS
(Dynamic Resource Scheduling)

• All virtual server I/O will be concentrated onto the SAN

• A resilient storage subsystem will be required to ensure that the availability

of the virtual server environment.

The IBM N Series storage subsystem provides hardware and software that
provides significant advantages when used in a virtualised server environment:

• IBM N Series FlexVol technology enables you to create multiple flexible

volumes on a large pool of disks, aggregate I/O across all physical disks on
NFS, iSCSI or Fibre Channel. You can dynamically allocate storage and
add storage capacity as needed without downtime.

• IBM N Series FlexClone provides instant clones of VMs without impacting

server resources and requiring additional storage capacity. This allows you
to take copies of production virtual servers for testing, validation and
development

Page 6
 Server Virtualisation and Storage Management

• IBM N Series Snapshot allows VMware users to make fast and space-
efficient backup copies of active data with negligible performance and
storage overhead.

• IBM N Series SnapMirror protects data and accelerates recovery, by

providing a simple, efficient way to replicate data between systems in
multiple locations using multi-transport (IP and FC) and frequency choice
(synchronous, semi-synchronous, and asynchronous) options while only
transferring changed blocks.

• IBM N Series Raid-DP (Double Parity) offers significantly more protection

than single parity schemes (including RAID 5 and RAID 4), with zero to
minimal cost and performance impact. Compared to RAID 1 (mirroring),
RAID-DP offers equivalent data protection against double disk failure at a
fraction of the cost of mirroring. Compared to RAID 6, RAID-DP offers
better performance and ease of use.

The IBM N Series storage system provides a unified storage solution that
incorporates data management, snapshots, real-time backup and disk-to-disk
server backup. This simplifies the management of the vital storage component
of a virtual infrastructure. The IBM N Series is ideally suited to a virtual server
environment.

Of those companies
participating in the 2001
Cost of Downtime Survey:

• 46% said each hour

of downtime would
cost their companies
up to $50k
• 28% said each hour
would cost between
$51K and $250K
• 18% said each hour
would cost between
$251K and $1
million
• 8% said it would cost
their companies more
than $1million per
hour.

Page 7
 ServerVirtualisation and Storage Management

Virtual Server Backup and Recovery

Backing-up and ensuring that the backups are recoverable for large numbers of
virtual servers can be a difficult task. The task can be more difficult because of
the need to ensure that backup I/O does not affect production I/O resources.

Multiple modes of backup – mirroring, point in time snapshots, offsite point in

time snapshots – can be extended to all virtual servers and ensure recovery.

Existing Server Backup IBM N Series Comprehensive Server

Backup Model

At what point is the Using the IBM N Series in a virtual server environment enables you to
survival of your implement a variety of backup and recovery approaches and techniques across
company at risk? the entire virtual infrastructure seamlessly using one common interface and tool.
• 40% said 72 Because backups are disk-based, they are directly recoverable.
hours
• 21% said 48 IBM N Series Snapshots do not result in any data movement and have any
hours impact on I/O, unlike the type of Snapshots available for other storage systems.
• 15% said 24 Therefore they are ideally suited for use in a virtual server environment.
hours
• 8% said 8 hours Industry Standard Snapshot Model IBM N Series Snapshot Model
9% said 4 hours
• 3% said 1 hour
• 4% said within
the hour
Cost of Downtime
Survey Results,
2001

SnapVault software can be used to move SnapShots to a separate storage

location to act as a disk-to-disk backup.

Snapshots can be restored using SnapRestore software.

The N Series solution can support thousands of snapshots. Unlike other storage
subsystems, the N Series does not suffer any performance degradation. The

Page 8
 Server Virtualisation and Storage Management

performance impact of snapshots using other hardware platforms can involve a

60% or greater reduction in I/O performance.

With the IBM N Series, multiple separate snapshot schedules or policies can be
maintained, such as hourly snapshots retained for several days, daily snapshots
retained for several weeks, weekly snapshots retained for several months and so
on.

IBM N Series SnapMirror

SnapMirror provides simple, flexible, and cost-effective replication software for

disaster recovery and data distribution.

SnapMirror can replicate virtual servers from one virtual infrastructure to

another to facilitate very rapid recovery and business continuity.

Exceptionally powerful, yet easy to implement and manage, IBM N Series

SnapMirror software combines disaster recovery and data distribution in a
streamlined solution that supports today's global enterprises. SnapMirror is a
very cost-effective solution with efficient storage and network bandwidth
utilisation, and provides additional value by enabling you to put the DR site to
active business use.

SnapMirror offers a straightforward set-up that you can easily replicate across
all your IBM N Series storage systems. Once installed, SnapMirror requires
minimal management, and SnapManager software ensures effortless replication
of application-consistent snapshots.

SnapMirror can mirror data from FC storage to ATA storage, reducing storage
costs. To reduce network usage, SnapMirror works with Snapshots to send only
changed data blocks to the disaster recovery storage.

SnapMirror enables space-efficient copies to be created on the disaster recovery

storage for other uses such as testing, development, and quality assurance,
without affecting the production system. SnapMirror also enables centralised
backup of data to tape from multiple data centres, reducing investment in tape
infrastructure as well as offloading the production system from tape backups.

IBM N Series FlexVol and FlexClone

FlexVol volumes are virtual dynamic storage entities that can grow and shrink
without being constrained by the underlying physical storage.

FlexClone uses ShapShot technology to create writable clones of FlexVol

volumes. Clones can be created nearly instantaneously and with little I/O and
storage overhead. FlexClone technology can be used to create copies of
production systems for testing, validation and development.

IBM N Series RAID Double Parity

IBM Storage System N series introduces double-parity RAID, named RAID-

DP. This section provides an overview of RAID-DP and how it dramatically
increases data fault tolerance from various disk failure scenarios. Other key

Page 9
 ServerVirtualisation and Storage Management

areas covered include how much RAID-DP costs, special hardware

requirements, and converting from existing RAID4-based volumes to RAID-
DP. This document presents a double-disk failure recovery scenario to show
how RAID-DP both allows the volume to continue serving data and recreates
data lost on the two failed disks.

Traditional single-parity RAID offers adequate protection against a single

event, which could be either a complete disk failure or a bit error during a read.
In either event, data is recreated using both parity and data remaining on
unaffected disks in the array or volume. If the event is a read error, then
recreating data happens almost instantaneously, and the array or volume
remains in an online mode. However, if a disk fails, then all data lost on it has
to be recreated, and the array or volume will remain in a vulnerable degraded
mode until data has been reconstructed onto a spare disk. It is in degraded
mode that traditional single-parity RAID shows its protection capabilities have
not kept up with modern disk architectures. RAID Double Parity significantly
increases the fault tolerance from failed disk drives over traditional RAID.

IBM System Storage N series uses a new type of RAID protection named
RAID-DP. RAID-DP stands for RAID Double Parity, and it significantly
increases the fault tolerance from failed disk drives over traditional RAID.
When all relevant numbers are plugged into the standard mean time to data
loss (MTTDL) formula for RAID-DP versus single-parity RAID, RAID-DP is
more reliable on the same underlying disk drives. With this reliability, RAID-
DP approaches RAID1 mirroring for fault tolerance, but at RAID4 pricing.
RAID-DP offers businesses the most compelling total cost of ownership storage
option without putting their data at an increased risk.

Conclusion

An effective storage platform is essential to the successful implementation and

operation of a virtualised server infrastructure.

Server virtualisation requires high-performing central storage. With virtualised

servers data is moved from local to central storage. The design and operation of
the storage component becomes of critical importance. This is particularly true
for organisations wishing to take advantage of high availability, centralised
backup and business continuity and disaster recovery features of server
virtualisation.

IBM storage is ideally suited for use in a virtualised server environment because

High Performance IBM storage solutions offer consistent high performance

with no performance degradation when taking system
snapshot.

Instantaneous Backups With Snapshot technology, you can make instantaneous,

consistent backup copies of active data with negligible

Streamline Storage VMware administrators can supply storage to virtual

Allocation to machines in seconds while achieving the highest level of
Virtual Machines storage utilisation.

Page 10
 Server Virtualisation and Storage Management

Thin Provisioning IBM storage systems offer powerful, flexible provisioning

capabilities that give IT storage specialists the ability to
lower cost while meeting capacity, utilisation, and
performance requirements.

Raid-DP RAID-DP offers significantly more protection than single

parity schemes (including RAID 5 and RAID 4), with zero
to minimal cost and performance impact. Compared to
RAID 1 (mirroring), RAID-DP offers equivalent data
protection against double disk failure at a fraction of the
cost of mirroring. Compared to RAID 6, RAID-DP offers
better performance and ease of use.

Disaster Recovery You can protect data and speed recovery using SnapMirror.
SnapMirror provides a simple, efficient way to replicate
data between systems in multiple locations using multi-
transport (IP and FC) and frequency choice (sync, semi-
sync, and async) options while transferring only changed
blocks. In the event of a disaster, a remote site can use the
SnapMirror copy of the data to restart critical virtual
servers with minimal disruption to ongoing operations.

Flexibility IBM storage solutions give you the ultimate in flexibility by

enabling you to choose iSCSI, FCP, or NFS connectivity
within a single storage platform. IBM systems can also mix
drive types within a single system so that you can tier your
virtual machines (VMs) within a single architecture, giving
users flexibility and investment protection.

Efficient Cloning of Using the FlexClone tool, you can instantly create VMs
VMs without consuming any additional physical storage. These
clones can be used for production and test and development
environments.

Page 11
ServerVirtualisation and Storage Management

For more information, please contact:

alan@alanmcsweeney.com

Page 12