IBM PROTECTIER: FROM BACKUP TO RECOVERY

SOLUTION PROFILE

IBM PROTECTIER: FROM BACKUP TO RECOVERY
NOVEMBER 2011

When it comes to backup and recovery, backup performance numbers rule the
roost. It’s understandable really: far more data gets backed up than ever gets
restored, and backup length is one of most difficult problems facing
administrators today. But a reliance on backup numbers alone is dangerous.
Recovery may not happen as frequently as daily backup but recovery is the entire
reason for backup. Backing up because everyone does it isn’t good enough. Backing up for
compliance won’t cut it. Backing up because you’ll get fired otherwise isn’t the point. Because when
a recovery operation looms -- especially if the recovery is large and the potential loss is huge -- fast
recovery performance becomes more urgent than backup ever was.

Now, we know that backup is an important challenge in and of itself. Backup is the foundation for
recovery and slow backup performance can threaten the integrity of the entire backup and
recovery process. Today’s greater data volumes already take a long time to backup which threatens
service levels. IT must address these important issues by demanding acceptable backup
performance. But for all the attention we pay to backup performance, recovery performance is
equally crucial -- yet it remains an afterthought.

Why? Because it is frankly harder to plan for and deploy recovery than it is to install backup.
Recovery is not just its speed but is also its replication options, its bandwidth impact, and its
disaster recovery potential. To make it even more complicated, recovery planning must account for
different service level agreements. It must appropriately provide for every scenario from “we can
wait a week for this application to come up” to “if we lose the last half hour of transactions then we
might as well just go home.” Complicated, yes. But the consequences of not preparing for recovery
are far too high to take the easy way out.
Fortunately there are backup and recovery products out there that can make recovery planning far
easier and much more successful. One such product is IBM ProtecTIER, which has one of the fastest
recovery rates in the business. In this paper we will cover the ProtecTIER deduplication system’s
very fast restore speeds and excellent DR architecture. These capabilities and more make
ProtecTIER an exceptional leader in the crowded backup and recovery market.

Recovery Isn’t Just a Good Idea
Disaster recovery (DR) is the process of returning to normal operational levels following a serious
loss of applications and data. The reasons behind disasters come in lots of different flavors
including physical disaster, external hacking, administrator errors, computer-caused data loss or
corruption, and even internal malfeasance (fancy way of saying disgruntled employee revenge). In

Copyright The TANEJA Group, Inc. 2011. All Rights Reserved. 1 of 6
87 Elm Street, Suite 900  Hopkinton, MA 01748  T: 508.435.2556  F: 508.435.2557 www.tanejagroup.com

Solution Profile

all of these cases DR is accomplished by restoring applications and data to acceptable usage levels
without serious negative consequences to the business.

It’s a hard job. The modern data center struggles with hugely growing volumes of data that threaten
data protection service levels. When disaster strikes, the sheer enormity of dealing with this volume
of data (not to mention applications, systems, users and networks) can be overwhelming. Yet in
spite of the urgency of recovery, IT often concentrates on backup performance when evaluating
backup and recovery technology. Recovery planning is an afterthought and may be relegated to the
DR document that is gathering virtual dust on a network share.

IT can handle quick and dirty file restores and the occasional volume recovery from backup, no
problem. But then the Big One hits: the critical database goes down and stays down; the hurricane
hits the central data center; the earthquake knocks the server rack to smithereens and takes
attached storage with it. These things happen, and IT must be able to restore applications and data -
- and they have got to restore them fast.

RECOVERY READINESS
There are a set of requirements that any company serious about recovery should meet. They
include:

 Recovery performance. Recovery speed is the simplest of the recovery metrics and is crucial
to timely restoration. Tape-based volume recovery can be very slow and combing backup
catalogs for individual files is no picnic either. Virtual Tape Libraries (VTL) let users use disk-
based recovery without making massive changes to their backup infrastructure. Adding
deduplication deeply compresses the backup volume and is a basic method for speeding up
disk-based backup and recovery.

 Recovery Time and Point Objectives. A recovery solution must address both Recovery Time
Objective (RTO) and Recovery Point Objective (RPO). RTO is the maximum amount of time that
a given application can be unavailable without damaging the business. Under 12 hours is a
common RTO for priority applications and truly crucial applications may be set as low as a zero-
hour RTO. RPO is the maximum allowable time that the replicated copy may lag behind the
primary data. A period of 24 hours is common given nightly backup schedules, but in the case
of critical data 12 hours down to zero data loss may be the only acceptable RPO.

 Replication. DR requires replicating data to a remote site for fast restoration to the primary
site or for host failover. Replication is fast and bandwidth-efficient, protects data integrity, and
enables quick virtual cartridge recovery at the secondary DR site.

 Bandwidth. Companies with remote DR sites need wide bandwidth but must control costs.
Replication operations should optimize bandwidth to write and restore the replicated data
within service level guidelines, at a price that won’t break a reasonable budget.

 Failover and failback. Failover/failback is the process of stopping replication from the
primary site and switching operations over to the remote site. Once the primary site is up
failback operations transfers processing and data back to the recovered production site. The
more automated the process the better, since minutes matter when critical applications become
unavailable. Look for recovery platforms that help to automate this process using policies.


Solution Profile

 DR testing. Disaster recovery can be difficult to test over time. There are multiple points of
failure to consider: LAN and WAN network connections, communications, secondary site status,
operational plans, automation triggers, and more. No one technology can test everything and
some concerns such as telephone availability will be out of IT’s hands. But data protection
technologies should offer DR self-testing abilities, which removes some of the onus from
overworked IT.

IBM ProtecTIER and Recovery Readiness
One of the best ways we know to accomplish these levels of recovery readiness is IBM’s ProtecTIER
deduplication system. ProtecTIER offers essential data protection with deduplicated backup and
very fast recovery rates. Features include powerful and flexible replication, highly automated
failover and failback, bandwidth optimization and DR testing capabilities. These features work
together to make ProtecTIER an excellent choice for backup and also for recovery.

ProtecTIER builds the foundation for recovery by dramatically reducing backup time: it replaces
tape with high performance backup disk, and deduplicates incoming data for big capacity and time
savings. Processing is extremely fast using an efficient index that resides in memory.
Recovery speeds are exceptionally fast at an up to sustained 2800 MB/sec (10 TB/hr) or above,
which lets it restore data very quickly when needed. The same rate of speed operates over the WAN
with replication and bandwidth optimization. Since ProtecTIER replicates only unique and
deduplicated data, it relieves the intensive load on the wide area network, and also provides
bandwidth control options for administrators. Replication can occur during its own time-window
or simultaneously with backup, which allows IT to fully protect the deduplicated data while it
enters primary ProtecTIER storage. ProtecTIER replication operates in one-to-one, many-to-one
and many-to-many modes to provide complete flexibility for data protection and disaster recovery.

Let’s take a closer look at ProtecTIER and the recovery enablers we mentioned above: recovery
speed, recovery time and point objectives, replication and bandwidth, failover/failback and DR
testing. It is this combination of capabilities that provides tremendous recovery advantages to
ProtecTIER users.

EXCEPTIONALLY FAST RECOVERY PERFORMANCE
ProtecTIER’s restore performance is even faster than their already fast backup speeds of
sustainable 2000-plus MB/sec. Recovery speeds hit 2800 MB/sec (10 TB/hr) and higher
sustained recovery performance. ProtecTIER’s architecture enables this exceptionally fast
performance by only storing unique, deduplicated data. When it comes time to recover, data
restores efficiently and quickly.

This is very good news, especially for Tier 1 applications like databases that require
immediate or near-immediate recovery. For example, ProtecTIER is a popular choice with
SAP administrators who back up large databases twice daily because they cannot afford
data unavailability or corruption. ProtecTIER’s fast backup and recovery speeds greatly
benefit data protection and enable quick recovery of mission-critical applications.

MEETING RPO AND RTO
ProtecTIER fulfills RPO and RTO for even the most demanding recovery requirements.
Replication schedules may be set to the proper service agreement level for any given


Solution Profile

application, which may range from immediate RPO or RTO to hours or even days depending
on the application priority.

RPO: A less critical application might be all right with restoring data from the point of the
most recent backup, which might have run a maximum 24 hours ago with a single daily
backup or a maximum of 12 hours ago
with a twice-daily backup. But Tier 1
applications may require RPOs within a
few minutes. Zero-loss RPO scenarios ProtecTIER Customers
should be fully mirrored to redundant
systems that take over immediately #1: FROM TAPE TO PROTECTIER DISK
should there be an interruption in This company had backed up files to tape
processing. For these circumstances, for many years. One large recovery effort
ProtecTIER supports running replication involved 1 million files stored on 10GB of
simultaneously with backup. tape. The restoration took over 17 hours to
complete.
RTO: Using ProtecTIER to restore from
disk instead of tape automatically speeds The company quickly made the switch to
up the recovery process to the tune of a disk-based deduplication backup using IBM
few hours instead of several days. As ProtecTIER. They expected a faster recovery
with RPO, one size does not fit all and IT time but they did not expect the extreme
needs to assign priority to application recovery speed of their next restoration
requirements. ProtecTIER enables fast project, which was even larger than the
and flexible recovery options for differing first. This time they had to restore 1.6
service level agreements. For example, IT million files stored on 34GB on their
can use ProtecTIER to create fully ProtecTIER system. The grand total of
redundant backup and restore systems recovery time? 1 hour and 38 minutes; a far
with immediate failover and up-to-the- cry from the previous tape-based project.
second data processing.
#2: TRIPLED DATA GROWTH
POWERFUL REPLICATION An IBM customer’s data had tripled in just a
ProtecTIER offers integrated IP-based few years. Over a petabyte of this data was
replication with the flexibility of one-to- contained in mission-critical databases.
one, many-to-one and many-to-many Over the years frequent tape-based backup
choices. (Many-to-one copies data from had resulted in over 7 PB stored on physical
multiple source repositories to a single tape. Backup to tape was taking 15 hours a
destination repository; many-to-many day and recovery from this volume size was
provides bi-directional replication extraordinarily challenging.
between 2-4 systems in a replication
grid.) All of these replication options The company adopted IBM ProtecTIER
allow IT to optimize backup data gateway clusters. In combination with TSM
protection between data centers, DR sites they experienced much faster backup time
and remote offices. with better disk space reclamation and a
smaller physical footprint. And recovery?
Users may schedule replication by using Recovery time for their crucial Oracle
either preset times or concurrent to database applications was slashed by more
backup and deduplication. They may also than 50%.
choose to manually launch the process.
Best practice is to set scheduled

Solution Profile

replication options in ProtecTIER’s policy engine so it can automatically identify data status
and priority in the replication queue. ProtecTIER’s fast replication -- up to 128 cartridges
simultaneously, each containing 32 concurrent data streams – provides for fast replication
and fast recovery even over the WAN (256 cartridges and 64 concurrent data streams with
a ProtecTIER dual-node cluster).

OPTIMIZED BANDWIDTH
ProtecTIER preserves replication bandwidth by only replicating deduplicated data that is
new and unique. It also provides optional bandwidth management features that allows IT to
support the maximum replication transfer rate allowed for a specific repository. This
capability reduces bandwidth needs by 90% and more compared to uncompressed,
undeduplicated data transfer.

This results in huge bandwidth cost savings, which allow users to protect all of their
applications and not just a few chosen business-critical ones. Administrators can afford the
level of bandwidth that they really need for restoring data between remote sites. Optimized
bandwidth is crucial for this level of disaster recovery, where a remote host must restore
critical data to a recently recovered primary site.

AUTOMATED FAILOVER/FAILBACK
Failover and failback events require extremely high recovery performance. This requires
recovery speed at and between multiple sites in several situations: 1) restoring data from a
replication site to an operational primary site, 2) during failover at a secondary site when
restoring backup cartridges to the new host and 3) during failback when the secondary host
restores data back to the restored primary host.
These scenarios all require high levels of preparation including secondary DR sites, creating
ProtecTIER policies, and establishing replication procedures. But once the prep work is
done, ProtecTIER will quickly accomplish failovers, failbacks and data restores as needed to
get critical applications back up and running on time.

NATIVE DISASTER TESTING
ProtecTIER provides disaster recovery testing capabilities for the replicated repository at
the DR site, which makes IT’s job much easier. For example, IT might test DR settings by
starting the remote ProtecTIER host and running a typical subset of deduplicated backup
jobs. Production volumes will remain in read-only mode to prevent errors during testing.
ProtecTIER also provides a command line interface (CLI) to allow many tools and
capabilities to the user including checking for uncompleted replication jobs.

Taneja Group Opinion
Backup performance is relatively straightforward and it is simple to pay attention to the numbers
surrounding it. Yet because recovery is more complex and more infrequent than backup, it goes
begging far too often. But recovery is backup’s end game and poor recovery practices result in
sharp revenue losses, lost productivity, failed regulatory compliance, and unrecoverable critical
data.
Before this happens to you, look to backup and recovery systems that offer deduplication, high
backup performance AND high recovery performance. Make sure that those high recovery speeds


Solution Profile

work locally for restoration data at the primary site and remotely for big disaster recovery. Then
add the really hard questions: Does the system have native replication that I can customize to my
needs? Does it optimize bandwidth so I can afford the level of DR protection I need? Can it
automatically failover to a secondary site and then failback, and can it do these things fast?

When the answer to all of these questions is an unqualified Yes – as it is with IBM ProtecTIER – then
we strongly suggest that you look very carefully at this platform for your backup, recovery and
complete disaster protection needs.

This document was developed with IBM funding. Although the document may utilize publicly available material
from various vendors, including IBM, it does not necessarily reflect the positions of such vendors on the issues
addressed in this document.

The information contained herein has been obtained from sources that we believe to be accurate and reliable, and
includes personal opinions that are subject to change without notice. Taneja Group disclaims all warranties as to
the accuracy of such information and assumes no responsibility or liability for errors or for your use of, or reliance
upon, such information. Brands and product names referenced may be trademarks of their respective owners.


IBM PROTECTIER: FROM BACKUP TO RECOVERY

More Related Content

What's hot (18)

Similar to IBM PROTECTIER: FROM BACKUP TO RECOVERY (20)

More from IBM India Smarter Computing (20)

Recently uploaded (20)

IBM PROTECTIER: FROM BACKUP TO RECOVERY