Taneja Group Technology Analysts
Summary Findings
For disk to disk backup (D2D) applications, Taneja Group finds that
the Pillar Axiom storage system from Pillar Data Systems
constitutes a differentiated storage platform, largely due to its
ability to combine primary and secondary storage into a common
footprint. We believe that this differentiation leads to distinct
advantages in scalability, performance, reliability, and customer
flexibility.
Key Disk Backup Considerations and Results for the Pillar
Axiom
Key
Consideration
1. How will this D2D offering scale and
adapt over time?
2. How will this D2D offering impact my backup and recovery
operations?
3. How does this solution justify itself in economic terms?
4. How will this D2D offering integrate with our existing
infrastructure?
Pillar Axiom 500
Pillar Axiom scales to 307
RAW TBs; modular design allows flexible growth of capacity and
performance.
Pillar Axiom’s ability to reach wire speed without introducing
bottlenecks ensures faster backup operations; its HA design ensures
high level of recoverability.
Pillar Axiom’s ability to host primary and secondary storage on
one platform ensures acquisition and management savings.
Adding a Pillar Axiom can be painless due to its zero
footprint.
Introduction
Based on recent Taneja Group
surveys and research of the data protection community, we see Disk
to Disk (D2D) backup solutions both augmenting and replacing
tape-based solutions in companies of all sizes, across all industry
verticals. The data storage vendor community has responded to
this
strong customer demand for improved data protection with a
bewildering range of disk-based solutions.
If you are an enterprise purchaser of D2D technologies, we highly
recommend developing a group of selection criteria questions that
can be put to each vendor under consideration.
Four Key D2D Evaluation Questions
Just four
questions can greatly streamline your D2D evaluation process and
make sure you find the appropriate disk-based data protection
offering for your environment. As a result of over 100 end user
conversations and surveys of more than 700 IT professionals
conducted over the past 18 months, Taneja Group has identified the
following four questions that you, the IT professional, should pose
to every vendor selling a D2D offering:
1. How will this D2D offering scale and adapt over time?
2. What is the impact of this D2D offering on my backup and
recovery operations?
3. What is the economic rationale for this D2D offering?
4. How will this D2D offering integrate with existing
infrastructure?
We will discuss these questions in detail, and then examine a
particular storage system and D2D vendor, Pillar Data Systems, in
context of these evaluation criteria.
Question 1: How will this D2D offering scale and integrate over
time?
As the result of end user interviews conducted with enterprise D2D
end users in 2005, Taneja Group has identified “Scaling and
Integration” as one of the most critically overlooked selection
criteria.
What we find most interesting is that many end users do not
adequately model for scalability limitations in their new D2D
deployments before actually deploying these technologies. According
to one IT Director at a leading US-based manufacturing firm that
recently deployed a virtual tape D2D
appliance, “We significantly underestimated how quickly we would
exceed the capacity limitations of our initial deployment. We’re
going back to buy another appliance two quarters ahead of
plan.”
While comments like that one may be good news for D2D appliance
vendors, its bad news for IT managers who have to live with higher
than expected capacity growth in their disk backup environments.
The reason for this state of affairs is simple: Because D2D is a
new technology, many enterprises have not developed accurate
capacity planning strategies for this new “disk tier” in their
storage architecture.
One of the little-noted culprits adversely impacting smooth
capacity growth in D2D environments is something we call
“retention creep.” Taneja Group finds that many enterprises,
once familiar with their initial D2D deployment, seek to extend
rapid recovery for mission critical applications by
retaining several more weeks or months worth of data on “hot
disk.” While this is a fantastic benefit of the entire D2D
paradigm, it opens the door to potential capacity issues for a
large number of solutions that were optimized for far shorter
retention periods. Even if you establish strict retention policies
on a per-application basis, you can find yourself rapidly exceeding
your highest capacity and scaling estimates for D2D.
Related to scalability is the ease of integration of the D2D
solution. We find that the best D2D solutions provide the end user
with maximum flexibility to support heterogeneous backup
applications, and avoid locking them into a single access
methodology (e.g. supporting diverse interfaces such as virtual
tape, object model, file, disk volumes, etc.). Enterprises should
understand whether or not a prospective D2D solution requires
changes to their production server environments in the form of
host-based agents. In most cases, Taneja Group finds that
satisfaction rates with D2D solutions are highest when the
integration of the solution was virtually seamless, requiring
little to no alteration of the existing production environment or
backup methodologies.
Another issue of scalability is the ability to scale performance.
As the number of disks grows in storage systems, performance
bottlenecks can begin to emerge. It is important that as capacity
scales, system performance on a per-disk basis scales along with
it.
Follow-Up Questions
To address these issues
of scalability and adaptability, Taneja Group suggests these
follow-up questions when discussing scalability and integration
issues with D2D vendors:
• Is the vendor selling a D2D appliance? If so, what is their
maximum capacity per device and does that suit our capacity plans?
How do they address scaling across multiple appliances?
• How modular is the offering in the addition of new capacity
increments?
• What retention periods can we realistically support with this
offering? Is that period adequate or overkill?
• Does scaling this D2D solution require any additional
investment in clustering or replication software?
• Does this solution lock us into one access methodology? If we
want to add or change in the future, does this pose new integration
issues?
Question 2: What is the impact of this D2D offering on my backup
and recovery operations?
While most IT managers know that D2D data protection will bring
improvements over tape, the devil is always in the details.
Fortunately, in the D2D world, we are now amassing ample evidence
to chart the benefits of this technology in a comparative
fashion.
Over a 24-month period, Taneja Group has conducted enterprise
end-user surveys in the areas of disk-based data protection; the
top improvements resulting from a D2D data protection deployment
are listed as follows:
Top Perceived Benefits of D2D Data Protection
1. Resolve our backup window challenges
2. Enhance our recoverability
Because we find these two top benefits are so pervasive across
virtually every potential D2D purchase, we believe they are worth
exploring in more detail.
1. D2D and Backup Windows: There is no question that a D2D offering
can radically impact backup window challenges in a positive manner.
In the vast majority of end user scenarios, the amount of time
required to conduct a backup job can be cut by at least 50 percent,
sometimes more, depending on
the quality of the D2D solution, the workload, and backup
methodologies.
Real Customer Example:
A medium-sized retailer expanded business hours to include
weekends, effectively losing the backup window they once enjoyed.
Upon deploying a D2D solution in 2004, they reduced their backup
job time to 1/5th the duration it had been in the tape world,
providing a very comfortable margin vs. available window. This
customer reported disk speeds of 80 MB/sec, close to five times the
speed they were achieving with tape.
Contrary to popular conception, it is not simply the “fast”
disk media itself that is responsible for the increase in backup
job speed. Because D2D solutions support multiple streams of data
writes to disk media, they operate entire backup workloads in
parallel. This capability is one
of D2D backup’s primary advantages over traditional tape
solutions. Virtual Tape software has emerged as one of the
principal interfaces by which this type of streaming takes place in
a D2D environment.
Follow-Up Questions
Consider using these
follow-up questions for prospective D2D vendors in order to assess
their potential impact on backup windows:
• What kinds of performance can we expect from your solution,
given our workload profile?
• Will we have to alter our backup methodologies to achieve your
stated performance gains?
• How does your D2D solution enable streaming, to what degree,
and what will that mean for our environment?
2. D2D and Enhanced Recovery: Along with increasing the speed of
backup jobs, establishing an immediate recovery capability is
ranked nearly equally as one of the top benefits of D2D technology.
Taneja Group finds that IT managers often report disk-based
recovery speeds of three
to five times those they achieved with tape-based architectures.
However, enhanced recovery means more than mere speed gains. Users
with D2D solutions also report to us that they achieve close to 100
percent successful recovery operations, compared to 50 to 80
percent successful recovery from
tape-based solutions. In tape environments, one bad tape can
destroy the integrity of the entire restore process.
Additionally, reduced media management issues in a disk-based
environment means that IT managers can quickly retrieve individual
user data with great ease, an activity that can become
prohibitively time consuming in a tape-based world.
Real Customer Example:
A large marketing services firm discovered through a 2003 auditing
exercise that an unacceptable portion of their data was not
recoverable from tape. By deploying D2D technologies for mission
critical applications, they passed their 2005 audit with 100
percent demonstrable recovery of the formerly exposed data
sets.
Follow Up Questions
Consider putting these
follow-up questions to your prospective D2D vendors in order to
assess the offerings impact on your recovery operations:
• What kind of reliability and data integrity checking does your
D2D solution provide?
• Based on the performance metrics of your D2D solution and our
recovery profile, what speeds can we expect?
• Does your D2D solution provide equal ease of use for both
incremental and full recovery of data?
Question 3: How does this solution justify itself in economic
terms?
Because a D2D solution represents a significant and critical
investment for the data protection environment, Taneja Group finds
that the majority of end users apply some manner of formalized ROI
analysis to their purchase.
From an ROI perspective, any D2D solution can be measured against
the variables of Technology, People and Processes.
Technology
Beginning with the new disk-based
backup technology under consideration, you should walk through a
detailed financial analysis of what the offering will mean in terms
of capital acquisition and deployment impact. In this regard, users
should look carefully at the incremental hardware and software
costs that will be associated with the offering over its useful
life (e.g., three to five years). Will the usage model for the
potential disk technology likely create more or less investment in
your existing tape media? What about backup software licensing?
Next, we encourage users to look at the solution’s operating
expense, factoring in a range of factors such as floor space, power
consumption, and field replacement costs. Taken together, these
technology factors can provide a better view into the costs and
benefits of the technology in the context of your deployed
infrastructure.
People
The next category, often overlooked,
is people. Quite simply, what will be the impact of this D2D
offering on your data protection team? Based on the solution’s
capabilities and intended usage model, will it create the need for
more headcount or less? Is there training involved in the offering,
and if so, how disruptive, extensive and ongoing will that training
be? What kind of estimates can be made regarding the productivity
of employees with this solution deployed? And finally, what are the
additional support requirements for this offering from a team
perspective? Is this yet another storage
“silo” that the IT team will have to install, manage, and
upgrade over time? Will they be additive or synergistic versus
existing technologies already deployed in the data center? All of
these human factors have costs and trade-offs that will determine
if the given solution will truly add net value to
the organization.
Process
Lastly, and perhaps more
importantly, we ask you to examine what the overall impact of the
solution will mean from an IT process perspective. Will the
solution require a disruptive shift in methodologies or support
your existing methodologies? Given best estimates based on the
environment and the
solution’s performance profile, what will be the impact on backup
and restore speeds? What is the general economic impact of these
improvements on the IT workflow? And finally, what is the
reliability impact of the potential technology once deployed? Will
it increase overall uptime and reduce IT touches or require
additional human interaction to ensure uptime?
All of these factors of Technology, People and Process considered
together constitute a general snapshot of how you should approach
the ROI equation for your disk-based data protection
deployment.
Question 4: How will this D2D offering integrate with our existing
infrastructure?
The last critical question for all vendors under consideration is
how the D2D offering will integrate with the infrastructure already
in place. Will it be easily deployed and easy to manage? What about
change management issues over time? We often find that the largest
challenges you will face in your
first D2D deployment will result from unforeseen integration issues
between various IT elements.
On this note, it must be said that not all D2D technologies are
created equal with regards to ease of integration and management.
Taneja Group often finds that too many end users do not perform
adequate diligence on D2D offerings before making their first
purchase. This inattention to offering
details can lead to backup-management problems later, and we
therefore encourage a thorough evaluation of the entire available
range of D2D options.
Follow-Up Questions
Some guiding questions
for prospective D2D vendors regarding backup management
include:
• What access methods are supported by this solution (e.g.
virtual tape, volume, file) and can we change between them
easily?
• Does this offering provide easy UI integration for recovery
operations?
• Does this offering support easy, bi-directional data flows
between disk and tape for archival and retrieval of content?
• Does this D2D solution help or hinder our existing change
management practices?
• If the solution is an appliance, what does it mean for our
network?
• If the solution is software, how will it integrate with our
host and storage environment?
• Does the solution have a good support service? Can we get
access to help when we need it?
• How effective is the relationship between this vendor and our
backup vendor(s)?
A Vendor in Context: Pillar Data Systems
Having established four guiding questions and supporting follow-ups
that should be posed to any D2D vendor, we will now apply these
questions to a real world vendor. In this section of our report, we
provide our summary opinion on Pillar Data Systems and their Pillar
Axiom storage system
offering, viewed from the context of our four evaluation questions
for D2D.
Based in San Jose, California, Pillar Data Systems is a
next-generation enterprise storage systems vendor, providing
integrated SAN, NAS, and data protection storage in a single
solution. Before putting our four questions to Pillar, we first
reviewed its Axiom product line. We can report that this platform
struck us as uniquely suited for D2D data protection for several
reasons:
• Axiom is a thoroughly modular storage platform. Every Axiom
deployment utilizes three components that can be arranged depending
on workloads: Pilots (system management modules) Slammers (data
mover modules) and Bricks (RAID capacity modules). Depending on the
classes of storage, usage models, and workloads, these three
elements can be structured for optimal price-performance. In a D2D
context, this means the ability to avoid bottlenecks, on the
fly.
• Axiom supports both primary storage and secondary storage on
the same platform. This unique feature lets the Axiom behave as
“two or more devices in one.” In a D2D context, this
‘multi-tenancy’ capability means that IT teams can scale
secondary storage needs as they occur,
adding capacity as required.
• Axiom has significant performance and reliability intelligence.
First, Axiom supports D2D Quality of Service (QoS) configurations
that optimize sequential (LUN or file-based) reads on a system
level. Second, Pillar has spent significant engineering energy
improving the reliability
of industry-standard drive technologies. These two factors add up
to a speed and reliability advantage over most existing D2D
offerings.
Bearing these key differentiators in mind, we can now turn to our
four evaluation questions, once again.
1. How will this D2D offering scale and adapt over time?
2. What is the impact of this D2D offering on my backup and
recovery operations?
3. What is the economic rationale for this D2D offering?
4. How will this D2D offering integrate with our existing
infrastructure?
Question 1 in Context: How does the Pillar Axiom D2D scale and
adapt over time?
Conclusions:
Very High Scalability. Taneja Group believes that one of the most
notable strengths of the Axiom D2D solution is its extremely high
scalability. Every Axiom chassis currently scales to 307 RAW TB,
all of which can be dynamically allocated between tiers of primary
or secondary storage, over time,
based on IT management requirements.
Flexible Capacity Planning. We often find that early adopters of
D2D solutions were forced to plan capacities around the set
capacities available within the D2D appliance. When more capacity
is required, the enterprise has to invest in a new device. We
believe that the Pillar solution to this challenge is ingenious and
useful to end users: Scale based on needed capacity, all from
within the same array. In addition to the added benefits of
managing scale, there are a range of benefits resulting from the
fact that Pillar does not charge incremental license fees for
increasing system capacity.
In summary, when measured in terms of sheer flexible scalability,
we feel that Pillar compares very favorably against the range of
smaller appliance based solutions on theD2D market today. Because
the Axiom system is both highly modular, down to the 1.6 TB brick
level, and capable of supporting 256 usable TB in a single
high-performance chassis, it is a force to be reckoned with in the
realm of scalability.
Question 2 in Context: What is the impact of the Pillar Axiom
system on my backup and recovery operations?
Conclusions:
Wire-speed Performance,
Bottleneck Free: When we examined the Axiom architecture from a D2D
perspective, it became clear to us that Pillar has developed a very
clever way to virtually guarantee the elimination of bottlenecks
for high-speed D2D: Bricks (RAID Controllers) can be added to
increase I/O bandwidth. Specifically, each Axiom can support up to
four Slammers with four FC ports each, attached to as many as 64
Bricks, achieving a theoretical bound of 3 Gb/sec.
While reality is always different from optimal potentials, we know
that this architecture already places Axiom D2D far ahead of most
competitors. As a result, an enterprise could scale their
mission-critical disk retention to well over 100 TB and still know
that they will not suffer from performance degradation. In short,
their backup jobs and recoveries would be assured and their
methodologies will be consistent. This is both powerful and
appealing, in our opinion.
Granular Performance Tuning: Because of Axiom’s ability to assign
disk classes within the chassis based on linear velocity at the
disk level, it is possible to create a very high performance
mini-tier for D2D. In this scenario, a very high performance
recovery tier might be designated for mission-
critical D2D, and then backed up into a lower performance, more
economical archival tier. To our knowledge, this is the only
architecture of this kind available today within D2D data
protection. We believe it will be particularly attractive for
database protection and archival.
High Reliability: As discussed above, recovery of data matters as
much as backup job speed in the D2D world. While we know that Axiom
is fast, it is important to note that the solution is also highly
reliable, providing reliability that far exceeds off-the-shelf SATA
disk-based D2D offerings. Given the
Pillar focus on drive-level reliability, we would fully expect to
see their D2D customers register near 100 percent success in all
Axiom-driven recovery operations.
Question 3 in Context: What is the economic rationale for the
Pillar Axiom D2D Application?
Conclusions:
Technology Flexibility and Savings: The Axiom D2D solution is a
multi-tenancy solution, capable of supporting both primary and
secondary solutions in the same array. Further, the capacity of a
single chassis is 256 TB. The cost savings of an Axiom system are
most dramatic when compared
at an aggregate dollars-per-Gigabyte level across the entire
infrastructure. For example, rather than paying $30/GB for
fully-burdened, mission-critical primary storage and $10/GB for
fully-burdened secondary storage, the Axiom provides a single
solution for well under $10/GB. All these factors
together translate into a range of economic savings in terms of
granularity of scaling, capital outlays, savings on software
licensing, and ongoing management costs.
Efficient Teams: We have seen even modest D2D appliances create
synergies in team effectiveness. These efficiencies result from a
reduction in physical media management on an ongoing basis. In the
case of Axiom D2D, we believe this team effectiveness should be
amplified because of two factors:
the combination of management control for both primary and
secondary storage within the same console, and the dramatic
reduction in device management requirements resulting from
Axiom’s multi-tenancy advantages. In other words, instead of
buying a larger quantity of storage appliances,
including D2D, IT teams can buy a single Axiom and deploy it in a
cross-functional manner.
Process Optimization: Backup and recovery processes have
traditionally been synonymous with a range of physical media
management steps. For larger deployments, there is no question that
managing multiple physical devices ultimately becomes a significant
factor in process gains for D2D
deployments. We anticipate that as Axiom D2D deployments scale to
the hundreds of terabytes, there will be a compounding effect as
extremely large “online” disk capacities are made available to
production environments within the same array. This will open the
door for Axiom D2D to provide
benefits for not just backup/recovery, but online archive and
mission critical production data, all in one physical device.
Question 4 in Context: How will the Pillar Axiom D2D integrate with
our existing infrastructure?
Conclusions:
Zero New Footprint: Because the D2D functionality can be added
within the Axiom array, there is the potential for end users to
simply add capacity to their existing array. For those customers
newly acquiring Axiom for D2D applications, the extremely high
scalability of the solution will mean less
footprint growth versus capacities deployed (most D2D appliances
have between 2 TB and 50 TB maximum capacities.)
Interface Flexibility: Axiom D2D supports both “disk to disk”
and virtual tape interface methods.
Additionally, the Axiom platform supports all leading backup vendor
systems and can simply treat Axiom D2D bricks as target
capacities.
UI-driven configuration. D2D-optimized parameters for LUNs and file
systems are available in the Axiom Configuration Manager, making it
easy for end users to deploy D2D functionality within their
array.
Common Management Interface. Axiom supports UI/CLI command
commonality, meaning that IT managers can access and manage their
Axiom D2D environment from virtually any existing top-level
management platform or application.
Conclusion
D2D data protection has become a
very hot market with many competitors. With competitors comes
variation and complexity. To deal with this, Taneja Group
recommends that you develop a common set of four questions that you
can pose to all vendors. By examining each vendor in a common
context, the relative appropriateness, differentiation, and value
of various vendors’ technologies become visible. Without a common
evaluation structure, you run the risk of selecting a D2D offering
that may be attractive, but not optimal.
For the second portion of this report, we examined the Pillar Data
System Axiom platform in the context of D2D data protection. Based
on the four evaluation questions we posed, we then shared our
conclusions regarding Axiom D2D.
We believe that Axiom is substantially differentiated, in no small
part due to its ability to combine primary and secondary storage in
a common footprint. As scalability, performance and reliability
become even more critical in the evolution of D2D over the coming
years, we believe the Pillar
approach will prove increasingly forward-looking.
We highly encourage enterprises exploring D2D offerings to evaluate
Pillar Data System’s Axiom system. As many end users begin to
move beyond appliance-based thinking about D2D, we think that
Pillar might have exactly what they are looking for in the Axiom
platform.
