| |
|
ATTO Diamond Array Outperforms
Nexsan ATAboy2 Array |
|
Tech Brief
Writes (i.e. Backup) Enterprise customers
with multi-
Terabyte backup needs and next-to-no backup
window need
storage devices with outstanding
Write-performance
capabilities. Hands down, the ATTO Diamond Array
outperforms the Nexsan ATAboy2 array in Writing
data in
single and multi-host environments.
Specifically, the ATTO
Diamond Array reaches its maximum advantage
(319%) over
the ATAboy2 in larger Write environments (4
hosts or more).
The real-world implication is that customers can
back up an
additional 378 GB/hr. using the ATTO Diamond
Array.
Reads (i.e. Restore) In today’s 24x7x365
business
environment, every minute of downtime can lead
to millions of
dollars in lost revenue. Therefore, after
sustaining a data loss it
is imperative that the data be restored as
quickly as possible. In
all host-scenarios, the ATTO Diamond Array
outperforms the
Nexsan ATAboy2 in Reading data by as much as
210%. In a
real-world network environment, this translates
into the
customer being able to restore an additional 454
Gigabytes of
data per hour by choosing the ATTO Diamond Array
as their
disk-based backup & restore solution. |
Overview
Extensive performance testing was done to
obtain a “head-to-head” comparison of the ATTO
Diamond
and Nexsan ATAboy2 storage arrays. To do so,
both products
were configured as RAID level 5 arrays with 4
Logical Unit
Numbers (LUNs) each. Both units were equipped
with
redundant 2-Gigabit Fibre Channel interfaces.
The arrays were
configured so that all LUNs were available to
both FC
interfaces to ease configuration and maximize
theoretical
performance. High-performance Windows® 2000
hosts were
connected to each array’s FC interface via ATTO
FC3300 2-
Gigabit Fibre Channel Host Bus Adapters (HBAs).
Framework
There were two main objectives of the tests.
The first series of tests determined the maximum
sustainable
throughput of each array in a RAID 5
configuration. Both
arrays were tested at varying I/O Queue Depths
and block
transfer sizes in both a 100% sequential Read
and 100% Write
I/O pattern. The second, more
application-specific series of
tests measured performance in a simulated
Enterprise diskbased
backup & restore environment. With 1 MB being
the
industry-standard block transfer size for
disk-based backup &
restore software, the performance of both arrays
was tested at
this block transfer size with 100% Write (i.e.
backup) and
100% Read (i.e. restore) I/O patterns. In all
tests, the I/O
Queue Depth was adjusted from 1 to 16 to
represent the traffic
generated in multi-host environments.
Assumptions
Due to the almost infinite combinations of
hardware and software available in an Enterprise
disk-based
backup & restore environment, a third-party disk
utility was
used to generate the I/O traffic representative
of a multi-host
environment. To obtain the performance results,
ATTO
Technology used the industry-accepted IOmeter
disk utility
to simulate the network environment and measure
array
performance.
Results
Series 1 – Both the ATTO Diamond and
Nexsan ATAboy2 arrays reached their maximum
sustained
Read performance in a 100% sequential Read I/O
pattern,
block transfer size of 4 Megabytes and a I/O
Queue Depth of 2.
The ATTO Diamond Array outperformed the Nexsan
ATAboy2 array with 23% better performance, or
sustained
Read performance of 242 MB/sec. for the ATTO
Diamond and
197 MB/sec. for the Nexsan ATAboy2. In all out
Write
performance, the ATTO Diamond Array once again
outperforms. Maximum sustained performance for
the ATTO
Diamond Array in the RAID 5 configuration was
153 MB/sec.
While the Nexsan ATAboy2 maxed out at sustained
Write
performance of 81 MB/sec., the ATTO Diamond
Array’s
maximum Write performance was realized at 100%
sequential
Write I/O pattern, block transfer size of 1
Megabyte and an I/O
Queue Depth of 4. The Nexsan ATAboy2 array’s
maximum
Write performance was realized at 100%
sequential Write I/O
pattern, block transfer size of 4 Megabytes and
an I/O Queue
Depth of 16
Series 2
In this more specific series of tests, the
arrays were
measured for performance in a simulated
disk-based backup &
restore environment. The I/O pattern was 100 %
sequential
Write (i.e. backup) and 100% Read (i.e. restore)
with a block
transfer size of 1 Megabyte. I/O Queue Depth was
adjusted
from 1 to 16 to accommodate for the effect of
multiple hosts
initiating backup and restore I/O requests. In
all instances, the
ATTO Diamond Array outperformed the Nexsan
ATAboy2
array in Writing (i.e. backup) data from the
hosts. In one
instance, the Diamond Array was up to 319%
faster in Writing
data than the ATAboy2 (see Figure 2). The ATTO
Diamond
Array’s Read (i.e. Restore) performance also
outperformed that
of the Nexsan ATAboy2 array (as much as 210%) in
all
simulated multi-host test environments (see
Figure 3).
Conclusion
IT administrators are continually being
tasked with the challenge of reducing backup &
restore
windows in today’s ever-expanding global
business
environment. As such, many IT managers are
moving toward
disk-based solutions as a means to increase
reliability,
flexibility and performance in their backup &
restore functions.
As a result the IT manager’s traditional choice
of using tape is
being relegated to long-term backup and
archival. This shift is
creating huge opportunities for disk-based
solutions to gain a
stronghold in the Enterprise Data Center as
dedicated near-line
storage devices. The advanced performance/costs
ratio of diskbased
solutions makes these systems particularly
attractive to
IT managers across the globe.
As early pioneers of the disk-based backup
trend, both the
ATTO Diamond and Nexsan ATAboy2 arrays are often
compared to one another. When it comes to
choosing the best
disk-based storage solutions for their needs, IT
managers
should immediately be drawn to the ATTO Diamond
Storage
Array. The Diamond Array’s high performance,
excellent
reliability and proven interoperability make the
ATTO
Diamond Storage Array an ideal choice for
near-line backup
solutions in Enterprise Data Centers.
Performance Advantage
The reason for the large performance advantage
in the ATTO Diamond Array over the Nexsan
ATAboy2 is in how the RAID 5 groups are
configured.
The ATAboy2 contains 14 drives. When this array
is
configured for a RAID 5 partition, all 14 drives
are utilized.
When using large capacity arrays like the ATTO
Diamond and
the Nexsan ATAboy2, most users will want to
partition the
available capacity into more than one volume.
The ATAboy2
takes the 14 drive physical RAID 5 LUN and
divides it up
across all the drives. Each Partition consists
of a portion of all
14 physical drives. When more than one partition
is accessed,
the drives will thrash and performance is
effected significantly.
The bottleneck becomes a function of the drives
seek ability.
When RAID 5 partitions are created in the 24
drive ATTO
Diamond Array, each LUN contains its own
physical drives. If
configured for three LUNs, each LUN will consist
of eight
independent drives. Performance to all three
LUNs
simultaneously can occur independently and is
not affected by
the drives seek performance.
|
