Meet the Data Processing Workflow Challenges of Oil and Gas Exploration with Advanced Data Storage Solution Profile
Halliburton Landmark SeisSpace Software and
Hitachi Storage Solution Match New Levels of
Sophistication in the Energy Industry Match
To meet growing worldwide demand, oil and gas
exploration and production (E&P) organizations are under
greater pressure to find new sources of energy. And while
exploration has always been expensive, today’s programs
are in increasingly hostile environments, making speed to
discovery much more urgent.
To expedite efforts, E&P organizations rely on
sophisticated geophysical technologies. Some of these
include reverse time migration, waveform inversion, and
3-D and 4-D downhole sensors to support making higher-
quality decisions. Growing volumes of 3-D data must be
analyzed in shorter time frames to make efficient use of
resources deployed in the field.
This explosive growth in data volumes presents new
processing challenges. To get the most out of raw
data, most E&P organizations have turned to 3-D
seismic processing applications. Many look to the
Landmark SeisSpace software, in particular.
A division of Halliburton, Landmark designs seismic
processing solutions to scale from field quality control
up through full volume, real-time production processing.
These software applications can be optimized for spe-
cific processing throughput requirements. In particular,
Landmark software is optimized for tasks, including:
■■ General quality control (QC), target investigations,
and field-specific data integrity workflows.
■■ Conventional time processing, Kirchhoff calcula-
tions, and amplitude versus offset (AVO).
■■ Production scale processing.
■■ Seismic coverage validation for illumination stud-
ies, acquisition planning, and targeted imaging
■■ 3-D prestack time and depth migration, velocity
analysis imaging, and finite difference forward
Meet the Data Processing Workflow Challenges of Oil
and Gas Exploration With Advanced Data Storage
Unlike conventional processing technolo-
gies, Landmark offerings place a special
emphasis on high-performance and inter-
active processing algorithms for today’s
high-performance computing environments.
Organizations need a shared network data
storage and management solution that
scales to accommodate the growing data
from seismic equipment and leverage paral-
lel performance characteristics of Landmark
SeisSpace. This solution will also provide
performance to feed these high-throughput
To better understand the potential per-
formance gains the right storage solution
can offer, Hitachi Data Systems joined with
Landmark to set up a test bed to experiment
with different system configurations. Exploiting
the unique features of the network storage
solution accelerated workflows, significantly
cutting the processing time required to derive
results. Additionally, the testing found that the
Hitachi system could run both primary and
secondary Landmark storage workloads at
the same time. This was something no other
vendor has been able to achieve.
When trying to match a suitable storage
solution with a seismic analysis solution, it is
important to keep in mind that you cannot
rely on narrow benchmarks. Real-world
application workloads are complex, and
overall analysis throughput can vary greatly
over time, depending on something as minor
as an application’s configuration settings.
With these issues in mind, we jointly exam-
ined the challenges, nuances and potential
benefits when integrating and optimizing
seismic analysis systems. Storage and data
management solutions and high-throughput
workflows were also tested.
In particular, the tests explored how to take
advantage of specific application features
to boost analysis workflows and reduce the
typical processing times for analysis.
The test searched for ways to exploit
Landmark SeisSpace performance enhanc-
ing capabilities. Landmark SeisSpace was
designed with new parallel-distributed-
memory architectures in mind. It supports
the JavaSeis prestack format, which allowed
for the development of algorithms that are
suited for true volume processing. The key
to these parallel efficiencies lies in the soft-
ware’s ability to leverage the parallel memory
I/O benefits of JavaSeis.
Essential components to the feature are
a storage filer and network throughput
capacities that are unlikely to become over-
whelmed by I/O transactions or storage
speed from seismic processing.
However, there is a requirement for an E&P
organization to take advantage of the soft-
ware’s enhancements. The organization
must ensure a delicate balance of a com-
puting system’s processing capabilities and
the IT infrastructure’s bandwidth and IOPS.
To evaluate the impact of fine-tuning a
storage solution to match the performance
capabilities of the software, Hitachi Data
Systems set up a test bed infrastructure
(see Figure 1).
The initial setup consisted of a Hitachi NAS
Platform (HNAS) 3090 cluster consisting
of a single storage pool containing 180 x
600GB 15k SAS drives. There were 10GbE
link aggregation control protocol (LACP)
connections into a 10GbE switch and 2
Fibre Channel connections per HNAS 3090
node to the back-end storage.
The testing compute nodes consisted of 32
Linux-based systems (CentOS v5.6) that were
each 1GbE attached. Each compute node
had 2 mounts, to a primary and secondary file
system. Each file system resided on its own
EVS (enterprise virtual server), which enabled
easy migration of the mount points.
With this configuration, 2 baseline testing
runs were conducted. The 1st was a read/
write test against seismic shot data; the
2nd was a read/write/sort function against
a similar subset of data. The 1st test run
completed in approximately 55 minutes,
and the 2nd ran in excess of 4 hours. These
results were consistent with previous expe-
riences, but with a performance edge over
other storage vendors.
Only Hitachi Data Systems with networked storage
has demonstrated the ability to meet requirements for
Landmark SeisSpace primary and secondary storage with
a single solution. This solution allows an organization to
consolidate its storage infrastructure.
Figure 1. The test configuration with Hitachi NAS Platform 3090 met the performance requirements
for both the primary and secondary storage for Halliburton Landmark SeisSpace software.
Innovation is the engine of change,
and information is its fuel. Innovate
intelligently to lead your market, grow
your company, and change the world.
Manage your information with
Hitachi Data Systems.
A number of configuration changes were
then made to fine-tune the performance.
The 1st change was to upgrade the existing
HNAS 3090 networked storage system
to the latest release of the HNAS system
software v8. One thing that sets Hitachi
Data Systems apart from competitors is our
firmware approach, with hardware accelera-
tion through field programmable gate arrays
(FPGAs). This capability allows adminis-
trators to change characteristics normally
associated with hardware through a soft-
ware upgrade. The HNAS system software
also helps end users analyze data access
patterns and then improve performance.
At each stage of the testing, standardized
performance reports were gathered against
the primary and secondary file systems.
The Landmark team adjusted networking
parameters in the compute nodes. The NFS
mount parameters were optimized for larger
block sizes, which resulted in up to a 15%
The parameters for sparse file system
functionality were also adjusted. SeisSpace
requires sparse file functions for accurate
application reporting, which includes the
capability to report the actual space used
(sparseness) versus the assumed (thin pro-
visioned) space utilization.
In subsequent tests, performance results
peaked at near the specified HNAS 3090
performance (72,921 IOPS; 1,100MB/sec
throughput without the performance accelera-
tor). At this point, all EVSs were also migrated
to a single physical node to demonstrate the
same performance, even without the failover
ability of a 2nd cluster node.
The original read/write shots test decreased
from 55 minutes runtime to just over 20
minutes runtime (at 1,035MB/sec through-
put), a 63% improvement (see Figure 2).
The 2nd, a sort test, also yielded more than
60% performance improvements.
The tests also demonstrated that the HNAS
3090 system (even as a single node) could
run both sets of Landmark workload (pri-
mary and secondary) simultaneously. No
other vendor has been able to successfully
maintain this performance.
Hitachi NAS Platform
Hitachi NAS Platform is an advanced, and
integrated, network attached storage (NAS)
solution. It is a powerful tool for file sharing
as well as file server consolidation, data
protection and business-critical NAS work-
loads. With HNAS, you can solve challenges
associated with data growth while achieving
a low total cost of ownership (TCO).
■■ Powerful hardware-accelerated file
system for multiprotocol file services,
dynamic provisioning, intelligent tiering,
virtualization and cloud infrastructure.
■■ High performance and scalability: up to
2GB/sec and 140,000 input/outputs per
second (IOPS) per node up to 16PB of
■■ File-level virtualization in a global name-
space isolates the user from technology
or vendor dependencies. It also enables
unified access to data stored on storage
systems from other vendors or Open
Source solutions like Lustre.
■■ Policy-based, universal file migration
simplifies deploying new technology and
migrating data, without impacting applica-
■■ Seamless integration with Hitachi SAN stor-
age, Hitachi Command Suite and Hitachi
Data Discovery Suite for advanced search
and indexing across HNAS systems.
Figure 2. After applying best practices
configuration testing, performance was
improved by 63% by using HNAS 3090.
■■ Integration with Hitachi Content Platform
for active archiving, regulatory compli-
ance and large object storage for cloud
■■ Simplifies your IT infrastructure by allow-
ing you to consolidate NAS devices or file
servers and migrate data by policy across
multiple vendors and technologies.
■■ Reduces the complexity of storage man-
agement and lowers your TCO.
■■ Significantly improves efficiency, agility
and utilization across NAS environments
through advanced virtualization and data
■■ Offers exceptional performance and
improves productivity for Halliburton
Landmark SeisSpace environments.
Figure 3. Highly scalable Hitachi Unified Storage
150 with Hitachi NAS Platform.