Enterprise Data Storage 2019
Future-Proofing Your
Infrastructure
Enterprise storage has always been a your storage infrastructure needs to achieve,
challenge, and that continues in 2019. and understand trends that you can adopt
The biggest challenge: to future-proof today, you will continue to manage a
CONTENTS
your storage investments so you purchase working infrastructure that will fulfill your Current State of Enterprise
storage technology that serves you now, business’s information objectives. Data Storage......................................... 2
and will continue to serve you into the near
future. This report will start with the current state Growing Trends in Enterprise
of enterprise data storage and its primary Data Storage......................................... 3
Frankly, it’s not easy to do this level of enabling technologies. We will continue
strategic planning with data growing so with strategic trends for 2019, and our final Future-Proofing Your Storage
quickly, and accelerated storage technology section will be how to future-proof your Infrastructure: 6 Key Tips................... 6
development. However, if you know what storage infrastructure today.
Enterprise Data Storage 2018: Optimizing Your Storage Infrastructure © 2018 QuinStreet, Inc. • 1 •
ENTERPRISE DATA STORAGE 2019: FUTURE-PROOFING YOUR INFRASTRUCTURE 2
The Current State Of
Enterprise Data Storage
C urrent enterprise data storage is
built around three major challenges:
storing massive amounts of data,
SSDs allow for much faster ingestion and
initial processing, and HDD’s are still the
workhorse of the data center and store
IT is also increasingly aware of threat to
cloud-based data. As data moves to the
cloud, whether as backup and archives
protecting massive amounts of data, and active data relatively cheaply. Tape is or SaaS data, it is ultimately the users’
managing massive amounts of data for seeing an important resurgence for long- responsibility to protect large volumes
value and retention. term big data retention and security. of online data that is likely spread across
(Tape’s so-called air gap defense is simply multiple availability zones and clouds.
Storing Massive Amounts of impossible for malware to bridge.)
Data Managing Massive Amounts of
These media types remain the same Data
User-created data, streaming data, and whether you’re storing data to your data
sensor data combine to push petabytes center or to the cloud. Even after IT has stored and protected
of information into data centers data, they must still manage it for cost
every hour of every day. Incoming Protecting Massive Amounts control, compliance, and business value.
and outgoing data require significant of Data
transmission and ingestion speeds Different data priorities and types require
during movement, and enough storage Physical and cyber threats loom large for differing levels of performance and
capacity and performance at rest. IT organizations. On the enterprise side, capacity. Compliance is an important part
data loss or corruption can easily cost this picture, and IT its automatic toolsets
In that last sentence, “significant” is no hundreds of thousands of dollars in lost that can help them identify data lifecycles
monolithic rule, but depends on data value, fines, and damaged reputation. and retention dates. IT also needs to
type and priority. This in turn drives IT needs to protect data from natural locate and classify stored data, and
expenditures on bandwidth and on disasters and malware attacks, and provide analysis tools so end-users can
storage media. protect mobile devices against loss, mine information for business value.
destruction, or theft.
Projected Data Storage Needs
Supply Demand
50000
40000
30000
Exebytes
20000
10000
0
2012 2013 2014 2015 2016 2017 2018 2019 2020
The years 2019 and 2020 are expected to be periods of exceptionally rapid data growth. Data: Harvard Business Analytics
Enterprise Data Storage 2019: Future-Proofing Your Infrastructure © 2019 QuinStreet, Inc. • 2 •
ENTERPRISE DATA STORAGE 2019: FUTURE-PROOFING YOUR INFRASTRUCTURE 3
Growing Trends in
Enterprise Data Storage
A mong all the many evolving trends
in enterprise data storage, the
following six defining trends stand out.
make careers out of developing
new malware code that they they sell on
the dark web to their hacker customers.
safe and can securely restore backup data
to a compromised network.
Each of the below has a deep impact on KEY TAKEAWAY: Attackers are getting
the entire data center and are driving The most popular seller is ransomware more sophisticated, and InfoSec must
significant storage innovation. because it is quick and easy to code, and stay ahead of them. The answer is not
simple to deliver through infected emails. a single tool but combining security
1. Finding Dark Data If a user clicks on a link or downloads a technology, constantly adapting security
file, the ransomware code takes root and frameworks, and consistent user training.
Dark data simply means unstructured spreads. More sophisticated ransomware
files that are stored but not easily attacks also create attack loops which 3. Software-Defined Storage
discoverable, making them essentially inject malware code not only are too (SDS)
useless for compliance and analytics. IBM hard drives and servers, but also into
estimates that over 80% of business data backup. Software-defined storage is still going
is unstructured and most of it is dark, and strong coming into 2019. SDS decouples
this percentage is steadily rising. Storing Train your staff to stay alert for suspicious storage intelligence from the underlying
this much unused data costs money in attachments and download links. And storage hardware, which saves money
storage systems and cloud subscriptions. don’t forget tape’s “air gap defense.” because IT can buy commodity hardware
Unless malware attackers can launch and provide the storage intelligence via
Many IT organizations are moving their digital viruses through the the software defined layer.
to auto classification software tools atmosphere, a virus simply cannot travel
that discover and classify previously from a networked device to an off-line Intelligence includes policy-driven
dark data, then trigger actions such as cartridge. This means that even if online workload processing, intelligent data
alerts or automated data movement. backup is attacked, off-line cartridges are movement, load balancing, dedupe,
Auto classification works best with an
information governance framework. Software-Defined Storage
Once your data handling framework is
in place, you can use auto classification
software to effectively locate, classify, and Control Plane
act on previously hidden data. Policy Based Management
KEY TAKEAWAY: With an information
framework and auto-classification
Virtual Data Plane
software in place, IT is now able to take Virtual Datastore Virtual Datastore
intelligent actions such as alerting
admins, automatically moving data,
optimizing data organization, proving
compliance for sensitive documents, and
creating a more efficient repository for Virtual SAN
search, analysis, and security.
2. Cyber Security
Cyberattacks are difficult to trace and SSD HDD SSD HDD SAN/NAS
are profitable for attackers. In fact, it’s Software defined storage decouples storage intelligence from the underlying storage
becoming so popular that developers hardware, enabling far greater flexibility in storage environments.
Enterprise Data Storage 2019: Future-Proofing Your Infrastructure © 2019 QuinStreet, Inc. • 3 •
ENTERPRISE DATA STORAGE 2019: FUTURE-PROOFING YOUR INFRASTRUCTURE 4
replication, snapshots, and backup.
SDS is not exactly cheap – admins still SDS will serve you well when you
need to spend money on performance
and capacity in commodity hardware.
But it is cost-effective, enables centralized need to simplify complex multi-
storage management, and frees IT from
storage vendor lock-in. Typical use cases
include:
vendor storage, extend the life
L egacy application environments.
The business is not ready to retire
of legacy components, and carry out
the application but would rather not
house it on expensive storage either. a cost-effective technology refresh.
Instead of choosing between high-
priced intelligent arrays and low-cost
basic infrastructure, they run the security, policies, and provisioning out a cost-effective technology refresh.
application on commodity hardware pools across a single logical
with intelligent data services. infrastructure, and can easily add 4. Internet of Things.
components without moving data and
ig data. Big data environments house
B load balancing. IoT refers to multiple data input sources
large volumes of data from sources at the edge of a physical domain: offices,
like business applications, streaming void vendor lock-in and future-proof
A homes, and factories for now; later even
data, and machine sensors. These investments. SDS enables companies smart cities. IoT depends on millions of
environments can be very expensive to add storage devices to the SDS pool sensors busily producing petabytes of
to store on high-priced intelligent and management layer. This lengthens information and sending them over the
arrays. SDS provides scale-out storage the lifecycle of legacy storage devices Internet to a centralized repository.
pools for data lakes with centralized and avoids being locked into a single
management. Look for SDS offerings vendor’s storage systems. There, analysts classify, research,
that natively support big data and apply information to make new
frameworks like Hadoop and NoSQL. KEY TAKEAWAY: SDS is not an automatic discoveries, improve business processes,
solution for all data storage. If your and produce information-propelled
ulti-vendor storage environments.
M existing arrays are working well, there is products and services. Factories are
Early SDS products lacked centralized no need to retire them early. Enterprise the primary business users of IoT, and
management for multiple vendor arrays serving high-transaction primary transportation is not far behind.
storage devices under their control. storage will always have a place, and
Now most SDS products can integrate specialized arrays for specific verticals Business-level IT organizations in
different devices under a single and use cases are popular. But SDS will transportation, factories, retail, and
dashboard, which saves significant serve you well when you need to simplify automotive industries invest in the
time and resources for IT. Centralized complex multi-vendor storage, extend necessary storage performance and
management enables IT to create the life of legacy components, and carry capacity and take security measures
to counterattacks. However, consumer
level IoT is vulnerable to cyber-attack,
which can potentially launch dangerous
malware into large portions of the
Internet.
Compute/Server Networking Storage
KEY TAKEAWAY: Advanced AI, rich
applications that leverage video and
other streaming data, and the Internet of
Things (IoT) are shifting tape and cloud
Hyperconvergence cold storage tiers towards long-term
Networking retention of massive data stores.
5. Artificial Intelligence
AI continues to grow in importance in
Compute/Storage Block the storage world. In the data center,
it drives big data analysis, automation,
and software applications. Outside the
Hyperconverged storage leverages virtualization to enable scalability and flexibility. data center, it is extremely important for
Enterprise Data Storage 2019: Future-Proofing Your Infrastructure © 2019 QuinStreet, Inc. • 4 •
ENTERPRISE DATA STORAGE 2019: FUTURE-PROOFING YOUR INFRASTRUCTURE 5
video surveillance, pattern recognition,
transportation, and autonomous vehicles
-- all of which depend on intelligent
enterprise-level storage.
I-driven applications that depend
A
on real-time data. AI has a big impact
on transportation and logistical supply
chains. For example, a transportation
company’s AI software data center
tracks trucking movement throughout
a region. The data center ingests large
volumes of real-time data on weather MACHINE
and traffic information and feeds LEARNING
the data to the AI software, which
directs drivers to adjust their routes
for the fastest and safest trips. Other AI
software automatically detects unusual
energy patterns to predict events like
power surges our power outages. This
allows data center admins to mitigate
problems before an event causes
significant damage.
utonomous vehicles (AVs).
A
Experimental AVs generate up to 100
TB of traffic and driving data every Storage systems that incorporate machine learning can adapt to changing variables
day. The data consists of both sensor over time.
information that allows the AV to react
to driving conditions, and statistics including social semantics and create, retire or rebuild containers at will.
on AV safety. Automotive companies sentiment analysis for social media
require high performance/high marketing, data governance, and There are two types of data in containers:
capacity storage to store and protect fraud investigations into digital the container image and application
massive data volumes. communications. data. The image is ephemeral, as is
some application data such as testing.
achine learning. Machine learning
M KEY TAKEAWAY: Your existing data But when a container holds persistent
is a subset of AI. The key to machine management tools may already data such as business databases, IT is
learning is that once it learns from a be using AI and machine learning. responsible for storing and protecting
knowledge seed set, it can expand When you decide to make a separate this persistent data. When a retired
its own learning as it operates in its investment in AI-enabled software, then container spins backup, its data should
environment. For instance, machine understand your business case and how be immediately available.
learning enables software toolsets to achieve ROI with a significant financial
to dynamically balance and optimize investment. For example, sentiment The three primary data protection
data on SSDs to compensate for flash analysis software can raise profits by techniques for storing persistent
wear-out. Over time, machine learning improving social media marketing. databases outside the container on the
intelligence improves dynamic But realize that the level of storage same server/host, store the database and
data placement without manual required for AI and machine learning a docker volume that enables replication
intervention. Machine learning also puts yet another burden on your storage and snapshots, or traditional backup
builds analytical models. Programmers infrastructure. software that backs up the persistent
create the initial analytical toolset, data in the container the storage
which independently adapts and grows infrastructure.
6. Container Storage
as it encounters new data. The result
is that the toolset issues increasingly KEY TAKEAWAY: Containers are a critical
Containers are virtualized, stand-
accurate conclusions and results to enabling technology for developers,
alone executable software that
business users. testers, and application portability. If the
shares operating systems and houses
application code and data, runtime, data is truly ephemeral, then container
attern recognition. AI recognizes
P usage does not have a large impact on IT
toolsets, libraries, and configuration.
patterns and, by extension, recognizes or the storage infrastructure. However, IT
Container applications are platform-
when those patterns are broken. should be prepared to store and protect
agnostic, meaning that they run equally
Pattern recognition works across persistent container data, treating it like
well in test/dev environments as they do
a broad area of business use cases any other type of retained data.
application deployments. Admins can
Enterprise Data Storage 2019: Future-Proofing Your Infrastructure © 2019 QuinStreet, Inc. • 5 •
ENTERPRISE DATA STORAGE 2019: FUTURE-PROOFING YOUR INFRASTRUCTURE 6
Future-Proofing Your Storage
Infrastructure: 6 Key Tips
E nterprise storage managers have
a lot to do. Everyday firefighting,
managing multiple systems, and dealing
fail. At worst, monitor performance
and troubleshooting so a failure won’t
be a disaster. At best, replace the
Smaller companies can create a single
content repository by storing data on
a single array, but this won’t work for
with end-users leaves little time for vintage hardware with modern storage the enterprise. What enterprise storage
proactive tasks like future proofing the systems. Look for systems that give you managers can do is use software
storage environment. However, when central management consoles, such tools to discover data on different
IT squeezes out the time to strategize as integrated systems from the same devices and manage it as a virtual
and upgrade, ongoing tasks will become vendor or software-defined storage. content repository. Search, eDiscovery,
easier and more effective. management, and governance tools
Also consider investing in failover operate in the virtual repository.
Storage admins are generally services with a Disaster Recovery as
responsible for six major domains: a Service provider. DRaaS isn’t the To do this, you will need to search for
1. Performance cheapest service in the world but losing data on the network and edge devices,
2. Availability and Reliability critical application availability for hours defensibly delete outdated files, and
3. Governance and Compliance and days is going to cost a lot more in then move the rest to the repository.
4. Control Cost money, time, and reputation – yours and Use enterprise search tools to locate
5. Software-Defined Storage your company’s. both visible and dark data, define it by
6. Cloud Data Protection metadata and content, and apply bulk
3. Data Management: actions such as delete or move.
1. Performance: Centrally manage data and use
Optimize your storage for speed. tools to understand what and Additional usage case are big data
where your data is. analysis and compliance. Invest
SSDs are very popular for tier 0-1 in analytics tools for big data that
performance. Most data centers still store
most of their active data on hard drives,
but all flash arrays are becoming more
popular as prices fall and SSDs get faster.
Modern tiering begins with the Tier 0
and Tier 1 all-flash tiers. From there, data
moves to a disk tier in the storage system
or to another networked system. From
there, aging data backs up or archives to
tape or a cool/cold cloud tier.
DATA
MANAGEMENT
This is a lot of data movement. Use
automated tiering tools to free up your
time, release aging workloads that slow
down production environments, and
save money on storage purchases.
2. Availability and Reliability:
Adopt DRaaS for cloud failover
and failback.
You can stretch out the life of legacy As storage systems grow in complexity, a central data management system becomes ever
hardware, but eventually it’s going to more important.
Enterprise Data Storage 2019: Future-Proofing Your Infrastructure © 2019 QuinStreet, Inc. • 6 •
ENTERPRISE DATA STORAGE 2019: FUTURE-PROOFING YOUR INFRASTRUCTURE 7
identify and analyze structured
and unstructured data. If you need
to investigate company data for
compliance, use pattern recognition
toolsets that recognize suspicious
communications in email, messages,
transcribed phone conversations, and
social media. CLOUD
STORAGE
4. Control Costs:
Smart virtualization and cloud
storage.
You can save money with smart
purchase negotiating, but don’t stop
there. Storing aging data in the cloud
can save significant money. Savings
aren’t automatic – you need to watch
your restore costs. But for aging data, Cloud storage is a great platform for increasing capacity, yet admins must not be lulled into
using the cloud for cool and cold thinking that the cloud provider is the true custodian of the data. That responsibility remains
storage tiers can save money on long- with the customer.
term storage. Look for data storage
services that index data for searchability
data protection are protecting mobile adding encryption to data in-transit
for extra compliance value.
endpoints, verifying that a good backup and at-rest, and backing up your cloud
has occurred, and shielding the local application data.
Virtualization is another popular
network against ransomware and other
technology to save money and
hacking attempts. Practice strong authentication such as
management time on storage
customizing AD by user and role and
environments. It’s by no means a pure
Modern backup software contains using multi-factor authentication. Enforce
cost play – virtualized environments
endpoint protection features and industry and corporate governance
still require hardware and software
verifies successful backup, and anti- policies on the cloud. Work with your
purchases, and training and
malware and intrusion detection cloud provider: some of these security
optimization take time. But an efficient
systems work against hacking measures may be covered in your
virtualized environment saves money
attempts. (Do not neglect physical agreement, and you can add additional
by requiring less management time,
security, as more than one serious loss security measures to your SLA.
fewer hardware purchases, and reduced
occurred because someone picked
power costs.
up a few laptops and walked out with Your best bet is to partner with cloud
them.) backup providers that backup directly
5. Data Protection: from your cloud data stores, ideally
You are ultimately responsible For cloud-based storage, remember using the same cloud user account. The
for securing your data on- that the data owner has the ultimate best services do more than simple long-
premises and in the cloud. responsibility for protecting data in term backup: they also add indexing
the cloud. Configure your stored data and searchability for backup and
The primary threats to on-premise and SaaS data for protection, such as archives.
Enterprise Data Storage 2019: Future-Proofing Your Infrastructure © 2019 QuinStreet, Inc. • 7 •