Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
Overview
Unstructured data is growing at an accelerated rate, and IT is feeling the pain. With machine-generated data, usergenerated data, and collaborative file data, IT has to deal with more sources and bigger files than ever before. In a
recent ESG survey of over 400 IT professionals, including storage directors, storage architects, IT managers, data center
architects, and operators working at midmarket (100 to 999 employees) and enterprise (more than 1,000 employees)
organizations, the rapid growth and management of unstructured data was cited by the largest percentage of
respondents as their primary storage challenge over any other storage issue facing IT (see Figure 1), making it the most
popular response.1 Of course, as data volumes grow, protecting that data becomes more challenging, so it is not
surprising that data protection came in as the second most cited primary storage challenge.
Figure 1. Top Ten Biggest Storage Environment Challenges
In general, what would you say are your organizations biggest challenges in terms of its
storage environment? Which would you characterize as the primary storage challenge for
your organization? (Percent of respondents, N=418)
15%
40%
11%
39%
10%
Hardware costs
39%
7%
25%
5%
25%
4%
Data migration
5%
6%
5%
All storage
challenges
25%
5%
Staff costs
Primary storage
challenge
10%
20%
19%
19%
17%
20%
30%
40%
50%
Source: ESG Research Brief, Key Challenges in Data Storage, December 2012.
When looking at the data from a company size standpoint, midmarket organizations were more likely than enterprise
organizations to call out the rapid growth and management of unstructured data, data protection, and running out of
physical space as their top storage challenges (see Table 1).2 The fact that midmarket organizations are more concerned
with their data protection strategies is consistent with previously conducted ESG research that revealed improved
backup and recovery as the top overall most important IT priority reported by midmarket organizations in 2012 and
beyond.
Table 1. Storage Challenges by Company Size
In general, what would you say are your organizations biggest challenges in terms of its storage environment?
Midmarket (100 to 999
employees, N=141)
47%
37%
43%
36%
31%
21%
7%
18%
Source: Enterprise Strategy Group, 2013.
It is pretty clear that doing more of the same when it comes to storing unstructured data just isnt going to cut it. The
overhead associated with traditional scale-up NAS systems, with limited throughput and drives, LUNs, and volumes to
configure, created the challenges IT is having with storing and managing data today. This is especially true for mid-sized
organizations that dont have the resources and expertise to spend all their time migrating data between systems to
perform storage optimization and management tasks. Next generation storage systems need to be easy, automated,
and scale-out to grow with the business. They must be optimized to save space, with deduplication and thin snaps. Of
course, they must be highly available, and protect the information. Exablox is promising all this and more with its
OneBlox solution.
Source: Ibid.
optimizing cash flow and asset utilization. This architecture also means that performance and availability scales with
capacity, so you dont suffer from performance bottlenecks as the system scales.
At the core of OneBlox is its new distributed object-based file system, architected from the ground up to be easy, secure,
and efficient. The secret sauce behind delivering the advanced functionality at scale is the use of a consistent hashing
algorithm. As with some other scale-out file and object storage systems, there are no volumes, LUNs, or RAID groups to
configure. The hardware environment is virtualizedas nodes are added (with zero configuration) to the global
namespace and absorbed into the system. Then they are presented as additional total capacity available to the users
and applications. Thats it. No LUN assignments, configuration changes, or boxes to check. Exablox claims that the initial
node can be up and running in about five minutes, and additional nodes added in about three minutes.
Exabloxs file system is full featured and automates a number of tasks. Some of the feature set includes:
Protection from physical failure: Drives fail, users kick plugs out of the wall, someone unplugs a wrong cable. These are
the physical failures IT needs to worry about. To protect from physical failure, OneBlox makes three copies of every data
object. When a file is written, it ensures that three copies are distributed across the drives in a node (in a single node
ring) or across multiple nodes in a ring. In a single node ring, the system can withstand two drive failures. It also means
that, in a three node or greater ring, two data paths, drives, controllers, or full nodes can fail without impacting data
availability. Because this is not traditional file storage using RAID, rebuild times are accelerated by leveraging all the
drives in the system to create new copies of the data objects as required, ensuring every file always has its three copies.
Extending the replication capability further, OneBlox can replicate to another ring, over WAN distances, and continue to
present a single global namespaceextending protection to entire ring and/or site failures.
Protection from logical errors: When it comes to data protection, logical errors are much more common than physical
errorsthings like accidental deletion or software bugs. OneBlox also addresses this issue with inline CDP (continuous
data protection) and creates a snapshot every time a file is written to OneBlox. It saves these files as versions, even if a
file is accidentally deleted. Every authorized user has access to all versions and can perform a self-service recoveryno
need to call IT. CDP is a fully integrated function, enabled by default, and cannot be accidentally turned off by an
administrator.
Space efficiency: OneBlox is one of the few solutions that provide inline deduplication of every file written as primary
storagethis is a significant advantage of an object-based file system. One would think that there would be tremendous
overhead associated with creating a snapshot of every version of a file, but OneBlox leverages inline deduplication
across the ring to minimize overall storage consumption and negate the impact of CDP.
Security: When a questionable drive is removed, there is always a concern whether the data on that device is
confidential. Exablox uses AES 256 level encryption to secure data at rest. This helps because as drives are removed from
the system, they are unreadable. Like CDP and deduplication, encryption is another feature that is enabled by default
and cannot be turned off.
Overhead is associated with the data protection methodology used. Because OneBlox creates three copies of data,
usable capacity is 33% of total capacity for any new file written. Deduplication will certainly make the equation much
betteras storage capacity does not get chewed up by duplicates, so system utilization could be much higher than 33%,
depending on the nature of the data stored and how much of it may be all or partially duplicated.
OneSystem
OneSystem is a multi-tenant cloud-based management system used to manage the OneBlox environment. It provides a
simple user interface that enables proactive monitoring and management of the entire Exablox environment.
Administrators no longer have to deploy onsite management servers, or patch and maintain softwarethey can simply
log into OneSystem from anywhere. Managed Service Providers will particularly enjoy the multi-tenancy that enables
consistent management across their many customers. This is a unique approach to management (at least for onpremises storage solutions) and should enable Exablox to rev software development more frequently as the physical
hardware is decoupled from the management software.
All OneBlox users will use OneSystem, so Exabloxs use of multi-factor authentication is reassuring to ensure another
level of security. OneSystem provides management services for the common user, group, and file share permission
tasks, but the valuable part is the collective proactive monitoring and reporting it delivers. It removes the configuration
complexity of many tasks, which will be a cultural change for many storage administrators.
Of course, there are some trade-offs with making things this easy. The default settings for these features cannot be
changed by the administrator. Enterprise storage administrators that want to turn knobs and dials and adjust everything
wont be happythis is the antithesis of what high-end systems deliver. But it will be fine for (in fact, desired by) midsized enterprises that may not have the staff of storage administrators to handle such tasks.
Market Impact
Exablox certainly addresses many of the pain points storage administrators face today. It invented a system that starts
small and grows with the business, is easy to use, leverages commodity hardware, and integrates security and multiple
levels of data protection. It almost sounds too good to be true.
No matter how good the technology seems, historically, it has been difficult for any new vendor to gain traction in
storage. The brand name vendors have a pretty tight stranglehold on their accounts, and will often do what it takes to
keep new vendors out. And storage administrators are (rightfully so) conservative when it comes to deploying new
storage technology. An if it aint broke, dont fix it mentality comes with being responsible for the corporate jewels,
the data itself. Often, that means making the safe decision and going with the known solution. But things seem to be
changing. Many of the core designs behind todays storage systems are (literally) decades old. RAID was invented in the
1980s. The pain points associated with managing data at scale, with accelerated unstructured data growth, and with
trying to store and manage that data in systems never designed for these use cases is creating an opportunity for new
storage technology to make inroads into IT. There is more willingness to try something new.
Of course, even though it is very cool technology, OneBlox is a new solution. It is not easy to create a scale-out file
system that manages data consistency across the cluster. Many have come and gone. Exablox will need to get proof
points into the market quickly. Those interested in kicking the tires can get in pretty easily. The system is expected to list
at under $10,000 for 32TB. It will be sold through channel partners who can source drives, or the user can source their
own drives. This is an interesting point. Administrators are often upset at vendor drive pricing. It is tough to pay vendor
drive prices when a 4TB drive sells for just over $200 at Best Buy. Of course, the system vendors will discuss drive
qualification and testing and the peace of mind that comes with knowing the drives are well vetted before they go into
the system. But the truth is that the secret sauce is in the software, not the drives. And if the software is designed at the
start with commodity drives in mind, it can mitigate potential problems. OneBlox appears well suited to take on that
challenge.