|

Next Generation Object Storage Summit 2013: Reconsidering Object Storage

Author(s)

Ben Woo

The ExecEvent

Current Overview

Object Based Storage (OBS) Systems have been available for many years. The basic concept behind OBS is the ability to present data in different contexts.

The term OBS was first introduced by IDC in December, 2007. Ben Woo and Rick Villars, co-authored the paper entitled Storage 3.0 – The Next Revolution in Storage Systems. That paper essentially defined what we know today as object-based storage systems. On June 18, 2013, leading vendors that offer OBS, along with a plethora of industry analysts gathered in San Francisco hosted by TheExecEvent to discuss the evolution of OBS and how OBS can gain better traction.

From the beginning, one thing was very clear. OBS means something slightly different to each delegate and sponsor. That said, at the opening panel consisting of four sponsoring vendors, all agreed that better awareness about OBS and providing more “on-ramps” were required to enable the OBS market to mature.

Two other panels were held over the two day event: the first being four analysts were gathered to discuss the market; and then a second panel consisting of analyst and vendors were gathered to discuss sales and marketing strategies to progress the OBS market.

While there was no resolution in terms of the definition of OBS, there was certainly a desire to come to some consensus. At odds, were taxonomical differences between OBS was essentially an access method; whether it was a “back-end” storage technology; where and how metadata is handled; and whether traditional file-based storage systems can be considered an OBS.

Even the usefulness of OBS and use cases for OBS were debated.

Areas of agreement included the fact that a user-definable extensible metadata schema was necessary; scaling is a key benefit; and not enough mainstream use cases have been documented. Another area of consensus was the fact that the confusion around what OBS is, and how it can benefit end-users is a big inhibitor for the growth and maturization of OBS in general.

To address the definition issue, emerging vendors called out larger market research firms including IDC and Gartner, and dismissed these firms’ ability to objectively define, evangelize, and help the OBS market to mature. All firms agreed that until a consensus can be found in defining OBS, that the market will continue to linger and lag behind other storage technologies.

Neuraspective™

Neuralytix proposed to attendees an alternative nomenclature for these storage systems – Object Addressable Storage (OAS). This moniker is more inclusive and better describes the underlying capabilities than Object Based Storage Systems.

We further propose that OAS must have the following attributes:

  • Provide the ability to share data between authorized systems, applications and users; and
  • Each piece of data (data object) must have some associated user-definable and extensible metadata.

Neuralytix does not believe that defining an access method (eg. REST) is in the industry’s interest. Access methods will evolve. Access methods popularity will fade in and out. While REST is interesting today, tomorrow, it may be CDMI, or some other advanced, yet to be named or defined protocol.

The requirement of location independence has also been dropped. An increasing number of storage vendors have put on their roadmaps, the ability to integrate cloud storage. As a result of this integration, the need to define location independence is no longer necessary.

To improve inclusiveness, Neuralytix also proposes to drop the self-referencing nature of OAS. This means that metadata can be stored either with the data object itself, or within a specific metadata server. In essence, separating the control and data paths for any given data object.

Specifically, Neuralytix excludes those storage systems that only provide a NAS (SMB and NFS) interface; irrespective of the underlying nature of the storage system. We believe that if the only access method is limited by a prescribed tree structure, and the amount of metadata is also prescriptive, then it fails to meet the criteria necessary to be an OAS.

Why Object Addressable Storage (OAS) in the first place?

There are multiple benefits of OAS.

It includes cloud storage

The storage industry is very conservative, and generally controlled by a very, very small number of vendors. According to IDC, the top 10 vendors represent over 70% of the market. What is not included are alternative storage providers, including cloud storage. This is a mistake. Cloud storage provides an alternative tier of storage to enterprises, with a new set of capabilities and service levels. By defining object storage as object addressable storage, it allows the industry to include cloud storage providers.

This significantly changes the market impact of OAS as opposed to OBS. Again, according to IDC, cloud storage is a significant part of the cloud ecosystem. Although dated, IDC observed that cloud storage services grew the fastest of all cloud services between 2008 and 2012, or a $5.6B market in 2012. $5.6B represents roughly a quarter of the external controller-based disk storage market, according to Gartner. Object based storage however, represents only an incidental and insignificant share of the overall storage market.

So the proposed OAS definition, which includes cloud storage, could potentially represent roughly 15% of the overall external+cloud storage market. This is a truly significant portion of the overall available storage market; and as such, both vendors and end-users need to take note of its significance.

Data reusability

Data reuse is very important in today’s evolving IT environment. By enabling data to be reused by multiple applications and users, the need to physically extract and/or transform data between application servers is eliminated. This reduces risk associated with any data movement and migration.

Proper data valuation

The value of data constantly changes. Typically, data has high value at the point of creation (or near the point of creation). This includes transactional data created at a point of sale. The data may retains its value through the reporting functions, but its value may drop significantly (nearing no value) when data is not being used, but archived either for regulatory or historical reasons. However, the value of data comes to life again, when it is used in predictive or historical business analyses, and returns to near zero value after the analysis.

With OAS, metadata can be set such that it helps to define the relative value of data at any given point in time.

Governance and Control

Since OAS can have user-definable and extensible metadata, information about access, use, and changes made to a data object can be journaled, and ultimately audited for internal controls or regulatory compliance. OAS also improves security, since precise, policy driven security can be applied to each and every data object. Virtual group access, can be granularly controlled at the data level.

OAS = Virtualized Data

The end-goal of OAS is simple. Data needs to be virtualized. Data virtualization is different to storage virtualization. In storage virtualization, capacity is abstracted from the physical media. Data virtualization is essentially the abstraction of data away from the application, away from the storage, away from the compute and network. Data objects can therefore exist (although not always persisted) independent of applications.

Application independence and abstraction is critical for reuse of data. It means that no data object is tied to any single application or server, thereby, being “liberated” (in Neuralytix terms) for use in any other authorized context.

Presenting data in alternative context is directly aligned with concepts around Big Data. Big Data is not necessarily voluminous. It simply looks at new ways of contextualizing data and information for the creation of innovation and new value. This is consistent with the reasons why OAS is important in the first place – which is to improve the value from data for any given enterprise.

Conclusion

Neuralytix is committed to helping the storage industry properly define, and advance the Object Addressable Storage (OAS) market. We believe that OAS will grow at a rate that is significantly higher (in terms of revenue and capacity) compared to traditional storage markets.

OAS vendors need to present end-users with more creative, more purposeful and broader examples of how OAS benefits enterprises. Conversely, enterprises need to recognize that already, the OAS market represents a significant part of the overall storage market (when considered in contemporary, and not archaic or traditional taxonomies).

Enhanced by Zemanta

Related Consulting Services

TAGS