Storage Peer Incite: Notes from Wikibon’s June 19, 2007 Research Meeting
Wikibon presents Storage consumerization: Not if but when. This week we look at the impact of new consumer-oriented, online storage services fielded by Google, Yahoo, and others. With online storage priced at $2 per GByte per year, an order-of-magnitude lower than the storage chargeback rates in many organizations, it will only be a matter of time, and that probably a pretty short time, before end-users, including the business executives who sign the checks to pay those chargebacks, start asking why internal IT cannot provide storage at rates and simplicity-of-use comparable to that available online. These services are already appropriate for use by small-to-medium companies (SMEs) and may find their way into the storage strategies of large organizational IT departments, seeking to reduce the cost and complexity of low-end storage. The question is how and how quickly corporate IT and the traditional storage vendors will respond to this new competitive pressure to cut cost and complexity in an area that until now has largely been isolated from those pressures.
One of the more important and contentious topics in IT infrastructure generally and storage in particular today is the question of the degree to which "consumerized services" -- storage and other IT services delivered directly to consumers by providers such as Google and Amazon.com -- start to affect critical design, implementation and management decisions for enterprise storage. The trends are clear: consumers are becoming more engaged in defining IT value. They are voting with their dollars at home and in social groups, even if they are not yet participating actively in the workplace, for simpler, more predictably set pricing and more aggressively packaged services as a superior way to gain access to technology.
Additionally, in the storage world we see a dramatic growth in unstructured data often associated with end-user computing as opposed to high-throughput transaction processing. In this user-oriented computing the consumerization effect will be much greater, because end-users can transfer their consumer knowledge directly into decisions regarding how end-user-focused applications are built, deployed and managed.
In the last few years consumer market players such as Google, Amazon and X-drive have fielded some very attractive for-fee storage services, appropriate not only for consumers but also for small-to-medium-sized enterprises (SMEs).
Also, the public infrastructure used in a number of different application domains is in many respects clearly becoming superior to the infrastructure found in many organizations, particularly at the networking level. Consequently, we expect to see a greater effort by third-party service providers to target storage as an opportunity to drive new for-fee services to users who have become used to accessing online services for free. For instance, Google provides Gmail for free to consumers but is making Gmail available as a for-fee service to organizations as well. With these come commitments and promises about how storage capabilities will be administered inside that contract.
We see typical price points of $2/GB/year for renting access to tier 2/tier 3 storage capabilities. This is much less than the $20/GB/year (or higher) price-point for tier 1 and TP-type applications or even the $7/GB/year for midrange tier 2 storage of many internal charge-back schemes.
Regardless of whether these services actually can meet enterprise needs in terms of support for enterprise applications, deduplication, backup, etc., end-users, who include the business executives who write the checks to pay IT's chargebacks, will be asking why internal IT cannot close the gap between what it charges and X-drive's fees for simple storage. Indeed, far from resisting such questions, IT may be asking its storage vendors similar questions. Increasing numbers of shops are struggling to maintain the multiple, complex, and sometimes redundant skill sets (i e., a different encryption algorithm for each storage implementation) imposed on them by their vendors. They may be willing to outsource low-end storage to companies like Google and accept the potentially lower SLAs that may entail to simplify their own environments.
Over the next 12 months we expect to see consumerization become more than just a discussion in enterprise storage circles and in subtle ways find its way into important RFPs. Within three years we expect sizable percentages of storage RFPs at the enterprise level to specifically require consumer-oriented capabilities as part of the decision-making process.
The real question for enterprises is: Will they be able to use products from traditional storage suppliers to build internal storage infrastructures that are comparable in functional simplicity and price to those available in the third-party market? The path to answering this question will be complex and challenging, but if storage suppliers do not put greater emphasis on simplicity of procurement, deployment, and management, they and many IT professionals will find themselves out in the cold within five years.
Action Item: Storage administrators must begin evaluating the packaging, pricing and procuring models of third-party storage services from non-traditional companies such as Google, Amazon.com, and Microsoft. A first step must be to answer the question: What is wrong with the pricing and packaging models provided by third parties? Why can't our organizations answer the challenge from those third parties today? Only answers to those questions can lead to an honest assessment of what reasonable, value-generating internal storage strategies can deliver.
Answer: Pretty far. Over the past five years, we've seen advertising income contribute to substantial innovation in the Internet software domain. Storage has been both an indirect and a direct beneficiary of this trend with the creation of infrastructures that storage services can exploit and development of new storage innovations such as the Google File System (GFS). While not available as a commercial product, GFS has gripped the computer science world and further advanced the idea of building massively distributed systems on the notion that components will fail and prompt recovery can be designed into the system.
Perhaps not coincidentally, these advancements have occurred simultaneous to an explosion in unstructured data.
The extent to which Google and other consumer service providers will encroach on the enterprise storage domain remains a function of: 1) how far businesses will allow advertising models to permeate corporate walls, and 2) how aggressively traditional storage vendors respond. It seems apparent that adoption by small- and mid-sized businesses is forthcoming, however delivering integrated sets of storage services for the enterprise will probably require active participation of established storage leaders. Remote backup, file services, remote replication, C-site services specialized archiving (e.g. email archiving) and compliance applications are likely candidates for new services models to emerge. Do established storage vendors have anything to offer here? You bet; the question is when and how fast will they respond.
Action Item: Established storage industry participants and upstarts in the ecosystem must not sit back and wait. Rather, they must respond to increasing pressure to adopt consumerized IT models. This means packaging offerings that support a storage services vision and include SaaS, channel partnerships with service providers, metering and monitoring software, more competitive pricing and attempts to leverage existing storage assets as applied to storage services.
For IT organizations inundated with complaints about why their applications and infrastructure can't be as simple, intuitive and secure as email from Yahoo, Microsoft and Google, some are pointing to Google's less-than-inspiring SLAs for hosted applications as evidence that a reprieve is in sight. Google's SLA terms do leave lots of wiggle room for the consumer giant such as:
- Promising penalties for downtime based on user error rates exceeding 5% but using server error rates as the adjudicating measure.
- Defining downtime as 10 minutes or more of contiguous interruption.
- Making credits contingent upon the customer's timely notification of service interruption.
What's more, Google has popularized the annoying software industry practice of keeping products in Beta perpetually.
Unfortunately these facts do not serve as evidence that IT can drop its guard and not worry about the ensuing consumerization onslaught. If anything, these terms and conditions are becoming the rule, not the exception. It is more likely that these realities point to an unstoppable scenario unfolding that conditions users to trade less stringent terms and conditions for simplicity, transparency, choice and scalability.
Action Item: Storage organizations must tip the scales and rebalance their service portfolios by benchmarking not only other organizations, but also leading consumer storage delivery models. The outcome must be flexible storage service models with simpler packaging, more competitive pricing, greater transparency of consumption and user choice.
Will IT organizations be able to install the Amazon Simple Storage Service to run on their own infrastructure? Will Google File System come shrink-wrapped? In the topsy-turvy world of consumerization the answer is probably not.
The economic pressures on companies like Amazon, Microsoft and Google to provide the best possible services for billions of end-users and millions of small companies means that they are unlikely to slow down and produce a product for thousands of large companies who will be reluctant to allow the traditional revenue streams that fund and drive these technologies. The best that large IT departments (say government departments) might expect is that a software services provider will provide an exclusive data center and network.
Is it likely that some as yet unthought-of alliance between a Telco, NetApp, Symantec and CISCO could combine skills to produce and market such solutions to large customers? Maybe in the short term, but as the consumer-led companies add more functionality, security, and integration to their offerings, and the software economies of scale kick-in, it will be an unfair fight.
This factor-of-ten price difference between public and private storage services will lead to significant tensions between IT and operation departments over the next few years, similar to previous tensions produced by microprocessors in mainframe shops.
Action item: CIOs should base organizational priorities on the reality that consumerization of storage is an unstoppable train. They should organize to embrace and qualify such services, and be proactive in showing where they are appropriate and where they are not. Strategies that just point out the inadequacies and missteps of consumer storage service providers will ensure that IT departments are seen as part of the problem.
Consumer Storage services for the right applications will be very appealing to department heads with tight budgets. A simple comparison of storage costs will probably show the ability to store five to ten times more data with the same budget. It is likely to be competitive even with an internal “bargain-basement” storage service with minimal controller function and high-density SATA disks.
IT will need to support users in evaluating the suitability of consumer storage services. The more that file-based data is shared among users, the wider the dispersion of users, the greater the number of devices accessing or creating the data (e.g., PCs, cell phones, in-store scanners) and the greater the autonomy of the user departments, the higher the likely value of using external storage services. Collaborative computing over long distances, email, email archiving and backup applications for users and small business are likely to be a great fit.
Equally, many applications will be unsuitable for storage services. The greater the write/read ratio, the locking rate and the number of I/O operations in a typical unit of computing (transaction, query), the lower the potential benefit. Block-based applications such as transactional database are very unlikely to fit.
Action item: IT should respond by encouraging pilot systems, analyzing the consumer storage services as they evolve, and qualifying them for usage. Critical parts of that qualification will be storage service function, performance, availability and disaster recovery. Some organizations will want to assess deletion and archiving policies, and methods of holding backup data and systems in escrow in case of events such as bankruptcy or seizure of assets. Care should be taken in excluding services because of lack of a feature; business managers will be better positioned to evaluate business importance.
While most agree that consumerization of storage (and, in general, IT) is occurring and will have important consequences, disagreements regarding the rate, scope, and scale of the changes often are significant and vituperative. Usually, however, these arguments reveal biases regarding what a decision-maker wants to happen to IT and not clear insight into a likely consumerization path. For example, for those who want to see IT organizations lose influence, consumerization promises to substitute third-party, online services for internal IT systems and personnel. For those who want to see IT organizations persist – or even grow – consumerization will catalyze new demand for IT capabilities but ultimately fail to deliver, thereby generating a new expansion of IT budgets and influence within businesses. However, the most likely effect of consumerization is a steady, not cataclysmic, expansion of complex IT offerings from online service providers, coupled with a rapid increase in emphasis on infrastructure-as-a-service product capabilities from traditional and new hardware and software suppliers. Ultimately, then, the "consumerization of IT" is best thought of as a style of usage and service delivery that will include increased use of third-party services coupled with a significant role for evolved IT organizations, and not mainly as a prescription for a radically new IT market structure.
Action Item: No IT capability should be retired simply because the "consumerization of IT" someday may obsolete it. Prudent business leaders should seek the benefits of simpler IT in whatever form it may take – service or product, delivered externally or internally.