Storage Peer Incite: Notes from Wikibon’s April 25, 2007 Research Meeting
This week David Floyer presents IBM's Image for a Storage Future. IBM's recent announcement of the Multilevel Grid Access Manager didn't receive much industry attention but the Wikibon community made some observations concerning distributed, software-led architectures that potentially harbor significant insights for the future of unstructured content storage.
On April 24th, 2007 IBM announced two products that are related mainly in that they both target archive and retention applications. The IBM System Storage Multilevel Grid Access Manager is a product that IBM is OEMing from Bycast based on the Bycast StorageGRID software product that provides a set of system storage services between the DICOM and other access methods and storage devices generally. The IBM Systems Storage DR550 File System gateway is a new optional device for the IBM DR550 that provides file archiving capability through the NFS/CIFS network file protocols interface. It takes a more general purpose hardware approach to archive and retention applicatons. Users should focus most of their attention on the more interesting of the two announcements which is the Grid Access Manager, a grid of devices that allows security and recovery across distributed nodes and sites.
The solution today is specifically targeted at large digital medical image applications that have very stringent requirements for duration of archive, security through encryption as well as other characteristics that are specific to the medical imaging world. This product is likely to be a success within the medical imaging community and will certainly begin tumbling some walls in that space where solutions have tended to be relatively hardware specific.
However Bycast is using a proprietary set of technologies and interfaces engineered and currently marketed specifically for medical imaging-like applications that might also be transferable to other areas such as homeland security applications. But these are unlikely to be easily transferred into image-orientated archive and retention domains, generally. Nonetheless, this is an important announcement in that the combination of IBM and Bycast has demonstrated the potential of a software-focused approach on mediating and providing a general mechanism for accessing image files in an archiving and retention setting. Users should begin exploring how this technological approach can impact their image archiving and retention requirements as well as other storage requirements.
Specifically we belive this is a validation of the overall Extensible Access Method (XAM) approach to providing a set of storage service interfaces between applications and devices that can handle both archive and retention as well as other storage needs. XAM in general and the Bycast/IBM solution in particular provide a powerful reference model for how storage solutions are likely to evolve over the next few years. Namely software led/device followed, which is a very different model from the storage market generally, which has been driven by what can be accomplished with the controller and devices and how software can be used to manage that controller.
Action Item: The Bycast IBM solution validates the overall XAM approach to software-led storage innovation. Users should become as familiar as possible with this reference model and immediately incorporate it into their ongoing storage strategy discussions. As well, buyers should start pushing vendors to indicate how a storage world driven first by software innovation and second by hardware innovation will affect products, solutions and go-to-market strategies.
Recently, we've seen the emergence of new storage architectures emphasizing software-based systems that provide high performance, scalability and resiliency by spreading data, file systems and operating systems across distributed infrastructure. While this approach is not generally being implemented for large transaction processing applications, within unstructured content applications it is rapidly gaining popularity.
IBM's recent announcement of its Grid Access Manager based on Bycast technology is an excellent example of this technique as applied to medical imaging and records retention. Customers in the medical industry that want to implement large imaging, records retention and archiving solutions are seeing Bycast emerge as an innovator and key technology provider, partnering with the likes of IBM as well as HP. IBM's announcement will serve to provide further credibility to this market offering users a services-led solution that appears to be competitive and compelling.
Customers with scale will likely be able to take advantage of this solution, finding it easier to automate disaster recovery and backup relative to more traditional, monolithic storage approaches. Smaller customers however must weigh the costs of adopting a turnkey solution with a heavy services emphasis against the incremental benefits.
Action Item: IT organizations within the medical community should begin to aggressively investigate solutions like IBM's Grid Access Manager and understand the implications of this type of architecture on mid-to-long term image and records retention strategies. However, users should not expect nor assume this specific solution will be leverageable to address more generalized storage any time soon.
One of the technology questions that IT organizations should ask about IBM’s Grid Access Manager Software announcement is “what type of software is it?” Is it a medical application that plays in the medical record vertical marketplace, or is it a storage application that plays as a horizontal component across the fixed record management marketplace? If a product has a significant presence in a market, it confers economies of scale and protection for adopters that the solution will remain cost effective over time. It is very unlikely that a product will become a market leader in both horizontal and vertical markets.
In the case of the IBM Grid Access Manager, the product is clearly positioned as a hospital vertical solution. However, the concepts behind the Bycast approach hold great promise for the management of unstructured data over time.
Action Item: Storage managers looking to adopt IBM's GMAS solutions will require simple storage skill-sets optimized around the acquisition and management of low-cost commodity storage that will be software-managed. Senior storage architects should study the IBM/Bycast and Google File System to understand its applicability in broader applications.
Both the Google File System and the IBM Grid Access Storage Manager from Bycast share some common assumptions. These are that commodity software and hardware components will fail, and that continuously available storage architectures should span multiple storage servers, multiple types of disk farms, multiple locations and multiple clients. This emerging reference model for unstructured data is a significant departure from traditional, monolithic single system approaches.
Action item: System architects should use this reference model for future storage infrastructures, especially for unstructured data. XAM is a standard that is likely to facilitate the adoption of this approach, so that applications can request storage services on demand and architects should actively encourage its adoption.
In a conference call with analysts this past week, IBM executives credibly positioned the new Grid Access Manager as a formidable offering within the medical imaging space. Notable is the product's innovative software-based storage management approach that offers great potential to address large customer records retention challenges in the highly regulated medical environment.
We see this announcement as lending credibility to software-led storage architectures and in particular the Extensible Access Method. As well, we see the possibility that the soluton could be applied to other areas such as homeland security and possibly spatial applications. However, users should be very careful and not expect this solution to be extended broadly to other industries and applications as IBM is likely to move slowly due to the complexities of generalizing this solution.
Action Item: While this announcement does signify further momentum for software-led 'spread everything' storage architectures, users should take a wait-and-see posture in regards to IBM extending the Grid Access Manager beyond the medical and other highly scrutized industries that manage relatively predictable data types. Only leading edge customers with sizable business problems will benefit from this solution in the near term.
Invention in the storage industry historically has revolved around two poles: (1) engineering bit density at lower cost; and (2) increasing overall system parallelism by offloading I/O work to storage system controllers. Consequently, the storage companies have evolved into two, main forms: (1) device manufacturers competing to gain manufacturing efficiencies; and (2) storage system suppliers differentiating themselves on the basis of highly proprietary controller function. While device companies will continue their technology and cost battles, recent, software- based invention is poised to dramatically alter the dynamic in the storage system business, as grid, virtualization, and XAM (eXtensible Access Method) standards technologies mature and moderate the historical advantages of "big controller" offerings. We liken the coming change in storage to the changes that faced network suppliers during the transition to IP-based networking. Some, like IBM, fought the change, and ultimately were driven from the business. Others, like Novell, were forced to abandon the proprietary hardware/software product strategies and compete as software companies (for a while). Others, like Cisco, pushed IP technology hard with a combination of hardware and software invention, making IP innovation safe for customers. Over the next 18-24 months, storage companies will be faced with having to choose similar roles, likely facing similar results.
Action Item: Emerging storage standards and invention will create a new class of storage system winners. The question is: who will emerge as the Cisco of storage by driving new open software storage standards through smartly engineered hardware optimization.