IDEAS Insights

The Real Story behind Sun’s Breakthrough Economics with Sun Storage J4000

Under the Open Storage banner, Sun just introduced several new products, including the next-generation “Thumper” storage server, the Sun Fire X4540, and a new line of economical storage arrays, the Sun Storage J4000 family.

Sun claims it delivers breakthrough economics with the Sun Storage J4000 family, based on the J4000’s SPC-1 benchmark $/IOPS results. However, this claim should not be interpreted as an Open Storage benefit. Although Sun Open Storage may indeed deliver very affordable solutions, the cost advantage of Open Storage would be predominantly enabled by the OpenSolaris ZFS (implemented in a solution such as Thumper), rather than the J4000 systems.

The J4000 arrays are essentially JBOD systems for direct server attachment, not a network storage platform. The J4000 can be a component in a Sun Open Storage solution (i.e., by attaching to a Sun Fire Storage Server that runs OpenSolaris), but on its own, it is not a complete Open Storage platform. In fact, the SPC-1 benchmark tests on J4000 systems did not involve OpenSolaris at all. The J4000 systems were attached to Windows servers as JBOD disk subsystems in the benchmark tests.

Additionally, due to the JBOD nature of the J4000, the claims made in the Sun press release comparing the $/IOPS costs of the J4000 with a SAN storage product were quite unfitting. In fact, the Sun J4000 should be compared with products such as the IBM EXP3000 disk enclosure rather than the DS3400 FC SAN array.

The IBM DS3400 is an entry SAN system, with redundant network storage controllers providing FC connectivity for multiple server hosts, as well as more advanced functionality (e.g. storage partitioning and copy services) than the J4000 JBOD systems. An apples-to-apples comparison should be made between the Sun J4000 and the IBM EXP3000 enclosure, as they are both SAS-based JBOD enclosures and are offered with the same SAS HBAs made by LSI (rebranded by Sun and IBM) for direct server attachment. An IBM EXP3000 Express Offering with a similar configuration to the Sun J4200 tested (one enclosure with 12 disks and a SAS HBA) would carry a very competitive price tag. Hence, the Sun J4000 SPC-1 results did not really present breakthrough economics, but merely reflected the cost delta between direct-attached storage and FC SAN storage.

Nevertheless, despite the unfitting comparison in Sun’s press release, the addition of the J4000 family certainly strengthens Sun’s competitiveness as a major storage vendor, and does enhance its Open Storage solutions. The J4000 provides the highest scalability among the current SAS-based disk subsystems from major vendors. As the new disk expansion options for the Sun Fire Storage Servers, including the new Thumper, the J4000 family significantly improves the scalability and subsystem performance of Sun Open Storage solutions. Furthermore, as NAS platforms, the Sun Fire Storage Servers with Open Solaris ZFS provide more sound competition to the popular Windows-based Storage Server solutions from other major system vendors, such as IBM, HP, and Dell (the offerings based on the Windows Storage Server software pre-integrated with their industry-standard servers).

HP File System Technology to Help Drive Linux into Open Storage

HP announced that it is releasing the source code for its Tru64 UNIX Advanced File System (AdvFS) under a license that will allow it to be integrated deeply into the Linux kernel. AdvFS is one of the most proven 64-bit system software technologies in the industry. It offers a number of powerful storage features, including the ability to configure flexible storage pools that can be shared by multiple file systems without needing to add a volume manager; the ability to expand, shrink or reconfigure file systems without taking them offline; flexible transaction log management, including the ability to log both metadata and data asynchronously or synchronously; availability features such as snapshots for consistent backups and support for recovering deleted files; and performance features such as online rebalancing of files and free space across storage pools. Moreover, these advanced features have proven in production for 16 years as a key component of Tru64 UNIX, which itself was one of the most advanced UNIX implementations on the market, before becoming retired in the fallout of HP's acquisition of Compaq.

HP emphasizes that its release of the AdvFS code into open source represents a technology contribution, and is not simply a port that will result in another competing file system for Linux, alongside ReiserFS or Global File System (GFS). The AdvFS code is being released under the same GPL Version 2 license used for the Linux kernel, which means that developers will be able to drive its features as deeply into the kernel as necessary to make them work effectively. HP is actually providing two generations of the AdvFS source code: the existing code taken from Tru64 UNIX, and work-in-progress code that HP produced during its abandoned effort to port AdvFS to its HP-UX operating system. HP is also providing an extensive set of design documentation, test suites and engineering resource related to the code. Linux programmers will be able to refer to this broad base of knowledge as they work on the next-generation Linux file systems currently under development, such as ext4 or BTRFS.

Indeed, it will take several years before the results of HP's AdvFS offering show up in mainstream Linux distributions. Aside from sheer altruism on HP's part, though, the contribution has several implications, both for HP and its customers, as well as the industry at large. First, the availability of proven AdvFS technology on a Linux foundation will make that OS more attractive to any remaining Tru64 UNIX users who may still be resisting HP's recommended migration to HP-UX (even if HP decided to offer innovative migration technology such as Transitive QuickTransit). If these users move to Linux, HP could still retain some as customers, which they could not if the users migrate to a competing UNIX system.

Further, by strengthening Linux storage management functions at the kernel level, HP will help to qualify Linux as a foundation for Open Storage, allowing it to serve as the core software for scalable and reliable network storage solutions that are based on standard hardware. The Open Storage concept has been promoted heavily by Sun with its OpenSolaris initiative, which allows users to couple the advanced storage functions in Solaris, i.e. the ZFS file system, with inexpensive, industry-standard hardware components. As the storage management function in standard operating systems become more powerful, such solutions could start to disrupt traditional enterprise storage systems, most of which are based on proprietary hardware. It is not possible to add ZFS technology into Linux, because it is licensed under Sun's Common Development and Distribution License (CDDL), which is incompatible with GPL v2. AdvFS matches several key ZFS features, and as Linux developers embrace AdvFS technology, it will become more feasible to offer Open Storage solutions based on standard Linux distributions, whereby solution providers add value with various data services (copy, replication, de-duplication, thin provisioning) on top of standard interfaces in the Linux kernel. By becoming fully engaged with efforts to enable Linux for open storage at the earliest stages, HP can start to position itself in a leadership role should open storage solutions start to generate serious momentum in the market.

Thin-Provisioning on SVC, Priceless

IBM recently released SAN Volume Controller (SVC) Version 4.3, adding several new features that greatly enhance the functionality and competitiveness of the SVC network storage virtualization solution. Most significantly, Version 4.3 added thin-provisioning capability to SVC with Space-Efficient Virtual Disks (SEV). SEV is designed to use physical storage capacity only when data is written to virtual disks instead of dedicating physical capacity to the entire defined virtual capacity.

Thin-provisioning technology has been very well received among customers. Besides the cost savings derived from improved storage resource utilization, customers are also vastly attracted to the simplified storage administration enabled by thin-provisioning. Appealing to the customer demand, major storage vendors have been adding thin-provisioning capability to their storage systems. Soon, thin-provisioning will emerge as a popular feature and a common customer requirement for storage systems. In fact, thin-provisioning support has been a work-in-progress slated for the next major release of the SNIA Storage Management Initiative Specification (SMI-S), which is expected in the 2008-2009 timeframe.

However, storage provisioning is a host-related task. Thin-provisioning is most effective when it is implemented at the same logical layer that is being presented to the hosts. In a virtualized storage network, hosts interact with the network virtualization layer instead of with individual storage arrays. The virtual representation of physical storage is provided by network virtualization engines such as IBM SVC, HDS USP, and EMC Invista.

HDS added thin-provisioning capability with the Dynamic Provisioning feature on its latest-generation USP V and VM systems. This initial release supports internal storage only, though HDS claims that Dynamic Provisioning will soon be supported on externally attached storage as well. Nevertheless, Dynamic Provisioning is a charged feature, priced according to capacity tiers. While EMC has be adding thin provisioning capabilities to its storage platforms (so far Celerra and Symmetrix), EMC currently does not support thin-provisioning on Invista. Further, any future support of thin provisioning on Invista could possibly be an added-charge feature, since EMC has expressed an inclination to charge for the newly added Virtual Provisioning feature on its Symmetrix arrays.

Notably, thin-provisioning capability is supported without an additional charge on IBM SVC. The use of SEV is included in the base SVC virtualization authorization license. Existing SVC customers can also take advantages of SEV without additional investment, as long as they are entitled to upgrade their SVC software to the latest version.

The no-charge addition of thin-provisioning capability on IBM SVC further solidifies its competitive positioning among storage virtualization offerings from leading vendors, as the most cost-effective and feature-rich alternative today. For a solution entry price of about $50,000, SVC enables customers to both transform their existing network storage infrastructures into a virtualized environment with unified manageability and benefit from the latest technology innovations, such as thin-provisioning. Conversely, $50,000 may not even cover the thin-provisioning enablement licensing charges on other alternative solutions.

Not Extreme Storage, but Beyond Storage

HP just announced a new solution offering, the HP StorageWorks 9100 Extreme Data Storage System (ExDS9100), positioned as an ideal platform to deploy streaming media applications. But don’t let the name fool you – the new HP system is much more than another storage array.

The HP ExDS9100 offers rack-mounted, factory-integrated hardware, including an HP ProLiant c7000 BladeSystem with up to 16 blades, as well as HP StorageWorks disk controllers and drive enclosures supporting up to 820 disks (820 TB of raw storage capacity with 1 TB disks). These hardware components utilize the latest industry-standard technologies to provide high-density computing power and storage resources in an environmentally friendly fashion. Although the HP hardware is impressive, it doesn’t make the new HP offering uniquely distinctive. And a near-petabyte storage platform is quite scalable, but not so extreme today.

The secret-sauce of the HP Extreme Data Storage System is the PolyServe Matrix Server technology, which HP now owns through its 2007 acquisition of PolyServe. A key component of this technology is the HP Clustered File System (previously known as the PolyServe File System). The Matrix Server architecture uses the clustered file system to provide high-speed data I/O to clustered applications up to 16 nodes. One well-known implementation of this cluster technology is scalable NAS file serving, using the cluster as a NAS server to transmit data via network file protocols (such as NFS and CIFS) to other application servers or network clients. This implementation – a NFS file serving solution – is available from HP as the StorageWorks Enterprise File Services Clustered Gateway. However, the performance advantage of this architecture is fully unleashed when the applications, such as databases or streaming media, are running on the clustered servers and transmit data via direct I/O to the file system (without the overhead of network file protocols) – as implemented in the Extreme Data Storage System.

The product concept of Extreme Data Storage capitalizes on HP’s extensive experiences in providing scalable storage solutions for a variety of customers using HP PolyServe coupled with HP storage arrays. The new HP ExDS solution is designed to offer customers optimal scalability, density, ease-of-use, and affordability for their digital media and Web 2.0 deployments. ExDS in fact offers more than just a high-density storage platform; it provides a clustered application platform that is optimized for high-speed data I/O. The HP ExDS is a great example of the value that system vendors can bring to the storage market beyond innovations to the storage systems alone. It provides customers with a complete, integrated, and fully supported system solution that reduces the total cost of storing and accessing business data.

HP Storage Focus: Virtualization, Automation, and Infrastructure Management

As the first calendar quarter of 2008 closes out and the second quarter begins, industry analyst conferences remain at the forefront of vendor activities. HP held its 2008 Technical Solutions Group (TSG) industry analyst summit on April 1st and 2nd at Boston’s Westin Waterfront Hotel. HP’s TSG group represents a close to $40b business that includes servers, storage, software, and services.  At last year’s conference, HP stated that it was on the path to grow TSG and this year EVP Ann Livermore reported a greater than 10% year over year growth rate for TSG as a whole. Enterprise Servers and Storage (ESS) is approximately half of TSG’s business and reported an over 9% year-over-year revenue growth.  It appears that so far HP is executing on its strategy to grow this once struggling business segment.
 
The overall solution set for TSG revolves around the areas of business information optimization (BIO), business technology optimization (BTO), and adaptive infrastructure (AI). BIO consists of products that provide information management, such as discovery & classification, information analytics, and information delivery.  BTO focuses on areas such as application management, performance management, and SOA governance. Adaptive Infrastructure, certainly not a new initiative for HP, focuses on the datacenter functional areas, such as management, security, and automation. The goal of these solution areas is to transform the datacenter to the next generation where low cost commodity-type assets result in an automated, 24x7, lights-out environment.  This is certainly a worthy goal, but not unlike the goals of HP’s competitors. HP touted AI several years ago and that initiative seemed to have gone quiet for a recent period, but is now back in the spotlight. The important point however, is that the explosion of data combined with global demands will absolutely require this type of next-generation datacenter model in order to adequately manage business applications in the future.

The role for storage in this overall strategy is extremely critical. Although this blog post is biased toward storage, it’s safe to say that while servers provide compute power and application support, storage is where the data is kept -- and data is the lifeblood of the business. HP describes its overall vision for storage in the next-generation datacenter as a technology to “bridge the gap between data explosion and storage infrastructure.” The underlying components are categorized as dependability, change-readiness, and cost-effectiveness. Dependability includes the information management functions as well as automation. Change-readiness provides the ability to dynamically adapt as necessary -- modularity and flexibility are emphasized here (this sounds strikingly similar to the initial definition of adaptive infrastructure). And finally, HP addresses cost effectiveness with industry-standard components and a unified management platform. The most critical piece to a cost effective infrastructure is the ability to scale. Scaling with traditional storage components is certainly achievable, but not economically. Only industry-standard components that can be seamlessly added and managed can scale to meet future demands in a reasonably economical manner.
 
Although there was some discussion of infrastructure virtualization, thin provisioning, and SaaS, HP didn’t provide a tremendous amount of product detail.  But of course, the level of content was reasonably appropriate for this analyst meeting. HP has identified many of the critical issues facing CIOs and IT managers, but so have many of its competitors. Nevertheless, HP’s loyal customer base combined with its vast resources has it well positioned to maintain its current growth path and lead the charge to the next-generation datacenter.