EMC Takes The Offensive In The Storage Wars

With its January blizzard of announcements, EMC has launched an across-the-board offensive. By doing so, the company is broadening its horizons as a wide spectrum information infrastructure company. That does not mean it has neglected its storage roots--far from it. Announcements at both the enterprise-class and midrange storage system levels have opened new fronts for EMC to contend against competitors.

Enterprise-class (and other) storage announcements tend to be about product refreshes, where the new models in the solution family are typically bigger (in capacity), faster (in performance) and more cost effective (in the sense of having a better ROI than the previous generation to encourage migration to the latest and greatest). That is not EMC's emphasis.

Instead, with these latest announcements, EMC aims to make Symmetrix VMAX the "world's smartest storage array" by leveraging software "intelligence," a.k.a. "smarts," to improve system features and performance and deliver measurable value to customers.

How can this be, and why should customers care? For starters, an IDC study sponsored by EMC in 2010 showed a 44X information growth over this decade while IT head count remains flat. The implication is clear that if the ability to manage this stunning growth has not already gotten beyond the ability of humans, it soon will. That means that managing storage can no longer simply be business as usual. And the key is not to just make management processes simpler (which is always a good thing), but rather to make the management of storage "smarter," in the sense of solutions multiplying the ability of human administrators many-fold.

The key to EMC's strategy, which it announced last year, is what the company calls Virtual Storage. Virtual Storage has three fundamental pillars: automation, storage federation and virtualization enablement.An example: EMC's approach to automated storage tiering reflects the difference between "simpler" and "smarter." For operational management, simpler tends to encompass taking fewer steps (such as clicks) and clarifying the understanding/execution of complex processes. That is good, but it is storage tiering that entails shifting data to the storage systems and media that best support it. For that, an analysis of information (such as which blocks are "hot," meaning those which require high I/O systems, and which are not) has to be analyzed over time.

How best to accomplish this? Automation in the form of software algorithms that examine data usage patterns over time and automatically make the assignment of which data should be placed on which tier is a "smarter" idea than doing things manually. The overall benefits of this methodology include raising service levels (as data that requires faster access gets it) while lowering costs (by putting data that does not demand fast access on slower, less expensive storage devices) and reducing manual administration (as storage tiering manually would take a considerable amount of time even if simple to do).

EMC is essentially turbo-charging its already fully automated storage tiering (FAST) technology with next generation FAST VP (Virtual Pools). FAST VP offers virtual pool-based, sub-LUN, extent-based data movement. In layman's language, that means that more granular (i.e., smaller) chunks--in this case, smaller than a megabyte--of data can be moved automatically, resulting in far more efficient placing of data in tiers. For example, if even a piece of a larger chunk (LUN) was hot, it would be classified as hot (to meet service-level demands) even if the rest of the data was "warm" (less I/O required) or "cold" (zero or very little I/Os required).

What are the practical results of this? EMC cited one example where using FAST VP resulted in the client using 70 percent less Tier 0 flash storage (the most expensive tier by far) and 50 percent more Tier 2 SATA (the most cost-effective tier) HDDs. The company also claims that FAST VP can determine where data should be placed much faster. Another point is that using FAST VP in conjunction with a Symmetrix VMAX system with three tiers (Flash, FC and SATA) is much more cost-efficient than simply using a traditional Symmetrix system using only Fibre Channel. EMC noted that its new technologies in a 140Tbyte VMAX system could result in the number of drives being reduced from 1,129 FC HDDs to 152 Flash, FC and SATA devices, resulting in 40 percent higher performance and 40 percent lower cost, while providing more beneficial environmental changes (a much smaller system footprint and significantly less power consumed).

Symmetrix VMAX is EMC's favored platform for the initiatives leading to the private cloud. The focus here is on a new usage model where storage resources can be non-disruptively and transparently used to access data on demand in local or distributed environments. A key to enabling this approach is around the Symmetrix Enginuity operating system, and EMC states that the latest version, Enginuity 5875, is the biggest software release in Symmetrix history, with more than 50 new major VMAX features and capabilities.One of the biggest is Federated Live Migration (FLM), a critical tool for moving data from one system to another. Obviously, EMC wants its customers to migrate from existing Symmetrix DMX platforms to the latest VMAX products, but performing complete technology refreshes can be time-consuming (taking months, in some cases), disruptive in the sense of application downtime during cutover and requires significant administrative resources. EMC claims that FLM-based migration can be accomplished with 75 percent less system downtime and eliminates application downtime. It almost goes without saying that this significantly reduces overall costs and administrative work.

Of course, this story is not complete without examining the third pillar of EMC's strategy,  Virtual Storage. Here, VMAX capabilities have been expanded remarkably, allowing a system to manage up to 5,000,000 virtual machines on a scale-out cluster. That is truly enterprise-class. But that is just the start. A wide range of complementary enhancements have been made, such as the integration of VMware APIs with EMC's Virtual Storage Integrator (VSI), allowing it to work in conjunction with VMware Vcenter. That simplifies virtual server management while simultaneously maintaining architectural control of the storage infrastructure. EMC claims much faster creation and greater scaling of VMs with fewer data stores, and a 3X improvement in backups, testing, and migrations.

Large IT vendors that have a significant presence in storage, including EMC, HDS, HP, IBM, Oracle and NetApp, have tended to support at least two product lines--enterprise-class (i.e., high end, where, while costs are important, capabilities are more so) and midrange (where, while functionality is important, cost receives greater attention). But any market segment can bifurcate. In the midrange, that bifurcation is between block-based storage area network (SAN) systems (in EMC's case, its Clariion arrays) and file-based network-attached storage (NAS) solutions (like EMC's Celerra).

Recently, the concept of what is called unified storage has seen a surge of acceptance in the marketplace. These systems combine SAN and NAS storage in the same box. To date, most vendors that offered unified storage would do so as only one of their choices. That is, customers could continue to order SAN- or NAS-based storage separately.

With the introduction of its new VNX family of unified storage, EMC is demolishing that scenario with a radical new approach. EMC VNX solutions support both block- and file-based storage, and will replace the company's long-established Clariion and Celerra product lines. For those who might have separation anxiety about losing a familiar brand, consider two things: One. a customer will still be able to buy current products for some time, and support for existing Clariion and Celerra products will be continued, and, two, the new VNX products have all the capabilities of both predecessor platforms.This is a stunning move by EMC, but one that makes perfect sense. No customers lose any functionality that they need, and if they require something they did not have before, it is easier to add that feature or capability than to buy a separate unit. Now, on to the specific announcement.

The new family consists of two separate yet similar siblings. The VNX-Midrange Unified Storage family supersedes the traditional Clariion and Celerra products. Critical to understanding EMC's decision is its belief that the move to highly virtualized servers demands new storage solutions. That means shifting from traditional systems with static RAID groups to those with self-optimizing storage pools. That will require the automatic optimization of both the placement of applications on VMs, as well as the automatic optimization of data on disk (which means the ability to move data as appropriate as applications are dynamically moved). EMC states that this can generate up to three times the performance--say, three times the users and three times the transactions in a SQL Server or Oracle application environment while using a VNX system. And, both Unisphere and FAST VP (both introduced for the Clariion and Celerra lines in the third quarter of 2010) can be used with VNX.

The VNXe-Entry-level Unified Storage line is designed specifically for small businesses, remote offices or branch offices, as well as certain departmental applications. This is below the midrange market.

Why is EMC so interested in these markets? First, because the entry-level market is larger than the midrange market, and the midrange market is (and is forecast to remain) larger than the high-end market. To steal a line from Willie Sutton, EMC is going where the money is. But the fact is that entry-level storage customers, most of which depend on IT generalists rather than storage specialists for administration and management responsibilities, need a solution like VNXe. So what does the company bring to the party? The answer is a lot of R&D that has already been amortized with higher-level products and can be adapted to fit the specific needs of the entry-level market. The VNXe has wizards for system deployment, management and integration with existing server and application environments. VNXe also leverages EMC Unisphere, a user interface that provides a dashboard to manage system components. Obviously, this is a product that EMC is not selling direct, so its partner ecosystem will be paramount in determining its success in this market.

The IT information infrastructure wars are going to continue hot and heavy in 2011, and we expect storage battles to be competitively fierce. EMC made a number of major storage technology announcements and acquisitions in 2010, but the year was also one of strategic directional shifts for the company on a number of fronts. From where we sit, 2011 would seem to be the year those pronouncements turn to actions continuing the implementation of the company's strategic vision.On the Symmetrix side, EMC is encouraging the movement of customers from their current DMX products to the VMAX products. FAST VP will not only facilitate but also automate tiering, which improves performance and saves costs. Federated Live Migration makes the transition from DMX to VMAX more attractive and aids that movement. Moreover, keeping up with explosive data growth without increasing headcount will likely require the movement to the cloud and automation, federation and virtualization.

On the unified storage side, VNX is a bold move. On the low end, VNXe represents a more attractive option than Clariion AX as it has more capabilities that might prove attractive to the entry-level market. This Swiss army knife approach (you want it, we can do it) makes it easier for selling partners to meet customer needs.

On the midrange side, VNX is more suited to the world of virtualization than the products that it is replacing, and also doesn't require a customer to declare fealty to either the block or file side of the house. Whichever way the market blows (say, file or block), EMC can meet the need. EMC was not the original proponent of unified storage, but it is alert to what the market demands. It has laid down the gauntlet to its unified storage competitors, and forces competitors that have separate block- and file-based systems to question their need to have something other than unified storage.

EMC's competitors are going to have to issue a call to arms. They may fire short-term salvos, but they are going to have to spend some time thinking about their competitive counter-offensives. Watching the competitive counter-attacks should make for a continuing storage war. Whatever happens, customers are likely to continue to get greater price/performance (however measured) from their storage investments.

EMC is currently a client of David Hill and the Mesabi Group.