Cisco’s introduction of the Unified Computing System (UCS) and its entry into the server market sparked a new arms race in the IT industry and has forced the heavy-hitters — storage, server and networking vendors alike — to come up with new strategies for data center convergence.
The benefits of unified computing platforms are clear: centralized management and provisioning of virtual resources on-the-fly while reducing physical infrastructure and overall cost.
The potential drawbacks, however, cannot be overlooked. According to industry analysts, users face the threat of vendor lock-in and storage vendors have a decision to make: pick a side.
Arun Taneja, founder and consulting analysts with the Taneja Group research and consulting firm, concedes that there are advantages to unified computing, but he’s very leery of the downside.
“Certain advantages accrue when you have that level of control over the vertical stack, from the application to the spindle, but the negatives far outweigh the positives,” says Taneja.
IT managers are accustomed to buying best-of-breed products and are confident that they have some leverage at the negotiating table. However, Taneja is fearful that power may evaporate if unified computing platforms take hold.
“It took us 25 years to move to a best-of-breed scenario in the data center. Now there are established standards for the interfaces between applications and systems. The moment you go the other way, a vendor would have enough power to say ‘I’m going to do something proprietary between these components or subsystems to get a 25% increase in performance.’ Users get locked-in in that scenario,” says Taneja.
Storage vendors represent an important piece of this puzzle as they may determine which way the pendulum swings.
“If the shift to unified computing happens, storage vendors are going to have to pick sides, and by picking sides they essentially become subservient to the larger vendor,” Taneja said. “What the big vendors are saying to storage vendors is: We are going to create the platform and you better decide which side you’re on.”
Each of the vendors making a push into unified computing has a piece, or pieces, of the stack and is plugging portfolio gaps with technology from partners.
Cisco’s UCS combines compute, network, storage, and virtualization resources in a single system based on a line of blade and rack-mount servers developed by Cisco.
Cisco’s servers are based on Intel’s Nehalem processors and feature “extended memory technology” to support applications with large data sets and virtual machines (VMs). The network connectivity for UCS is provided by Cisco’s Nexus family of switches, and is based on lossless, 10 Gigabit Ethernet (10GbE), enabling a unified fabric that consolidates LAN and SAN traffic on a single wire, which reduces the number of network adapters, switches and cables and decreasing power and cooling requirements.
The system can access storage via Fibre Channel, Fibre Channel over Ethernet (FCoE) or iSCSI. Additionally, users can pre-assign storage access policies for system connectivity to storage resources.
Cisco has recruited EMC, NetApp and VMware, among others, to provide the storage and virtual machine technologies that round out the UCS platform’s capabilities.
HP retaliated with its Converged Infrastructure Architecture, which includes a set of associated services and partner offerings aimed at creating a virtualized, on-demand data center.
The Converged Infrastructure Architecture integrates and virtualizes compute, storage, networking and management resources based on several HP technologies, most notably BladeSystem Matrix, FlexFabric and Virtual Resource Pools.
Using HP’s Virtual Resource Pool technology, modular standards-based systems can be provisioned on-demand to support enterprise, cloud and high performance computing (HPC) applications.
The underlying storage for Virtual Resource Pools is supplied by HP’s StorageWorks arrays, including the StorageWorks X9000 Network Storage System family, a series of scale-out systems based on technology from the acquisition of Ibrix.
The most recent entrant into the fray, Hitachi Data Systems (HDS) is planning to ship the Hitachi Unified Compute Platform in early 2011. Like the others, the HDS platform offers the combined management of compute, networking, storage and applications. The management layer will be made possible through an OEM deal with Microsoft, under which HDS will distribute Microsoft Windows Server 2008 R2, the Microsoft System Center suite and Microsoft SQL Server 2008 software.
The Hitachi platform will combine HDS software, storage and server products with Microsoft’s System Center infrastructure software, Hyper-V virtualization technology and SQL Server database, as well as networking products to serve up an automated, virtualized, “cloud-ready” platform, according to Miki Sandorfi, HDS’ chief strategist for file and content services.
The hardware includes the Hitachi Universal Storage Platform V (USP V), Hitachi blade servers, and Fibre Channel and Ethernet networking gear from HDS partners.
Dell has a different take on the next generation data center. Dubbed Virtual Integrated Solution, or VIS, Dell’s offerings include storage hardware, software and data management services based on open systems technology.
The solution can be composed of new or existing server, storage and networking technologies tied together with an offering called Dell Advanced Infrastructure Manager (AIM). The AIM solution dynamically allocates hardware and software resources based on workload demands. For example, VIS may include Dell servers, Juniper switching and Dell-EMC storage all tied together with AIM.
Dell is bringing to bear stacks made up of hardware, software and services aimed at managing internal IT and cloud computing environments. For Dell, it’s about tying legacy systems together with the latest server, storage and systems management technologies with an open systems approach.
“It’s easier for the OEM to build a fully integrated system, but the industry doesn’t slow down. If we made a fully integrated, Dell-only solution it would pretty quickly no longer be best-of-breed,” says Paul Prince, CTO of Dell’s Enterprise Product Group.
Oracle appears to be the one vendor with the entire IT stack in its arsenal, including applications, middleware, virtual machine technology, servers, storage, and database and operating system software.
Oracle’s offering – the Exadata Database Machine – is a pre-packaged hardware and software system complete with operating system software, virtual machine (VM) technology, and servers and storage from Sun.
Oracle’s president, Charles Phillips, has made it clear that the company is out to change the way IT is engineered, integrated and delivered to customers.
“This industry has a long history of building systems in a very labor-intensive, manual way.,” Phillips said. “They are very unpredictable, very unreliable and expensive to maintain and implement. Why don’t we do all that up front at the factory?”
Phillips believes it’s time to get back to the ways of IBM in the 1960’s by building complete compute environments. “We are going to recreate that reliable environment from years ago, but we are going to do it with an open systems platform.”
If Phillips’ view on the industry is correct, customers might be ready to embrace the one-stop-shop all over again. Tight purse strings always play a role in the purchasing process and the prospect of discounts through consolidation is alluring, especially in times like these.
According to Andrew Reichman, a senior analyst with Forrester Research: “There is definitely a desire to aggregate purchases. Consolidating more purchases may [yield] better discount percentages and support. Infrastructure is not black or white. There are shades of lock-in no matter what you do. The key is to be aware of it and keep an exit strategy in mind.”