Everything old is new again: Virtualization in 2005, Part II
Everything old is new again: Virtualization in 2005, Part II
By Mark F. Ewert published: Thursday, March 17 2005
Last month we explored the state of System and Datacenter Virtualization in 2005 – this month we review the two other major areas of the virtualization market: Application and Storage.
Application Virtualization
More than ten years after the invasion of personal computers into the modern workplace, businesses are still searching for the best way to manage and control the desktop. Application Virtualization provides yet another potential solution for IT Departments faced with the daunting task of providing customized workstations and applications in pace with ever changing business needs. Offering many of the capabilities once promised by such technologies as Citrix Metaframe and Microsoft Terminal Server, Application Virtualization enables technologists to centrally manage and deploy tightly-controlled, standardized applications and workstation computing environments.
Despite the obvious benefits, creating and maintaining completely standardized, up-to-date computing infrastructures is very difficult and in many cases simply not feasible for many organizations. For numerous business reasons, many IT departments must maintain multiple, typically incompatible versions of applications and
application configurations. As business needs seemingly change daily, IT departments must also be able to rapidly deploy and update applications without risking production. Modern businesses employ a shifting landscape of contractors, business partners, and remote employees all of whom may require different applications and
levels of infrastructure access.
Several years ago, Citrix Metaframe and Windows Terminal Server provided technologists the ability to deploy tightly controlled Windows workstation computing environments running on specialized, multi-user versions of Windows Server. Instead of running applications or even leveraging the operating systems installed on end-user PCs, Citrix Metaframe and Terminal Servers hosted instances of Windows 2000 or XP Professional for each user on the servers themselves. End-users then utilized either Thin-Client Terminals or stripped down Windows PCs running client-connectivity software to remotely access their instance of the Windows OS and applications running on the physical servers. As this shifted end-user application and operating system configuration from numerous distributed PCs to fewer centralized servers, it promised to reduce the cost and amount of time required to craft, maintain and control customized end-user computing environments. By removing the reliance on highly customized end-user PCs by reducing desktop requirements to merely the software or thin-client hardware required to connect to the server infrastructure, IT Departments also hoped to slash desktop support costs.
Unfortunately the Metaframe and Terminal Server solutions had several drawbacks that diminished or in many cases nullified any infrastructure cost or control benefits. Most notably, the shifting of applications and end-user operating systems from workstations to servers required IT departments to invest in extensive training and much more expensive Systems Engineers to handle the vastly more complex server configurations required. As the servers both hosted complex server software and provided end-user workstation and application environments, engineers had to take great pains to ensure each new application or configuration change did not affect other hosted applications or the performance and stability of the physical servers themselves. The consolidation of end-user computing environments from numerous PCs to fewer, centralized servers also dramatically increased the risk of user downtime. Without taking expensive steps to build, monitor and maintain highly fault tolerant, multi-server Metaframe or Terminal Server environments, the outage of a single server typically affected dozens of end-users. Metaframe and Terminal Servers also did not provide a good solution for organizations requiring different workstation and applications for different classifications of users, as each variation required dedicated server infrastructure to host it.
Application Virtualization provides solutions to address all of these drawbacks. Unlike the Metaframe and Terminal Server approaches, most Application Virtualization solutions aimed at end-users, such as those offered by Softricity and VMWare, do not require expensive server infrastructure. Instead, they provide IT departments a better way to centrally create and manage complex applications (Softricity) or workstation operating system configurations (VMWare ACE), but run them on distributed end-user workstations. Despite offering a way for organizations to replace PCs with thin-client terminals, Metaframe and Terminal Server were never able to supplant the deeply entrenched personal computer. By running virtualized applications on each end-user’s PC, Application Virtualization leverages investments in workstations and negates the need for complex server infrastructure. This also nullifies the downtime risks associated with Metaframe and Terminal Server since each user depends upon their workstation and not a centralized, multi-user server for their business computing environment. Shifting configuration requirements from critical servers back to the desktop also reduces the caliber and cost of IT resources required to support the user community.
Products like Softricity’s SoftGrid utilize a virtualization layer running on each PC to isolate virtualized applications from both the underlying workstation operating system and each other. By containing application components such as .dlls (dynamic linked libraries) and Windows registry settings that often conflict with each other within this virtualization layer, the Softricity solution enables IT departments to deploy applications without concern that they might interfere with each other or affect the stability of user PCs. This approach even enables users to run different versions of major applications such as Microsoft Office simultaneously, as is often required to support older Microsoft Access database applications for example.
Other solutions, such as VMWare’s ACE (Assured Computing Environment) instead of virtualizing just the applications create an entire virtualized operating system and application environment which runs on a virtualization layer installed on each user’s PC. Through VMWare ACE, end-users utilize this virtual operating system to run their standardized business applications, instead of accessing these applications installed directly on their underlying PC. As only a functional VMWare ACE virtualization layer is necessary to run the user’s computing environment, technicians can deploy very simply workstation configurations to the desktop that can be rapidly re-deployed in the event of workstation hardware failure.
Both the Softricity SoftGrid and VMWare ACE solutions provide IT Departments ability to centrally create and manage end-user computing environments running on distributed PCs at a fraction of the cost or complexity of the Metaframe and Terminal Server solutions; or even running the applications directly on the workstations themselves. Most significantly, both solutions enable IT departments to craft standardized end-user computing environments in a central test environment and then deploy identical copies of these configurations to the virtualization software
running on each distributed PC. Softricity’s SoftGrid is specifically focused on centralized virtual application deployment and management as it is able to automatically deploy applications and application updates leveraging the Windows Active Directory for user identification and deployment security. VMWare’s ACE does not provide such automated functionality, instead relying on administrators to replicate VMWare ACE virtual machines to distributed PCs using essentially any network file replication service.
By simplifying the task of centrally creating standardized end-user computing environments, both VMWare ACE and Softricity SoftGrid not only negate the need for armies of application technicians to physically visit PCs for software installations and updates, they enable adherence to application standardization by eliminating repetitive installation configuration mistakes and preventing end-users from affecting the virtualized environment with other software installed directly on their workstations.
Softricity SoftGrid and VMWare ACE also easily accommodate the business need for diverse computing and application configurations for different classifications of employees and business partners. Instead of continuously customizing individual PCs for different needs, technologists simply create and distribute different virtualized workstation or application environments to individual users or user groups as required. If different applications are needed, IT departments simply deploy different standardized virtual configurations to the specific desktops without having to risk changing the software directly installed on the workstations themselves.
Although other technologies have promised to simplify desktop and desktop application management, none have come as close to fulfilling this promise as Application Virtualization. In 2005, more IT Departments increasingly familiar with System and Storage Virtualization will begin to utilize Application Virtualization as a key tool in the never-ending battle to meet business workstation and application demands. To help meet this burgeoning demand, more vendors will undoubtedly enter this new segment of the virtualization market, providing technologists an increasingly diverse array of Application Virtualization solutions.
Storage Virtualization
Storage Virtualization is currently one of the hottest areas of interest in the technology marketplace. Faced with an ever growing amount of data required to be accessed by an increasingly diverse number of users, applications and platforms, IT managers are hopeful rapid developments in storage virtualization and management will settle into cohesive, viable solutions for their storage woes.
Storage Virtualization is slightly different from the three other types of virtualization covered in this series of articles, as it is less the application of old virtualization concepts to new technologies as it is simply the latest iteration of a decades old evolution of storage management, combined with exciting advances in storage capacity and connectivity.
In this first quarter of 2005, all major enterprise storage vendors are offering some form of storage virtualization. These vendors include venerable giants like IBM and Sun Microsystems, storage specialists Hitachi and EMC, storage-market newcomers such as EqualLogic and network-giant Cisco Systems.
The challenge for all storage virtualization vendors in 2005 is to maintain market distinction while meeting their customer’s interoperability needs. Vendors must also clearly demonstrate, not simply promise, a return on investment (ROI) on what is often a revolutionary restructuring of an organization’s storage and data.
In order to understand the storage virtualization market it is important to understand what storage virtualization is, and what it isn’t. Storage Virtualization in its most ideal form is technology implemented in hardware and/or software that masks the underlying technical complexity, incompatibility and geographic location of disparate storage systems with a cohesive, unified and heterogeneously accessible virtual storage platform. This enables organizations to utilize their distributed investments in storage, previously trapped in islands of underutilized servers or on incompatible file and storage systems, as a central, cohesive pool of storage resources which can be efficiently allocated and simultaneously accessed by heterogeneous computing platforms such as Windows, Linux and Unix.
By enabling storage managers to create virtual volumes composed of physical disks located in different storage arrays, storage virtualization reduces the amount of physical disk space trapped in underutilized storage systems. Physical disk space remaining on one array can be combined with space on another to create a virtual storage volume. Since these virtual storage volumes exist between the physical storage systems and the computing platforms utilizing them, they also enable previously incompatible systems to access and utilize the same physical storage. For example, by enabling a farm of Windows and Linux Web and Application Servers to efficiently utilize the same storage systems, system architects require less physical storage and are able to leverage the unique applications and strengths of each computing platform without duplicating data.
Implemented effectively, the efficiencies introduced by Storage Virtualization yield immediate gains in the utilization of existing storage investments, offer a potentially significant reduction in future storage expenditures and enable once disparate computing platforms to access the same data.
It is important to understand, however, that Storage Virtualization in its pure form is not a specific physical storage platform, system, or protocol. In the past, most vendors’ storage virtualization solutions have required that specific vendor’s disk arrays, certified controllers and storage switches. Ideally, Storage Virtualization should not be tied to one vendor’s specific equipment. Indeed, requiring organizations to wholesale replace existing investments in storage to move to a vendor’s storage virtualization platform significantly decreases ROI gains. In 2005, organizations are increasingly recognizing the need for enterprise-wide, interoperable storage solutions capable of leveraging their existing investments in multiple vendors’ storage technology.
Current product offerings from most storage virtualization vendors recognize this trend. IBM provides its TotalStorage SAN Volume Controller, TotalStorage SAN File System and TotalStorage Productivity Center as components of its TotalStorage Open Software Family. The SAN Volume Controller enables technologists to combine SAN storage systems from multiple vendors including EMC, Hitachi and HP into a cohesive, virtualized storage system. The SAN File System then allows this virtual pool of heterogeneous storage to be accessed simultaneously by heterogeneous computing platforms including Windows, Linux and Unix. Finally, the TotalStorage Productivity Center enables simplified centralized management of the virtual and physical storage infrastructure.
Hitachi Data Systems offers the TagmaStore Universal Storage Platform. Named Product of the Year by both Storage Magazine and SearchStorage.com, the TagmaStore is a crossbar storage switch (a high-performance switch fabric that supports multiple simultaneous point-to-point communications between individual storage network devices), disk array and intelligent storage management software platform in one. Implemented logically between an organization’s heterogeneous computing platforms and its multi-vendor storage systems, the TagmaStore Universal Storage Platform is able to aggregate up to 32 petabytes of heterogeneous SAN, NAS and even mainframe storage into a cohesive virtual storage environment.
Sun offers a very similar product with its StorEdge 9900 system. In fact, the StorEdge 9900 series is based on Hitachi’s Lightning 9900 SAN and Storage Management product line. Last year, both companies announced a strategic partnership to share storage experience and technology, which will undoubtedly benefit customers of both storage vendors. As a further example of this partnership, both companies simultaneously announced support for EMC Symmetrix DMX 800, 1000, 2000, 3000, Symmetrix 3000, 5000 and 8000 and EMC Clariion storage systems in their Storage Virtualization Systems.
EMC is notably the furthest behind in providing products to virtualize and manage heterogeneous storage systems. EMC's Storage Router offering is not scheduled to be released until the 2nd quarter of this year. Running as intelligent software within leading SAN switches from vendors such as McData, Cisco and Brocade, the EMC solution promises similar functionality as the Hitachi/Sun solution at a fraction of the cost. However, as this product is yet to be released, specific details on pricing and the brands and models of compatible third-party storage systems are not available, making it difficult to evaluate any potential cost benefits.
Cisco is taking an approach similar to EMC with its MDS-9000 Series of multilayer SAN switches and MDS-9000 SAN OS 2.1 software. A relative newcomer to the storage market, Cisco is utilizing decades of heterogeneous network expertise to create high-performance storage networking switches running a highly-flexible storage operating system capable of connecting SANs from multiple vendors. By supporting numerous storage connectivity technologies and protocols including iSCSI, Fibre Channel, FICON (IBM Fiber Connection), and FCIP (Fibre Channel over IP), Cisco will likely carve out a healthy position as a leading provider of intelligent storage network solutions. To help bolster this strategy, Cisco based its SAN OS software on its famous IOS router software, reducing the learning curve for technologists already well familiar with Cisco’s network router technology. Cisco also designed its MDS-9000 family to support third-party storage applications. MDS-9000 components are already being combined with IBM and Veritas storage solutions to provide greater SAN compatibility and flexibility.
EqualLogic, another relative newcomer to the storage market, is the only notable vendor not focusing on virtualizing heterogeneous storage environments. Without the vast resources of its competitors mentioned in this article, EqualLogic is focused on doing what it does well: providing a soup-to-nuts SAN, storage management and virtualization package. As their solution includes all required components including hard drives, SAN switches and SAN software, EqualLogic did not have to focus on supporting numerous protocols and connectivity technologies. Supporting only iSCSI, the EqualLogic solution is best suited for organizations either without large investments in storage systems or who are looking for a less-expensive SAN solution than IBM, Hitachi/Sun or EMC to replace existing storage.
Once a highly proprietary area of the storage marketplace, Storage Virtualization in 2005 is promising a new phase of openness, offering IT Managers and solution providers incredible flexibility in efficiently managing and utilizing enterprise storage. This openness will certainly be put to the test throughout the next year as more and more organizations make investments in SAN and Storage Virtualization technologies. If the trend continues, by the end of 2005 storage may very well become the most optimized segment of the modern enterprise computing environment.
Within the two articles in this series we reviewed the four major segments of the Virtualization Marketplace: Systems, Data Center, Application and Storage. In a matter of only a few years, virtualization has moved from an obscure mainframe concept to the frontline of enterprise computing. Over the next several years the number of technologies and innovative solutions leveraging virtualization will dramatically increase and become commonplace throughout the typical computing landscape. The challenge for future innovators is to combine these four diverse types of virtualization into a cohesive, enterprise-wide virtual computing environment. Only then can true utility computing or “applications on demand” be achieved.
*****
Mark F. Ewert, a Freelance Solutions Architect, has been working with information technology for more than 25 years, with seven years hands-on experience with virtualization. Specializing in heterogeneous computing on every scale, Ewert won the 1996 NetWorld\InterOp Network Design Contest. He has designed and implemented hundreds of successful solutions for organizations of all sizes, including Fidelity Investments, the BBN Planet division of GTE Internetworking, and the State of New Hampshire Department of Health and Human Services.
