Storage virtualisation mitigates SAN complexity

SANs bring tangible benefits such as better connectivity, improved performance, flexibility and scalability. Yet, these benefits may be outweighed by the added complexity that a SAN brings to the table. Akhtar Pasha finds that storage virtualisation is the answer to that particular problem

Avijit Basu, marketing manager-Network Storage Solution, HP India says, “Storage, there’s never enough of it.” With businesses generating oodles of data year after year islands of data are sitting on servers and other secondary media, creating a management nightmare for companies and their IT administrators. Shailesh Agarwal, country manager-Storage, IBM India says, “With the proliferation of SANs, customers are actually seeing the benefits of performance, connectivity and distance flexibility. But now they are realising that they are not able to fully exploit SANs to dynamically manage storage resources.”

According to Shailesh Agarwal, customers are seeing the benefits of SANs, but they are also realising that they are not able to fully exploit SANs to dynamically manage storage resources

According to an IBM internal report, storage capacity is growing at a rate greater than 50 percent per year, but the ability to manage that storage is clearly lagging behind. Although price per gigabyte in storage hardware has plummeted, those savings are easily offset by the added management burden that results from having to cobble together storage systems and devices utilising proprietary methodologies. The lack of centralised management across server, storage and operating system platforms adds to the complexity, resulting in poor IT resource utilisation. Problem identification and resolution is often slow, painful, and costly.

This raises a question—is it still viable for companies to use the traditional approach of discretely allocating storage devices to application servers on a one-to-one basis or should they use Storage Virtualisation (SV)?

SANs lead to SV

In terms of growth, enterprises, especially in the BFSI segment, telecom, oil & gas, which were the early adopters of networked storage (SAN) are slowly realising the importance of storage virtualisation. Today, there are a handful of enterprises that face this problem out of the 25-30 large SAN implementations in the country. But tomorrow as complexity grows with companies having storage subsystems from different vendors, the problem will become more pervasive. As enterprises deploy SANs they will also need a SV solution. By 2004-05 around 60 to 70 percent of enterprises will be using some form of virtualised storage to take full advantage of SAN technology.

Managing storage growth

As an example, large enterprise SANs usually contain different types of storage devices in terms of the disks deployed, their performance levels or the functionality provided, such as RAID or mirroring. Additionally, enterprises may have storage devices from several vendors as a result of mergers or acquisitions. The result is that administrators need to configure storage to servers and keep track of which servers own or have access to what storage. Storage administration can be daunting as a SAN grows and storage administrators attempt to manage it manually.

These are the issues that enterprises want to solve with storage virtualisation. Data migration from an older storage subsystem to newer storage boxes can be hampered by lack of interoperability. Enterprises are demanding solutions that help them move data from applications that do not require a high-performance storage box to cheaper disk, based on policies. Agarwal says, “SV will help you add new functionality. For example, say an enterprise has 5 TB of data in a storage box. To do a flash copy, it needs 5 TB of data to be kept on the same box, hogging precious storage space. What an enterprise can do instead is to move the flash copy data to a low cost disk and use that for flash copy.”

Dr P Sambath Narayanan, storage and HPC Architect at Sun Microsystems India adds, “Replication of data becomes complex in a disaster recovery scenario if an enterprise is using say an EMC storage box in the primary site in Mumbai and a HDS box at the secondary site in Chennai.” Enterprises such as BPCL and HDFC that have disaster recovery sites in more than two locations will face problems such as these. Additionally, customers are asking for solutions that let them add more disk space to the network without bringing the application down or allow them to schedule the movement of applications to high performance storage boxes during their peak period without disrupting the network.

Step into the future

Each year capacity is growing by 50 percent and hardware costs are dropping by 30 percent but availability requirements being 99.99 percent. Enter storage virtualisation. Storage virtualisation addresses the increasing complexity of managing storage and will reduce associated costs dramatically. Its main purpose is the full exploitation of the benefits promised by a SAN. Virtualisation will become an enabler for sharing data, ensuring higher availability, providing disaster tolerance and improving performance. It will allow for consolidation of resources, provide policy-based automation and automated back-ups.

Rana Dutta, regional director-Asia Operations, Movinture Storage Networks, a storage integrator says, “Storage virtualisation separates the physical from the logical aspect of data storage. It combines different storage devices from various vendors into a logical or virtual storage device. Operating system, applications and users work with the virtual device, and don’t need to consider the limitations of physical storage.” In a nutshell, virtualisation gives you a new way to access, manage and use existing storage resources.

Storage virtualisation ensures that data is never tied to particular hardware devices—it can be anywhere. Removing the limits of individual devices, storage virtualisation facilitates a flexible storage environment. At the same time, you can use storage resources to the fullest extent.

When should virtualisation be done?

Storage virtualisation can be implemented in three different ways, based upon the architectural viewpoint—host-based/server-level, storage-based and finally, network-based. Companies design and build solutions according to their idea of virtualisation. Narayanan of Sun Microsystems says, “Each storage vendor is defining storage virtualisation to his own strengths.”

l Server-level: Abstraction at the server-level is by means of logical volume management of server operating systems. This is where a player like Veritas operates. Agendra Kumar, country manager, Veritas says, “SV is a high growth area for our Volume Manager product with customers such as ICICI Bank, the Bharti Group and the NSE. Volume Manager supports heterogeneous platforms—Unix, Windows and Linux and multiple storage devices from all vendors.”

l Network- /fabric-level: The IBM strategy is to move storage device management intelligence out of the server, reducing the dependency on having to implement specialised software such as a Logical Volume Manager (LVM), at the server-level. IBM has two offerings for this market. The first is a storage appliance called SAN Volume Controller (SVC). This is a dedicated bit of hardware with its own software that can be connected to SAN switch. SAN Integration Server (SIS) is the other option that is offered as a pre-configured combination of storage, SVC and switch for companies setting up new storage infrastructure. ING Vysya Bank is using IBM’s SVC solution. HP is offering Continuous Access Storage Appliance (CASA) at the network layer that can connect mixed storage environments, increasing capacity utilisation. CASA offers auto-RAID functions. Basu says, “We are aggressively pushing CASA in India and our target verticals are banking, telecom and extended manufacturing.”

Anil Valluri, director-Systems Engineering at Sun Microsystems says, “We recently introduced the PSX1000, a storage box that can aggregate heterogeneous storage boxes and create a storage domain that can be presented to a SAN so that servers in the network see a logical virtual environment for storage systems. Our strategy will be to go after companies that have implemented SANs.”

l Storage subsystem level: Disk storage systems can provide some level of virtualisation by sub-dividing disks into smaller virtual drives. Conversely, more storage devices could be consolidated together to form a larger virtual drive.

Virtualisation allows users to add storage capacity using inexpensive, commodity disk, tape drives and to dynamically manage those storage resources as virtual storage pools without worrying about where the data is physically stored on the back-end.

Bottom line

While not a near-term panacea for current storage management ills, storage virtualisation promises a real-time view of storage operations, enabling flexibility, interoperability, ease of storage management and scalability. SV will impact business. Better, more accurate and real-time knowledge of the enterprise storage portfolio’s live operations will enable a more agile IT infrastructure to deliver a higher business RoI.

akhtar@expresscomputeronline.com