The evolution of server virtualization is upping the ante for storage system capabilities. With the flexibility, high performance and reliability required to support these dynamic environments, server virtualization is driving the need to upgrade server, networking and storage infrastructures. According to IDC, the growth of virtual machines is outpacing the growth of physical servers with nearly 20 percent of all servers expected to be virtual by 2010.
For storage administrators, building a reliable storage network optimized for virtualization has been a complex, costly and time-consuming proposition. Traditionally, Fibre Channel Storage Area Networks (FC SANs) and Network Attached Storage (NAS) were designed to be static with large storage volumes connected to application servers via reliable and powerful fixed connections. In particular, NAS solutions are especially difficult to optimize for virtualization—the architecture requiring islands of storage that are scattered around the organization to be disconnected to the server environment when scaled for capacity or performance.
With the proliferation of virtualization, administrators must be able to provision shared storage resources quickly, efficiently and with little disruption to the business to keep pace with the dynamic needs of server environments. A fluid interconnect between environments is needed to support planned and unplanned spikes in traffic load and additional bandwidth requirements from the flexible and highly-utilized virtual systems. Organizations are also seeing an increase in storage volumes and in the number of systems that need to be connected to storage, creating additional scalability issues.
Luckily, iSCSI SANs are delivering the necessary capabilities to meet these challenges, providing organizations with increased flexibility, improved performance and availability, reduced management and enhanced shared storage security. With these advantages, iSCSI SANs are much better suited to support server virtualization than Fibre Channel SANs or NAS solutions which are difficult and disruptive to scale and have traditionally relied on a static, hardware-based fabric.
However, there are clear differentiations between varying iSCSI approaches. Here are some key things that storage administrators need to consider when deploying iSCSI SANs to support server virtualization.
Flexibility
The biggest challenge facing storage administrators is provisioning storage resources quickly and efficiently to keep up with the dynamic demands of the virtual server environment. For example, a file server that is repurposed as a transaction server to support increased retail activity around the holidays will have vastly different storage requirements one day to the next. And it is up to the storage administrator to make sure these dynamic systems have access to storage volumes with the appropriate configuration, bandwidth, permissions and service levels.
Fibre Channel SAN and NAS solutions do not provide the ability to quickly and efficiently provision storage resources, sometimes taking weeks to scale, restripe and configure storage volumes. The process is manual and usually complex, requiring the storage to come offline while administrators reconnect the volumes to servers. In these environments, storage is an inhibitor to a flexible data center architecture, creating a time-consuming, complex and disruptive bottleneck in the organization’s efforts to meet dynamic IT needs.
Instead, administrators should look for iSCSI SAN solutions that have virtualization engines embedded in the storage architecture. These engines automate many basic tasks associated with provisioning including adding storage volumes, restriping the data across nodes, reconfiguring the systems and creating the connections to the server environment. All this can be done automatically or from a central management console, giving administrators complete visibility across the entire storage network. It’s also done without disconnecting storage volumes from the virtual server environment, allowing business operations to continue while the systems are being provisioned or capacity is being added.
This automation dramatically speeds up the process of provisioning storage resources to virtual machines, providing the flexibility needed throughout the entire data center to support end-to-end virtualization strategies. As a result, organizations are armed with the tools they need to react quickly to market demands—whether the demands are changing day-by-day or hour-by-hour.
Performance
Virtualization dramatically improves server utilization rates with up to dozens of virtual applications sharing the compute load on a system. According to most industry experts, the average virtual server is 90 percent utilized compared to just 40 percent for static systems. The more processing each server does, the more storage volumes the system is likely to need access to, increasing both data processing and bandwidth requirements. This explosion in traffic is likely to cause bottlenecks throughout the data center infrastructure, impacting performance, availability and reliability during a time when those service levels are most crucial to the business.
It is important that storage administrators consider these performance demands when selecting a storage system. Because of their flexible, modular architecture, most iSCSI SAN solutions are able to scale performance in line with capacity. Each iSCSI storage node that is added to the storage environment also includes RAM, memory, its own network ports and disk drives, making sure these resources are not taken away from existing nodes already in production. These resources are also pooled across the storage environment, ensuring that administrators can keep up with dynamic traffic loads and varying business demands without degrading performance.
High Availability
High availability is also a concern when optimizing shared storage for virtual server environments. Many organizations prefer to deploy storage across multiple facilities, building redundancy in the storage architecture and protecting the business from a wide-spread data center outage. If one facility is rendered unusable due to a power outage, building fire or regional disaster, systems can be automatically failed over to a remote facility making sure business operations continue unabated.
Fibre Channel SANs solutions make it hard to support this ability to automatically fail over to another distributed facility. Fibre Channel is a hardware-centric protocol that relies on World Wide Name (WWN) tags to link servers to storage, requiring a static interconnect between storage and the server environment. Each time a virtual server is moved to a different machine--such would be the case in a disaster recovery scenario--the connection would have to be severed and reconnected to the new hardware.
iSCSI, on the other hand, uses iSCSI qualified name (iQN) tags to attach directly to the virtual machines rather than the physical server. This means that when a virtual machine is moved from one server to the next—or from one data center to the next—the iQN moves with it, keeping the interconnect intact.
Instead of deploying a high availability multi-site solution that relies on two volumes—one in each location—that have to be recovered and restored, clustered iSCSI SAN solutions simply deploy a single volume that manages two copies of the data—one in each site—that allows seamless failover. As a result, servers in the recovery location never have to be remapped to volumes in case of a disaster. It is a much simpler, reliable and cost-efficient high availability architecture—and one that will enable the virtual server environment to get back up and running much faster.
Security
An on-going concern among IT organizations is how to separate data accessibility between virtual applications on the same physical server. For privacy and compliance reasons, it is important that an administrator or user for one application be prevented from accessing data belonging to another.
Currently, organizations that still rely on Fibre Channel are able to put up a secure wall between virtual applications through a new technology called N_Port ID Virtualization (NPIV). However, this technology requires that organizations upgrade the fabric architecture—including both host bus adapters (HBAs) and fabric switches—an expensive, complex and disruptive option.
iSCSI, on the other hand, has this security barrier inherent in its architecture since it allows for storage to be connected directly to virtual machines, rather than the physical server itself. As discussed before, this distinction helps enable an end-to-end flexible data center architecture, and ultimately, a more robust data security strategy.
Manageability
Since multiple virtual applications can be deployed on a single server, there has been an explosion in the number of virtual servers since the proliferation of virtualization technology. The additional virtual machines in turn need to be connected to storage, leading to an expansion of the interconnect layer. These connections need to be managed carefully to avoid complexity and to ensure they are reliable, robust and secure. More servers also mean more storage volumes, only adding to the complexity.
With an eye on simplicity, storage administrators should consider the manageability benefits of iSCSI, which helps ensure that supporting high-growth virtual server environments does not cause unnecessary complexity in the storage fabric. The modular design of clustered iSCSI SANs gives administrators the ability to add, administer and replace nodes quickly, efficiently and without having to take systems off-line. This capability extends the flexibility and transparency of the server environment to the storage realm, making it easier to allocate storage resources in line with the dynamic needs of the server environment while optimizing the entire IT infrastructure for end-to-end virtualization solutions.
Given the challenges presented by growing virtual server environments, storage administrators are increasingly relying on reliable, flexible, powerful and secure iSCSI SANs. By looking closely at available iSCSI SAN options, these administrators can select the solution that will provision storage resources quickly, simply and efficiently while keeping up with the performance and high availability demands of the higher-utilized virtual environments.
John Spiers founded LeftHand Networks.