Latest News
Newsletter
Home > Storage management for virtualized servers
Server virtualization has often been problematic for data storage administrators, especially when it comes to tracking the relationships between virtualized servers and data storage assets. Many storage management products have been modified for this new environment, enabling users to monitor virtual servers, the applications they host and the storage they’re implementing.
Functional silos
Within each data center, there are typically silos pertaining to particular functional areas, each with a group of staff members dedicated to their management. There are people in charge of managing particular data center resources, including the network, servers, storage systems and virtualization. Each team focuses on managing its specific area and collaborates with other teams when it’s necessary to handle integration points between them. Should a new server need shared storage, the group dedicated to servers works with the storage group to provision storage and present it to the server.
In a traditional physical server environment, the relationships between storage and physical servers can easily be managed by the storage group; a logical unit number (LUN) created on a storage area network (SAN) is assigned to a physical server, and only that server uses the LUN. Things are very different in a server virtualization environment. Storage is perhaps the most critical component of a virtual infrastructure, so it must be implemented and managed properly for maximum performance and reliability. The relationship between server virtualization and storage is a tight one, so the management must be as well.
VMs add complexity to the storage environment
Virtualization revolves around multiple virtual machines (VMs) sharing physical resources. Virtualization file systems, such as VMware’s VMFS, let several physical servers simultaneously read and write to the same LUNs. This is enabled by a special locking mechanism that ensures that multiple hosts have exclusive access to each of their VMs on a shared LUN. Adding to the complexity of the environment are high availability and workload load balancing across virtualization clusters. In addition, VMware’s vMotion and Storage vMotion can move running VMs across hosts and across storage devices, respectively.
On top of the movement of VMs at the virtualization layer, many storage arrays also have an automated storage tiering feature constructed around tiers of devices with differing performance attributes, including solid-state drives (SSDs) and SATA drives that are pooled and presented to the host. The array transports data dynamically across tiers, depending on performance needs. This movement occurs at the storage layer, unbeknownst to the host.
While advantageous, features that allow for the movement of VMs can also cause problems for storage and virtualization managers, because of the dynamic nature of the relationships among VMs, their physical hosts and the storage device in which their virtual disk resides. These relationships have the greatest impact on troubleshooting and performance monitoring. Neither the storage administrator nor the virtualization administrator is able to see what occurs at the virtualization layer and the storage layer, respectively; as a result, neither administrator is able to see the overarching picture.
Source: TechTarget
