This article can also be found in the Premium Editorial Download "IT in Europe: Object storage: An elementary approach to file structure."
Download it now to read this article plus other related content.
Virtual servers and storage systems don’t have to exist in separate worlds; new tools and plug-ins provide single-console management of both virtual servers and storage.
Virtualized servers have created plenty of problems for data storage managers, not the least of which is keeping track of the relationships between data storage assets and virtual servers. Some storage management products have adapted to this new environment, allowing users to keep track of virtual servers, the apps they host and the storage they’re using.
Inside every data center there are typically silos related to specific functional areas, each with a dedicated group responsible for management. There are teams responsible for managing specific data center resources, such as the network, servers, storage systems and virtualization. Each group focuses on managing its own area and works with other groups when needed to handle integration points between groups. If a new server requires shared storage, the server team works with the storage team to get storage provisioned and presented to the server.
In a traditional physical server environment, the storage group can easily manage the relationships between storage and physical servers: 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. Server virtualization changes all that. But storage
VMs can complicate storage
Virtualization is about the sharing of a common set of physical resources among many virtual machines (VMs). Virtualization file systems like VMware Inc.’s VMFS allow many physical servers to read and write concurrently to the same LUNs. This is possible because of a special locking mechanism that ensures multiple hosts have exclusive access to each of their VMs on a shared LUN. Among server virtualization’s strengths are its features that provide high availability and workload load balancing across a virtualization cluster. Features such as VMware’s vMotion and Storage vMotion can move VMs while they’re running from host to host, as well as from one storage device to another.
To further complicate things, the movement of virtual machines on storage devices doesn’t just occur at the virtualization layer. Many storage arrays now have an automated storage tiering feature built around tiers of devices with different performance characteristics such as solid-state drives (SSDs) and SATA drives that are pooled and presented to a host. The array dynamically moves data across tiers based on performance demands. All of that occurs at the storage layer and the virtual host is unaware of the movement.
While the features that move VMs around are beneficial, they can cause headaches for storage and virtualization administrators as the relationships among virtual machines, the physical hosts they’re located on and the storage device where their virtual disk resides is a dynamic one. That will have the most impact when troubleshooting problems and monitoring performance. Because the virtualization admin is unaware of what’s occurring at the storage layer and the storage admin doesn’t know what’s happening at the virtualization layer, neither one gets to see the big picture.
This was first published in June 2011