The Storage Challenge for Virtual Machine Environments
Server virtualization technologies are being adopted at an accelerating pace with higher server productivity and more responsiveness as the primary benefits. However, storage virtualization technology has lagged that of server virtualization, presenting several challenges for storage administrators. For example:
- Management complexity – Virtual server environments can run hundreds, or even thousands, of virtual machines (VMs). In a typical environment an administrator would provision storage and I/O every time a VM is created, requiring complex management of many individual volumes and LUN connections.
- Storage I/O bottlenecks – When multiple VMs simultaneously access shared data this often creates I/O bottlenecks that lower performance.
- High cost of storage – Many times the reaction is to “throw hardware at the problem” since disks are perceived as inexpensive. While the cost per GB of disk drives continues to fall, the systems and software to manage them are unfortunately still expensive. Over-provisioning and segregating data in storage silos only compounds this problem.
- Availability – Mission critical applications are now commonly deployed on VMs; storage needs to support high availability for mission critical VMs.
The Gluster Solution
Gluster excels at storing and managing file data, and VM images are just that – files. Gluster provides a virtual storage solution that complements VM deployments by addressing the common drawbacks. The Gluster Storage Platform aggregates disk and memory resources in a unified global namespace, simplifying management, delivering excellent performance, and ensuring high availability. We do this with the superior economics of open sources software combined with commodity hardware
- Simple, centralized management – Gluster provides a unified global namespace; a single volume that can be accessed by thousands of VMs with automatic data distribution and I/O load balancing.
- Scale-out to eliminate storage I/O bottlenecks – Storage nodes are scaled horizontally in the storage cluster, automatic I/O scheduling and load balancing are used to spread data evenly across the cluster and avoid choke points.
- Lower storage cost – open source software plus commodity hardware plays an important role in controlling costs as it does throughout the datacenter. Additionally, storage can be consolidated and managed under the global namespace, enabling VM images and application data to be stored on the same infrastructure. This increases utilization and simplifies management.
- High availability – Gluster Storage Platform employs file replication to ensure VMs can survive in the event of a hardware failure. Replication is coupled with sophisticated self-healing to detect and correct VM image errors while running and without pausing I/O operations.