.shock - Fotolia
How can we configure a storage LUN for better virtual machine (VM) performance?
Selecting, installing, provisioning and maintaining storage for virtual machines can be a complex and demanding art, but there are some guidelines that can help organizations get better storage performance for virtual environments.
First, it's usually a good idea to create Hyper-V volumes or vSphere data stores that are as large as reasonably possible and employ thin provisioning techniques to minimize the actual physical storage applied to each VM. The idea is that it's easier to simply add physical storage as the thin provisioned volume runs low rather than exhaust the volume and need to re-create a larger storage LUN later. Remember that thin provisioned volumes may impose erratic or unpredictable growth demands, so it's important to deploy monitoring and reporting tools to track storage use and ensure adequate storage is always available to the thin volumes over time. This also allows the business to budget for more storage over time.
When architecting the storage LUN subsystem, consider using multiple paths between servers and shared storage volumes in order to ensure an acceptable level of storage availability. For example, a server may use multiple host bus adapters (HBAs) or Ethernet ports intended for the storage network. In addition, some environments may benefit from tiering or segregating the storage infrastructure. For example, use Fibre Channel HBAs for top-tier high-performance storage, and using Ethernet-based iSCSI or FCoE for lower-priority storage tasks. This prevents the high-end storage network from being bogged down by slower and less-efficient storage devices. The disk array itself should employ a suitable RAID level to ensure reliable operation (such as RAID 5 or even RAID 6). Performance may also be improved by using the latest virtual SCSI and SAS storage drivers.
Finally, be sensitive to the storage needs of the individual application and be sure to group workloads with similar storage performance needs on similar storage tiers. For example, simple enterprise applications can get by with access to SAS (or even SATA) disk groups using Ethernet FCoE connectivity, while storage-intensive VMs might benefit from higher-tier storage like solid-state disk groups on a Fibre Channel storage area network. Without this kind of separation, all workloads would basically get the same level of storage performance (which might be too fast and expensive for low-end VMs or too slow for demanding VMs).
Dig Deeper on Server hardware and virtualization
Related Q&A from Stephen J. Bigelow
Learn how load balancing in the cloud differs from a traditional network traffic distribution, and explore services available from AWS, Google and ... Continue Reading
Access management is critical to securing the cloud. Understand the differences between AWS IAM roles and users to properly restrict access to AWS ... Continue Reading
Containers have rapidly come into focus as a popular option for deploying applications, but they have limitations and are fundamentally different ... Continue Reading