Most enterprises are in the process of virtualizing their physical infrastructure or have already done so. The purpose of the transformation is to use underlying physical resources optimally and reduce capital expenses as well as operations expenses. However, companies need to consider some SAN management factors while
Requires Free Membership to View
transforming and during the operational phase so that application
performance is not impacted. Here are some things to keep in mind from a SAN management perspective
in virtual server environments:
- Identify the environment bottlenecks. As a single server will be hosting multiple virtual
machines (VM), ensure that you install more than two host bus adapter (HBA) cards on the server.
This SAN
management best practice ensures that the host to storage path does not become an application
performance bottleneck.
- Use an aggregation of measured workload to determine the storage protocol, redundancy
protection and array features to use, rather than an estimate. The best results come from measuring
your application’s I/O throughput and capacity for a few days, prior to moving them to a
virtualized server environment.
- Virtual provisioning should be used to supply storage to the physical servers. This will
optimally utilize storage, and it’s possible to see performance benefits. The logical unit number
(LUN) provisioned from virtual pools will stripe on all the physical disks in the pools, therefore
leading to more physical spindles in the backend.
- HBAs installed on the physical servers should use the latest firmware. As a good SAN
management practice, always install the latest HBA drivers.
- Refer to compatibility guides of various virtualization vendors, and install only compatible
HBAs.
- Install multi-pathing software on a physical server recommended by the storage vendor.
- Many storage vendors have included plug-ins to provision storage for virtual server
environments. Using such SAN
management plug-ins ease storage provisioning.
- It’s a good SAN management best practice to pool volumes created from the LUNs presented to
physical servers, especially in virtual environments. For example, create four LUNS of 100 GB from
FC 15000K RPM disks, four LUNS of 100 GB from FC10000K RPM disks, four LUNS of 100 GB from 7200K
SATA disks, and assign them to the server. You can create volume on the FC 15000K RPM LUNS and use
them for critical application VMs. Similarly, create volume on FC 10000K RPM LUNs and use them for
less critical applications. Volume created on 7200K SATA LUNs can be used for least critical
applications as part of this SAN management strategy.
- If you currently have one or two LUNs for the 10 or 20 VMs shared by your servers, you may
notice remarkable improvement by simply redistributing those VMs over four or six LUNs. If you have
a VM that is particularly I/O bound, you should consider moving its virtual disks to a dedicated
volume.
- If possible, fiber channel over Ethernet (FCoE) should be deployed for better SAN management
and performance in virtualized server environments.
- ‘Boot from SAN’ option should be used in all the servers for high fault tolerance.
- Many server virtualization products offer thin provisioning features, but it is recommended to
use the thin provisioning at the storage levels. Use of thin provisioning from server
virtualization software along with storage thin provisioning is not a recommended SAN management
practice.
- HBA vendors have configuration files, which need to be edited for better performance and
interoperability with different storage vendors. Information about tuning these SAN management
parameters is easily available on the storage vendors’ Websites.
- Single initiator zoning should be used throughout the virtual environment.
- Naming convention should be followed for creating SAN management zones, so that the name
identifies which server and storage the zone is meant for.
- Always backup the active zone set before committing any changes.
- In a virtual environment, make sure that the HBAs are compatible, and from a single vendor. This is essential to avoid interoperability conflicts.
| Anuj Sharma | |
About the author: Anuj Sharma is an EMC Certified and
NetApp accredited professional. Sharma has experience in handling implementation projects related
to SAN, NAS and BURA. He also has to his credit several research papers published globally on SAN
and BURA technologies.
This was first published in May 2011