5 Best Practices for Using Local Storage
There's been a revival of local storage use lately. Whether it's a new drive shelf connected to a local storage controller, adding local drives into an internal drive chassis or adding more drives to an existing configuration, I see many virtualization pros (both vSphere and Hyper-V) going this route.
From the hypervisor perspective, there are some compelling reasons to invest in local storage. For one, Microsoft's Storage Spaces and VMware's VSAN offer some new capabilities that weren't realistic options previously.
In addition, solid state drives (SSD), with increasing options and price decreases, have disrupted the space and allowed more local storage. Beyond the SSD, I see a lot of smaller virtualization installations taking advantage of local storage with minimal features. If you're considering adding some local storage, here are a few tips that can improve your experience:
1. Take the time to do a reboot. If you're adding disks or shelves to a system, you may need to do a reboot to get to a low-level control system for adding the new disks or setting up a RAID. What you want to avoid is having the array or drives set up incorrectly, so that on the next boot, there's some other problem you need to fix at the wrong time.
2. Use replication; it saves time. It may seem outdated, but a replicated virtual machine (VM) is a versatile way to move VMs to a new storage resource. For one, you can fail back to the old VM if something doesn't go as expected with the new storage (especially if you're experimenting with high-capacity, low-speed drives). Secondly, it can be done quickly via the failover process if you don't have a built-in migration capability.
3. Don't forget to update automatic power on. If you move VMs around, you should ensure that any automatic "power on" tasks are enabled.
4. If you're using slower storage, monitor before and after. It's tempting to add more drives with very high capacities; 6TB drives are available, for example, allowing lots of room for additional VMs. Be wary, though, since the IOPs don't lie; the number of operations that those drives can do are fixed. Remember that if you fill up slower storage with more VMs, they're all sharing those IOPs. Make sure you can qualify performance before and after migrating to slower storage.
5. Use modern hypervisors. There are any number of pitfalls to avoid in a virtualized environment, many of them can be avoided by using the latest versions vSphere and Hyper-V. Example: There was a time when native filesystems for ESXi and Hyper-V couldn't map a datastore to hard drive arrays larger than 2TB. Also, prior to the VHDX disk format, it wasn't possible to write virtual disks larger than 2TB. This led to a lot of worst practices like dynamic disks and datastore extents, and even Raw Device Mappings. Upgrading to the latest versions of your hypervisors will solve a lot of those issues. In other words, avoid deploying outdated hypervisors like ESXi 4.1 or Hyper-V Server 2008 R2. It may seem obvious, but it still happens.
Local storage is still a viable option for running VMs today, but don't go down the path of expanding a storage option without knowing what to expect. Have you reinvested in local storage? If so, what types of local storage are you using? Share your strategies around local storage below.
Posted by Rick Vanover on 08/18/2015 at 8:36 AM