Top Storage Tips for SME/SMB
Storage is undergoing a pivotal transformation caused by the combined pressure of virtualization on performance and the ever-shrinking IT budget. The biggest challenge of managing storage is that the combination of storage and server virtualization, while creating great flexibility and resiliency, also creates layer upon layer of abstraction which is difficult to understand usage and diagnose storage performance issues. This article intends to discuss those challenges from the storage perspective and offer guidance on how to manage your storage environment.
Virtualization and Storage
The big story is virtualization and storage and how well they play together. On the one hand, virtualization capabilities like thin provisioning and snapshots allow you to maximize your storage utilization while maintaining ready-to-go backups. However, all the goodness comes with a cost – random IO. Also known as the IO blender, virtualization effectively negates traditional methods to speed up arrays (like caching algorithms). Here are a few tips to help you balance between storage and virtualization.
Use thin provisioning to maximize usage on your arrays. Thin provisioning allows a host/server to use only the amount of storage from the array that it currently needs, rather than the amount initially allocated to it. This allows you to use the array capacity much more effectively because you can over-allocate the storage to multiple hosts, but the true storage used is just a fraction of the allocation – initially. However, too much thin-provisioning can be a bad thing as the more hosts sharing the storage, the more risk you introduce. Generally, you shouldn’t overallocate more than 50% of the real capacity of the storage. Thin provisioning can be done at the hypervisor level or the array level – you can do both, but it is not recommended.
Snapshots are a great technology that allows you to keep point in time copies of your VMs ready to go at a moment’s notice. They can also be easily copied or backed up without impacting the running VM. Note – snapshots are not backups! If you run your environment that way, it will bite you if you have a disk or array failure and don’t have copies on other media or devices. Snapshots can be done at the hypervisor or the array level as well.
To help handle the impact of virtualization of storage, a new API has been developed called VAAI. This technology allows for better communication between the hypervisor and the storage array, improving performance many times over. If you can upgrade your current array to leverage VAAI or are about to purchase a new array, this should be on your checklist. Most array manufacturers already support VAAI, so unless your budget is severely restrained, this is a no brainer.
Don’t work yourself into a frenzy about the type of technology - Block vs. NFS, Fibre Channel vs. iSCSI vs. FCOE – you really can’t go wrong with any of them (although array and switch vendors won’t say that). If you are considering a technology refresh, you will definitely want to take a deeper dive to understand which one is better for you.
Reduce Primary Storage Usage
The amount of primary storage you own is going to drive your costs – how many disks you buy, how much data you have to back up, how much electricity you are going to use, etc., so effectively managing it is your first line of defense against snowballing costs.