In the cloud, numerous services can all too easily consume an IT budget. Cloud storage, specifically block level storage volumes, can play an outsize role in that and can quickly become burdensome and difficult to manage on an on-going basis.
On the one hand, the cloud has brought about instantaneous creation and delivery of block storage, on the other hand, the cloud has instantaneous delivery. This is a blessing and a curse. All too easily, storage can be provisioned that is unused or over spec’d.
When considering different disk types, cloud providers offer a dizzying array of options, dedicated drives with provisioned IOPs, solid state drives, hard drives and cold storage, to name but a few. On top of that, each cloud instance typically might have three or four disks attached. With this scale and complexity of options, it’s no wonder that storage disks oftentimes sit untouched for months or years accruing greater and greater cloud bills.
How can one manage this dynamic target of optimal tiering for a constantly changing number of disks?
It starts with the continuous collection of critical metrics that drive decision making. For block storage, tracking IOPs (input/output operations per second) and throughput for each storage disk is the starting point. However, it is not so simple. Here are some of the dimensions and considerations as one optimizes cloud storage disks for performance and cost.
First consider what tier to select. The cloud providers offer many options, and they don’t simple lead to more IOPs and throughput. Some scale capacity and IOPs independently, others require an increase in disk capacity to deliver more IOPs, some have uptime service levels, others charge for transactions. All these considerations can create quite the headache when solving for this multi-dimensional problem. The only way to garner trust is by making the right storage tier selection every time.
Second, understand the impact of the action itself. When and how can I make the change to the storage and application? Certain changes in storage tier are reversible, depending on the original and new tier others cannot be revoked. At the same time, some actions require downtime, while others do not. Scaling storage tiers back and forth without downtime reaches the pinnacle of pure elasticity, the constant pursuit of meeting resource demand while spending the least amount of money possible.
Here at Turbonomic, we help our customers reach that ultimate pursuit. Every hour of every day, Turbonomic answers the question, what storage tier is best? To do so, it continuously collects the percentile utilization of IOPs and throughput and scans all available storage tiers and sizes to select the optimal one with pinpoint accuracy. Of course, not all storage tiers are available to each instance. The instance might not support a certain disk type, or not all disk tiers are available in a certain region. Even the volume itself influences what storage tiers are available, depending on if it is a boot or data disk. All these considerations are seamlessly factored into the analysis.
Recommending the correct tier is one thing. The next step is acting on that decision. Turbonomic can do so with a simple click of a button. As shown above, a user can quickly review the recommendation, act on it, and realize the performance improvement or savings. Even better, Turbonomic can streamline this process through automation by acting on these storage disk changes on a user’s behalf. A user simply flips on automation for some or all storage disks, potentially during a set recurring window, and Turbonomic takes the pending scale actions. Now the storage disks are continuously set to the optimal tier without any tedious calculations, manual analysis or time-consuming reviews.
The complexities of storage disk tiering can be overwhelming, which leads to apathy on the individual costs of storage disks. However, that can quickly lead to regret when a monthly, quarterly or yearly cloud bill comes due. Turbonomic is here to help manage this problem at scale, tackling the analysis, scheduling pending actions and automatically making changes enabling you to rest assured that your cloud storage is optimized.