What is Thin Provisioning and Should You Utilize It?

Since the advent of computing, humans have generated a lot of data. So much data that we are now living in what is being hailed as the Zettabyte Era – a time when we’ve collectively generated zettabytes of data. And in case you didn’t know, a zettabyte is 1 sextillion bytes or, to put it into numbers, 1,000,000,000,000,000,000,000 bytes. In fact, according to Statista, about 181 zettabytes of data will be created by the year 2025.

As any cloud user knows, the more data that you generate, the more storage you use. And the more storage you use, the higher your cloud bill is. Luckily, there are ways to keep your cloud capacity to a minimum. In this post, we’ll look at thin provisioning – how it works, what challenges you might encounter, how it might affect your cloud performance, and determine if it is right for you.

But first: What is thin provisioning and how does thin provisioning work? Thin provisioning is a method of allocating storage based on user requirements on-demand. As the user consumes their allocated space, they are allocated more. If they delete files or transfer data, the freed-up space is reclaimed and repurposed for other workloads. Division of storage space is virtual and flexible so there is no need to determine allocation in advance.

This contrasts with thick provisioning – a typical method of storage allocation on-prem. In thick provisioning, a set amount of space is reserved for each user as the disk partition is created. Let’s illustrate this point simply: Imagine a company that has 100GB on the network and each user needs to be allocated up to 50GB. With thick provisioning, only two users could fit on the drive — Even if both users end up only leveraging 10GB of space each. The remaining 80GB would remain unusable to others as they’ve already been allocated. Conversely, with thin provisioning, the unused 80GB could remain open to users for use.

What is the Advantage of Thin Provisioning?

Thin provisioning is one of the many great data services that can help enterprises get the most out of their cloud storage. For IT environments that use multiple hosts and applications, thin provisioning speeds up cloud storage provisioning, simplifies management, and reduces costs. There is a higher level of storage utilization since thin provisioned virtual disks take up less space and unused disk space can be repurposed for other needs. In turn, this means that there’s less of a need to purchase additional storage since you’re able to get more out of your current cloud resources.

While thin provisioning is more flexible than thick provisioning, both can still help account for growth. While thick provisioning requires that you plan ahead of time what your future capacity needs might be in order to avoid the time and headache of quickly adding more if growth doesn’t go as expected, thin provisioning allows you the flexibility to add as your needs change on-demand. Not only that, but it’s much faster to create thin provisioned virtual disks than thick disks.

Challenges of Thin Provisioning to Consider

With that being said, if you don’t adequately prepare for growth, you may find that thin provisioning can leave you scrambling. Capacity shortages can result in crashes and performance issues. Depending on how critical to the business the applications are, you might need to consider overprovisioning storage space – as a fail-safe in case additional storage is needed quickly. Overprovisioning means wasted and unnecessary cloud budget dollars. Overprovisioning is seen as less of a problem with thick provisioning. You set the disk size at the time of creation so you predictably know how much space you have available from the get-go, accounting for overprovisioning.

Another challenge of thin provisioning is the inability to automatically shrink the disk size when data is deleted from the virtual disk. If you want to reduce the size of the disk, you will need to do so manually.

The Relationship of Thin Provisioning to Performance

So how does thin provisioning affect performance? By shrinking your capacity footprint through the use of thin provisioning, you can increase the efficiency of your cloud resources. Meaning you can increase how much capacity is being used without investing in additional cloud resources.

Let’s look at a furniture and home goods ecommerce company that was able to reap the benefits of thin provisioning and other data services with Silk. The company had a corporate objective and strict deadline to get all of its data out of their on-prem data centers and into Google Cloud. The IT team started migrating their dozens of Microsoft SQL Server hosts to the cloud but kept hitting throughput limitations no matter how many cloud resources they assigned to them. With the deadline bearing down, they didn’t have time to refactor all of their Microsoft SQL Server databases. They needed a way to get faster performance and streamline the deployment of their cloud resources with a common configuration for each SQL instance (instead of every instance being customized based on peak workload requirements).

With Silk, the company was able to easily exceed their throughput requirements and get better performance with latencies that were even lower than their previous physical storage setup. This is due to Silk’s ability to dynamically scale performance as needed. And with Silk’s enterprise data services – including their thin provisioning capabilities – the company was able to standardize their mission-critical Microsoft SQL Server deployments without worrying about massively overprovisioning cloud resources. With Silk, the company saw 1.9:1 data reduction for a total reduction of 27.9:1.

How to Determine if Thin Provisioning is Necessary

So how can you tell if thin provisioning is the right choice for your organization? If your cloud footprint – and in your turn, your cloud bill – is ballooning out of control, then you need to invest in thin provisioning and other enterprise data services such as snapshots and data reduction capabilities. Unfortunately, while these types of data services are taken for granted in on-premises environments, they are often not a standard feature in the public cloud. Which means that you’ll need to invest in a third-party platform, like Silk, in order to achieve these data services.

But that isn’t the only way that Silk helps boost your cloud performance while reducing your costs. The Silk Data Virtualization Platform lives between your applications and the underlying public cloud storage infrastructure. It separates the layers for performance and capacity so that you never have to overprovision on cloud resources just to hit IOPS and throughput performance targets. The platform connects with compute virtual machines (VMs) over a higher performance compute network instead of the limited capacity data network of cloud infrastructure. This allows Silk to support the most performance-intensive workloads while eliminating the need to oversize compute VMs for better data network performance. Bonus: by rightsizing your compute VMs, Silk gets rid of the need to overpay for unnecessary database licenses since license costs often grow with the number of vCPUs in your compute VMs. Since Silk intelligently offloads tasks from network to compute, you see consistently dramatic reductions in latency and the max responsiveness of your applications.

Alongside thin provisioning, Silk’s suite of enterprise data services includes instantaneous, zero-footprint snapshots. These snapshots do not require a fully provisioned copy of the data, which makes it easy to access copies for various purposes (whether Dev/Test or Disaster Recovery) without paying for redundant data taking up valuable storage space. On top of helping to reduce costs, Silk’s snapshots are a great example for how Silk can improve the manageability and resiliency of customers’ data in the cloud.

One company who has taken advantage of Silk’s snapshot capabilities is Sentara. A not-for-profit healthcare organization serving Virginia and North Carolina, Sentara was looking to move its Electronic Health Record (EHR) system to Microsoft Azure. One issue they were looking to solve during this process was reducing downtime in the EHR’s SQL reporting engine. This reporting engine would go offline for 7-10 hours a night as it went into ETL. During that time, data was inaccessible to providers and patients who urgently needed it.

With Silk, not only was Sentara able to easily adopt Microsoft Azure for its EHR system, but it was also able to find a way to work around the SQL reporting engine’s ETL downtime. By taking an instantaneous zero-footprint snapshot of the database every night and mounting it to two other Microsoft SQL Server hosts, the team was able to reduce downtime from 7-10 hours per night to less than 15 minutes, giving end users nearly 24/7 access to their data.

This is only one example for how Silk helps companies improve the manageability and resiliency of its data in the cloud. Silk also boasts a self-healing architecture that proactively tracks cloud maintenance windows to avoid disruption as well as native machine learning-based monitoring that anticipates issues and concerns and proactively addresses them before they occur. Finally, Silk provides top resiliency with an active-active architecture that separates cloud management across zones and regions to eliminate single points of failure. So you can rest easy knowing that Silk is keeping the proverbial lights on, that you’re getting the fastest performance possible at the most cost efficient price point.