Block storage options with Azure Disk Storage and Elastic SAN
Secure the best price performance at scale for block storage options with Azure Disk Storage and Azure Elastic SAN, and benefit from Azure’s industry leading SLAs for high availability, resiliency and disaster recovery. Storage is a major component that determines the cost of running workloads.
Azure Disk Storage provides a comprehensive spectrum of block storage options for any workload, regardless of performance needs. This includes the new fully-managed Elastic SAN service in Azure to share storage across workloads in your network and give you a first-time alternative to running your SAN on-premises.
Azure expert, Matt McSpirit, shares how to choose and attach the right OS or data disk storage as you provision VMs in Azure.
Provision faster disks with low capacity.
Premium SSD v2 in Azure — now lets you decouple disk storage capacity from performance requirements. See Azure Disk Storage options.
Run your SAN in the cloud.
Establish shared and consolidated storage, accessible by hundreds of workloads. Check out the first-time alternative to running your SAN on-premises — Azure Elastic SAN.
Lower cost storage for workloads that don’t require stringent SLAs.
Check out Standard HDD and Standard SSD options for Azure Disk Storage.
Watch our video here.
QUICK LINKS:
00:00 — Introduction
00:51 — Workload requirements
02:34 — Premium SSD
04:12 — Premium SSD v2 & Ultra Disk
05:37 — Azure Elastic SAN
08:14 — Low-cost storage options
09:23 — Additional options
10:02 — Wrap up
Link References:
Get started at https://aka.ms/AzureDiskStorage
Check out Azure Elastic SAN at https://aka.ms/AzureElasticSANMechanics
Unfamiliar with Microsoft Mechanics?
As Microsoft’s official video series for IT, you can watch and share valuable content and demos of current and upcoming tech from the people who build it at Microsoft.
- Subscribe to our YouTube: https://www.youtube.com/c/MicrosoftMechanicsSeries
- Talk with other IT Pros, join us on the Microsoft Tech Community: https://techcommunity.microsoft.com/t5/microsoft-mechanics-blog/bg-p/MicrosoftMechanicsBlog
- Watch or listen from anywhere, subscribe to our podcast: https://microsoftmechanics.libsyn.com/podcast
Keep getting this insider knowledge, join us on social:
- Follow us on Twitter: https://twitter.com/MSFTMechanics
- Share knowledge on LinkedIn: https://www.linkedin.com/company/microsoft-mechanics/
- Enjoy us on Instagram: https://www.instagram.com/msftmechanics/
- Loosen up with us on TikTok: https://www.tiktok.com/@msftmechanics
Video Transcript:
-Welcome to Microsoft Mechanics. In the next few minutes, we’ll break down the essential things you need to know about block storage in Azure to help you to achieve the best price-performance at scale, all while benefiting from Azure’s industry leading SLAs for high availability, resiliency, and disaster recovery. As we’ve covered before on previous episodes, storage is one of the major components that determines the cost of running your workloads in Azure. As you provision your VMs in Azure, you need to choose and attach the right OS or data disk storage for your workloads. Here, Azure Disk Storage provides a comprehensive spectrum of block storage options for any workload, regardless of performance needs. And this includes the new fully managed Elastic SAN service in Azure to share storage across your workloads in the network, giving you a first-time alternative to running your SAN on-premises.
-Now, I’ll go through these options in detail in a moment, but first, to balance price and performance, it’s important to understand the requirements of your workload around four main areas. Starting with storage capacity, which refers to the size of your disk and how much data you need to store as you run your workload. As a rule in Azure, the larger your disk capacity, the higher the potential performance, although there are options that let you decouple capacity and performance, which we’ll get to in a moment.
-Then, beyond choosing the right capacity, it’s all about optimizing for performance. This means understanding how many IOPS or input/output operations per second you need. A highly transactional app with data frequently written to or read from disk storage, for example, will need far more IOPS than an app that processes data primarily in memory. In parallel, you need to think about your throughput, which determines how much data can be processed. Smaller sized operations require less throughput compared to large ones. Finally, critical to performance is latency, which is how long it takes operations to access the disk storage to read the data or write to it.
-As a managed service, latency can range from several milliseconds to single digit and sub millisecond performance, depending on the disk storage option you choose. If you aren’t sure about your workload requirements, you can monitor VM metrics for your workload. This can help you to gauge the average IOPS and throughput for your app, spot the times when your IOPS peak, and see how long those peaks last as well as how often they occur. That said, acceptable performance is workload dependent. For example, if it’s a mission critical or a production workload, you may want a consistent level of performance with low latency. Whereas if your workload is accessing storage less frequently, consistent performance and low latency may be less important. This determination will help you make the right price performance trade offs.
-So, with this context in mind, now let’s go deeper into available storage options. The majority of your workloads, including business critical ones, will benefit from two main options, Premium SSD and Premium SSD V2. For your IOPS intensive apps that don’t need sub-millisecond latency, Premium SSD provides the best balance of cost and performance. It’s ideal if you need high capacity and performance. Premium SSD allows you to choose from different disk capacity sizes up to a maximum of over 32,000 gibibytes and 900 megabytes per second in throughput. The larger the capacity, the higher the potential performance.
-The relationship between disk size and performance is an important one to understand. For example, if your app consistently needs 5,000 IOPS, you’d need to purchase and provision one tebibyte of capacity to achieve the performance you need. That said, there are several scaling options. First, you can accrue credit if your runtime performance is below provisioned, which allows you to burst up to 30 minutes for free to account for any peaks in your IOPS. Next, if you need more than 30 minutes and have more than 512 gibibytes in capacity provisioned, you can enable on-demand disk bursting which allows you to pay for burst performance when you need it for infrequent or unpredictable traffic spikes.
-Alternatively, you can also take advantage of performance tiers for planned performance scaling where you need sustained higher performance for a few hours or days like a Black Friday event. Additionally, as your data needs expand, you can also resize to larger disk capacity dynamically, without downtime. Now specifically for your data disks, this is where the latest generation of Premium Disk storage, Premium SSD V2, comes in. It delivers almost double the capacity of Premium SSD with a maximum disk size of over 65,000 gibibytes and provides sub-millisecond latency. But there’s an important difference.
-This option lets you decouple your disk storage capacity from your performance requirements, which means you don’t have to provision larger disk sizes to get the performance you need. You always get a baseline performance of 3,000 IOPS and 125 megabytes per second included. And you can dynamically scale up or scale down your IOPS and throughput as needed without downtime. In this case, up to 80,000 IOPS and 1,200 megabytes per second in throughput. This option suits a broad range of production database workloads such as SQL Server, Oracle Database, big data analytics, stateful containers, SAP workloads, and more.
-Now, of course, going back to our decision tree, if your workloads require higher IOPS and throughput and low sub-millisecond latency, then the Ultra Disk option provides the fastest and most flexible choice for the cost. It’s capable of a massive 160,000 IOPS and 4,000 megabytes per second of throughput, which can be provisioned in minutes. This makes it ideal for top-tier databases and transaction heavy, latency-sensitive workloads. Performance can be dynamically scaled to meet demand without disrupting your workloads. Next, if you have large on-premises SAN appliances and are looking for a cloud-native solution, with Azure Elastic SAN, you can establish shared and consolidated storage that is accessible by hundreds of different workloads. This eliminates the need to figure out exactly what disk type is needed for each workload. Azure
-Elastic SAN in fact is the first industry alternative to running your SAN on-premises, and there are many more advantages. First is the time to provision. Instead of the time it takes to acquire and configure hardware for your SAN locally, in Azure, you can set up and provision your SAN in minutes. As a fully managed service, all the infrastructure configuration is taken care of for you. Just like many Azure Disk Storage options, redundancy is built in. You can choose from locally redundant storage where three copies of your storage volume are stored locally in one datacenter or zonal redundant storage where those three copies are spread across different data centers within an availability zone, which gives you added resiliency should one datacenter go down.
-And you have the flexibility to easily and cost effectively optimize the size and performance of your SAN. To achieve this, first you select your base units, essentially the more you have, the more performance and storage you get. In this case, up to millions of IOPS and double digit gibibytes per second in throughput. And once your performance needs are met, at any point in time, you can increase your storage capacity separately, which is significantly less expensive than adding more base units.
-Once you’ve provisioned your SAN, you create a volume group followed by your volume. A volume group, which is a management construct, lets you easily define your network configurations and encryption settings. Inherent in these policies is your volume. Think of these like individual disks or storage units that you run workloads against. Your workloads can be attached to a single volume or stripe across multiple volumes for their performance and storage needs. And because Azure Elastic SAN uses the iSCSI protocol, those workloads can run on a broad range of compute resources in Azure.
-That said, perhaps the biggest advantage of Azure Elastic SAN is the ability to dynamically pool performance across your workloads. For example, you could have a 100 tebibyte SAN with 500,000 IOPS and a volume group with 10 one tebibyte volumes which are capable of 64K IOPS each or 640,000 IOPS in total. That said, if all 10 volumes were maximally loaded, it would be capped at 500,000 IOPS which is the size of the provisioned SAN. However, it’s unlikely that each volume would max out at the same time. Maybe three would be loaded at capacity while others have varying loads. The key here is that performance is pooled and balanced to support the workloads that need it.
-I began by explaining the options for your production level and mission critical workloads, but of course, there are lower cost storage options for workloads that don’t require stringent SLAs. Now, where your IOPS and throughput requirements are less demanding, and consistent performance and low latency isn’t an issue, Standard HDD is going to be the most cost efficient and inexpensive option. Here, IOPS and throughput are variable with up-to ranges that max out at 2,000 IOPS and 500 megabytes per second in throughput. This option is best where your workloads don’t require frequent data access as write latencies are under 10 milliseconds while read latencies are under 20 milliseconds. This, for example, may work great for a dev/test or back-up workload.
-That said, for single-digit millisecond latencies, your next option is our entry-level Standard SSD. In fact, Azure leads the industry with this SLA for SSD disks. Here, the performance range is higher with up-to ranges that max out at 6,000 IOPS and 750 megabytes per second in throughput. With this option, just like with Premium SSD, you get credit-based bursting for up to 30 minutes for free to handle peaks in your IOPS. In fact, this can work well for your less frequently used web servers or lightly used small databases.
-So those are your storage options, but there are a few more things worth mentioning. First, Premium SSD, Premium SSD V2, Standard SSD, and Ultra Disk storage options can be shared, which means that you can connect them to multiple VMs. This helps you to run clustered or high-availability applications much more cost effectively. Second, once you understand your disk storage requirements and you’ve decided on your preferred storage option, you need to choose a matching VM IOPS and throughput performance level to avoid capping your storage IOPS and throughput. Using the Azure Virtual Machine selector, you can specify your requirements along with the technical specifications of your VM. And the tool will identify the best VM and disk storage combination for your needs.
-So those are essential things to know about block storage options with Azure Disk Storage and Azure Elastic SAN to help you to achieve the best price performance at scale while benefiting from Azure’s industry leading SLAs for high availability, resiliency, and disaster recovery. To learn more, check out aka.ms/AzureDiskStorage and aka.ms/AzureElasticSANMechanics. Keep watching Microsoft Mechanics for the latest in tech updates. Subscribe if you haven’t already and thanks for watching.
