PostgreSQL® and Planning for Those Unknown Events 

“You won’t need it for every instance. But when you do, you’ll notice the difference right away.” 

Sure, this quote may be squarely about Instaclustr Managed PostgreSQL on Azure NetApp Files (ANF) but the underlying lesson certainly rings true for countless other (read: non-technical) situations as well: 

• Going for a hike on a trail you already know like the back of your hand? You probably won’t need a compass, but it could prove rather handy if you become disoriented 
• Driving around in snowy conditions all winter? You won’t need a sleeping bag now, but you’ll certainly want one if you get stranded while driving in a blizzard.  
• The kids have all had their lunch which should hold them over until dinner? Better have a snack ready just in case (because no one likes to deal with a hangry child). 

It’s a relief to know you have a ready-made solution in case things go haywire. (Take it from me, having grown up in in a place where it snows half the year, you absolutely want that sleeping bag in your car just in case as the alternative is rather chilly).  

Having a logical backup plan in place for a known situation is one thing (see: “sleeping bag,” above); establishing a catch-all remedy for an endless variety of situations is both impractical and prohibitively expensive. This is especially true for a sprawling database management system like PostgreSQL. 

So, here’s the problem: We’re in the age of cost optimization—higher interest rates, uncertain macroenvironment, AI, and the bourgeoning of new, expensive technology, to name a few—and that’s not going away anytime soon. But this certainly doesn’t stop our PostgreSQL databases from growing larger and more complex by the day.  

How can we maintain top-notch performance and control over our PostgreSQL instances–but still be fully prepared for those unknown scenarios–while somehow managing to reduce costs? (show me one CFO who encourages their IT department to spend more company money and I’ll show you the Loch Ness Monster). Is that even possible? 

That’s exactly where Instaclustr Managed PostgreSQL on Azure NetApp Files comes into play: Faster speeds, better reliability and control, and best of all: substantially lower transaction costs. 

Just as our engineer said, you won’t need ANF for every PostgreSQL instance. But when you do, you’ll notice the difference right away. (And your CFO will notice the cost savings, too). 

What Is Azure NetApp Files? 

Based on NetApp’s industry-leading ONTAP storage management software, Azure NetApp Files is a cloud-native file storage system ready to handle the most demanding file workloads with sub-millisecond response times. Providing the highest performance and lowest latency enterprise-grade file service in Azure, ANF comes with built-in availability, rapid disaster recovery capabilities and data protection—all offered on a pay-as-you-go (PAYGO) consumption basis on Azure, so you’re never stuck over provisioning.  

Incredible power aside, ANF was designed with flexibility and scalability in mind. Users can easily create and manage high-performance, highly available, and scalable file shares, quickly and seamlessly replicating data across all regions.  

While ANF works great for a number of use cases and offers a ton of other cool features, we noticed 3 intriguing things:

1. High throughout for large file transfers,  
2. Ability to handle many random read and write operations with high concurrency,  
3. Extremely low latency made it a perfect fit for database management. 

These gave us an idea: we already run our Instaclustr Managed PostgreSQL instances on Azure Premium Disk, so why not try running instances on ANF? Given the power of ANF, we had a pretty good idea of what to expect running it on our Managed PostgreSQL platform 

But even we were surprised by the results. 

Developing Azure NetApp Files for Instaclustr Managed PostgreSQL 

There are several reasons why so many organizations use PostgreSQL as their database of choice—its seemingly-endless extensibility, ease of use, a strong and vibrant community of contributors, among many others—but one reason stands out above all: It just works. As databases go, PostgreSQL is relatively straightforward to use. And with its unparalleled extensibility, it’s easy to fine-tune your setup to match your use case. 

However, challenges still abound: 

• Organizations want the financial benefits and agility of moving to the cloud but replicating what they have already built into their on-prem data centers—SLAs, capabilities, and protection—is way too cumbersome and costly. 
• After moving to the cloud, organizations had to over-provision their PostgreSQL due to storage layer limitations, leading to a substantial increase in spending on VMs–completely vanquishing all cost savings from moving to the cloud in the first place. 
• Normal cloud backups are taking far too long—often 20 hours or more and well outside the time window allotted—resulting in massive delays and downtime, slowing overall performance, and negatively impacting business. 

We know that customers want to reduce infrastructure TCO for their most demanding applications and avoid adding operational complexity like the plague, so we designed PostgreSQL-ANF to be simple to set up. 

Like, incredibly simple: when spinning up a cluster on the Instaclustr Managed PostgreSQL platform all you need to do is check the Azure NetApp Files Premium NFSv3 option.  

Setting up Azure NetApp Files is done with just the click of a button. 

That’s it. Boom. Done. It’s your same PostgreSQL instance—now backed with the power of Azure NetApp Files. 

The beauty of PostgreSQL is that it’s such a straightforward piece of software to manage despite the massive and demanding data loads that it can handle; adding any sort of operational complexity with ANF was always a non-starter, which is why we kept it as simple as possible (and without sacrificing any power). 

Performance Testing: Azure NetApp Files vs. Azure Premium Disk 

As part of the release for PostgreSQL-ANF, we upgraded the VM hardware from Dsv2 to Esv4. This had no impact on the premium disk performance, but it did come with the added benefit of slightly lower cost. 

Using pgbench to run our tests, we compared 2 instances with each running 1.5TB of data (a real-world representation of a production workload): 

• E8s_v4 instance running on Microsoft Premium SSD 
• E8s_v4 instance running on Azure NetApp Files 

After a bit of tuning and adjusting our parameters (more detail on that in our white paper), here’s what we found in our head-to-head testing for Read/Write workloads and Read-Only workloads: 

• Read/Write Workloads
  • The increased IOPS available on ANF showed a stunning improvement on the PostgreSQL application over Azure Premium Disk, including:
    • 167% faster TPS at 8 clients
    • 125% faster TPS at 32 clients 

• Read-Only Workloads 
  • The increased IOPS available on ANF was even more dramatic for Read-Only workloads than Read/Write workloads, including:
    • 325% faster TPS at 64 clients
    • 193% faster TPS at 16 clients

No matter the client counts, the results were always the same: Azure NetApp Files TPS performed significantly faster for both Read/Write and Read-Only workloads.  

In the end, there really wasn’t much of a comparison when it came to speed. Azure NetApp Files was the clear winner—but how does this impact total operating costs? 

Cost Breakdown 

Incredible performance is one thing, but we all know that benefit is immediately wiped out if it comes with a prohibitively expensive price tag (and what’s the point of having all that extra power if no one can afford it?).  

Given how much faster and more powerful ANF is compared to Azure Premium Disk, we calculated the monthly costs to be nearly the same. While you do pay more upfront for storage costs with ANF (which is to be fully expected as you’re making substantially more transactions), this is practically offset by the lowered instance cost.  

But the most striking change is the dramatic reduction in costs per TPS when using ANF: 

• Read-Only $/TPS
  • Azure Premium Disk: $0.67
  • ANF: $0.17 
• Read/Write $/TPS
  • Azure Premium Disk: $1.37
  • ANF: $0.64 

That’s more than a 50% reduction for Read-Only and 70% for Read/Write transactions—substantial cost savings that only grow larger as your workloads bigger and more demanding.  

Now throw in the benefits of a managed service—offloading the day-to-day work of managing your data infrastructure and freeing up engineering resources to focus on scaling your application—and achieve an even greater TCO. 

It’s Time to Reconsider a PostgreSQL Managed Service 

I get it: PostgreSQL is such a straightforward, easy-to-operate software that there really has never been much reason for organizations to switch to a managed service if they weren’t already using one. There must be a bigger reason to offload than just cost optimization (although that’s a pretty good reason in itself). 

Organizations demand a drastically enhanced performance experience, too, and Instaclustr Managed PostgreSQL on Azure NetApp Files delivers that on both fronts—and more: 

• High Availability: Up to 99.99% availability SLA means that downtime and disruptions are kept to an absolute minimum 
• High Throughput: Large file transfers are fully supported, ensuring your read and write operations are handled with high concurrency 
• Scalability: Quickly—and easily—scale up or down to meet the performance needs of your specific workloads as needed, and ANF automatically provisions the necessary throughput 
• Automatic Failover: If there’s disruption or a maintenance event, ANF supports automatic failover within a bare-metal fleet, ensuring your data is always available—even in failure 
• Snapshots: Create point-in-time snapshots of your data, drastically reducing your storage consumption as only the changes made since the last snapshot are captured 
• Support for Availability Zones: Deploy volumes in your Availability Zone of choice, allowing you to build high availability architectures  

But above all, you’re guaranteed the ultimate in flexibility: as a purely 100% open source solution, Instaclustr Manged PostgreSQL on Azure NetApp files means your data is never locked in. You have the freedom to control where and how you use your data, and never pay a single licensing fee.  

Power. Flexibility. Savings. PostgreSQL can do a lot of things extremely well, but planning for those unknown scenarios is challenging (if not outright impossible) due to so many things that can (and often do) go wrong. 

In a perfect world, IT departments would have an unlimited budget to give their databases the incredible power and unlimited storage to tackle whatever problem came their way.  

Welcome back to reality, where things aren’t perfect–and IT budgets are far from unlimited. That doesn’t stop organizations from demanding more power, enhanced flexibility, and increased savings—but that’s exactly what PostgreSQL-ANF delivers: 

• Power: 3x faster than before 
• Flexibility: quickly scale up or down when your workload demands it, giving you the utmost control of your database 
• Savings: TPS costs up to 75% lower than Azure Premium Disk, and no licensing fees with 100% open source 

Use that incredible power when you need it. Don’t pay for it whenever it goes unused. It’s as simple as that. 

We can’t solve every “what if” scenario, but Instaclustr Managed PostgreSQL on Azure NetApp Files comes pretty close.  

You certainly won’t need it for every instance. But when you do, you’ll absolutely notice the difference—and your CFO will notice the TCO savings too. 

Ready to experience PostgreSQL-ANF yourself? Try it out for free and spin up a cluster in minutes or reach out to our friendly Sales team and see how we can help with your use case. 

Coming up next: how to integrate NetApp BlueXP Classification into Instaclustr Managed PostgreSQL on Azure NetApp Files—and drastically improve your data governance abilities with the power of AI.