Insights, Best Practices and Forward Thinking from the Customer Facing Team, Solution Architects and Leaders on Extreme Performance Applications, Infrastructure, Storage and the Real-World Impact Possible

SQL Server and the WFA: Part 3 - Reducing CAPEX and OPEX through Simplicity

by VIOLIN SYSTEMS on October 21, 2014

Greetings!

We have previously discussed the Windows Flash Array's (WFA) performance and the importance of latency, and in this blog, we'll focus on how the WFA can deliver a simpler solution that helps transform the economics of your data center.

Having high throughput and the low latency are imperative, but they are only part of what's required to make an enterprise-class solution. Transforming the efficiency and economics of the data center, which ultimately yields reduced OPEX and CAPEX, is just as essential for overall corporate success. One way to improve efficiency, and thus reduce expense, is through elegant simplicity of design. In the ideal world, you want to lower costs for your company, improved SQL application response for users, and make your life easier. The WFA achieves this in a variety of ways.

diagram-flash-fabric-architecture

Application performance is a tangible measure of efficiency. One way to improve performance is to get data into server RAM faster so that it can spend more time being worked over by the application as opposed to being stuck in the seemingly interminable I/O wait purgatory. Our Flash Fabric Architecture™ (FFA) avoids the performance bottlenecks common in any disk drive to reduce latency, which in turn means you can support massive concurrency of users at an unprecedented scale. Higher utilization → higher efficiency → reduced CAPEX.

Next, WFA uses SMB Direct, which dramatically increases throughput and thus decreases the time to load data into RAM by shifting much of the I/O processing away from the application servers' CPUs to the network interface and the storage solution. With SMB Direct, you can reduce application servers' CPU overhead by up to 30%. In other words, you now have 30% more processing capacity for your SQL Server workloads, with your existing servers. The combination of our FFA, SMB Direct, and the unique kernel-level optimizations of Windows Storage Server 2012 R2 tuned for the WFA hardware delivers performance that is not available on any other vendor's all-flash SMB Direct solution. Again, higher utilization → reduced CAPEX.

diagram-smb-direct

Given the historic limitations of storage, any serious database deployment was based upon direct attached storage (DAS) or block-based storage area networks (SAN) solutions. The low overhead of block-based storage was able to deliver the performance, at least initially, so best practices dictated the use of DAS or SAN for database implementations. The tradeoff, however, was in the flexibility, simplicity, and ease of storage management. As a result, each application directly or indirectly had to deal with issues of expanding and contracting capacity as needed and a host of other cumbersome storage management tasks.

With the WFA and its support for SMB Direct, you can have simplicity of a file-based environment but without sacrificing the performance to which you are accustomed with block-based solutions. Rather than directly mapping to fixed volumes, you can instead map to shares, whose underlying storage can be easily moved around by admins without changing share names or impacting applications' or users' references to the data. As a result, storage moves up into the Windows world, natively accessible through simple and flexible share concepts, and easily managed within Windows Clusters. You can easily scale your storage while maintaining consistent performance to accommodate SQL Server growth, all within a simpler, easier, (and dare we say) user-friendlier environment. Simplified management → less effort → reduced OPEX.

OK, that all sounds fine, but what about data locality, storage performance tuning, managing hot spots, and all of the other issues that keep DBAs busy at work and sleepless at night? With Violin's FFA, these are quaint vestiges of the past as the architecture renders them irrelevant. Since Violin All Flash Arrays are not based upon disk drives, they are not beholden to the architectural limitations of the past, which makes storage simpler and less risky to setup and manage. There are no disk groupings, data locality issues, performance mapping or tuning required. You can enjoy the simplicity and immediacy of a zero tuning experience. That's right. All of your workloads will perform at the maximum speed of the array, every time, with no storage tuning or up front architecture planning. Your DBAs can cross at least one item off their daily worry list, and you no longer need to renew that expensive data locality software license. A simpler architecture → less effort → reduced CAPEX and OPEX.

Oh, that FFA also means there is ample performance so that you can undertake routine maintenance such as backup, index maintenance, etc. during production hours without negatively impacting primary operations. Moreover, when your maintenance needs include updates to the WFA operating system, Cluster Aware Updating enables you to patch one node at a time so your databases and related applications remain online throughout the process. No downtime → increased ROI.

The last measure of efficiency we're going to discuss is integration and manageability. With management based on the Microsoft toolkit, existing Windows admin staff can provision storage resources. You can rapidly deploy new development, test, and Q/A scenarios with the knowledge that roll out and roll up will be quick and simple, and the test beds will match the speed and architecture of your production environment. You can reduce the number of tools and skill sets needed to maintain your SQL Server environment by using the native tools, i.e., System Center, PowerShell, Active Directory, etc. that you have already deployed in your Microsoft environment. In addition, you will be squarely on Microsoft's technology roadmap for not only your servers and applications, but also your storage and its management. Fewer tools and skillsets → reduced CAPEX and OPEX.

As you can see from this and the previous blogs, we can do a lot in 3U of rackspace to not only boost your SQL Server performance, but transform the economics as well. The combination of a simpler architecture, blazing performance enhanced by SMB Direct, file-based access, Microsoft cluster integration, and administration with the Microsoft toolkit yields a fundamentally different and simpler SQL Server storage solution (pardon the alliteration) that is a true enterprise-class solution. Finally, you can have storage that is so fast that it doesn't require you to change your behavior; you can run tasks during the day while you're at work, so why not invest in the storage solution that allows you to do so?

In the quest for maximum SQL Server performance, you have two paths from which to choose:

  1. The path on which you arrived at your present environment, or
  2. A new approach that transforms not only your SQL Server performance, but your data center economics as well

You already know the results of path 1; simply repeating the same choices will not yield a different result. With path 2, you might be pleasantly surprised by how the WFA could not only increase your SQL Server performance with greater flexibility on a simpler architecture, but also transform your data center economics by reducing CAPEX and OPEX through greater efficiency and utilization.

Cheers!

For an in-depth discussion and more details on how all-flash storage can transform your SQL Server environment check out this webinar. It's long, but you can watch it in slices as time permits.

Learn how Violin and Microsoft collaboratively optimized this solution to leverage WFA's unique performance profile so you can run your SQL Server environment in a Flash.

For more information on the WFA and SQL Server, read this Solution Brief: Accelerating, Virtualizing, and Scaling SQL Server with Windows Flash Array.