What happened?  Everything seems faster all of the sudden!  Read on to see why...

Previously in this series I wrote about our how we had uplifted our core software infrastructure to Windows Server 2016, System Center 2016, and Exchange 2016 and foreshadowed some big changes coming soon to the hardware infrastructure.  Every few years we refresh our baseline hardware for servers and storage.  We tend to this simultaneously so that we are implementing a complete solution from as few vendors as possible using components that are certified to work with each other.

We are now in the process of phasing out our current traditional SAN hardware with all-new, all-flash arrays.  The new storage infrastructure has a network throughput of 200 GB/s and is capable of supporting up to 85000 IOPS on the backend.  This new array infrastructure will entirely replace the storage used for our production Hyper-V, SQL, and File services.

We are implementing entirely new 14th generation Dell servers and retiring all existing hardware servers supporting Hyper-V, SQL, and File services.  This could not have happend at a better time.  Recently, the Meltdown and Spectre CPU vulnerabilites that we mitigated a few months back by applying new BIOS updates to all our existing infrastructure resulted in a measurable impact on performance (as was expected).  We opted to apply these updates and live with a slight performance impact as it is always better to be safe than speedy.  We also knew we wouldn't have to live with it too long either as the new servers we would be moving too were less affected by the BIOS updates than our existing servers.  On top of that older Windows operating systems were more impacted than the newer Windows Server 2016 and we already had a migration project underway for that.

The migration is happening in two phases. 

The first phase is a migration of our Hyper-V virtualization infrastructure to the new flash arrays and servers.  That phase is now complete.  All customer virtual machine workloads have been moved to the the new hardware.  The migration was completely transparent and required no downtime or changes to client configurations.  The impact was immediate and measurable.  The performance of both the arrays and the new servers far exceeds that of the legacy hardware even before the Spectre and Meltdown patching.

The second phase is the migration of our SQL Server instances and File services to the new flash arrays and servers.  This phase is going to be a bit more intensive than the Hyper-V migration and as our existing SQL Servers were already highly-tuned juggernauts with moderate utilization it will likely have a less noticeable impact on performance than the Hyper-V migration did.  That said the new SQL Server server hardware has double the RAM and CPU capabilities of the existing hardware and the network bandwith is being massively enhanced so those more resource intensive workloads should see a noticeable change.  In preparation for the move to the new hardware we will be eliminating our SQL Server 2008 R2 instances as they have been out of mainstream support since 2014 and extended support will expire in less than a year.  We will be contacting any customers still on this legacy platform and working with them to migrate their solutions to newer versions of SQL Server where possible.  Once the migration is fully complete we will have only SQL Server 2012 (which is supported through 2022) and higher.  This phase is actively underway and is expected to take several weeks if not months to complete.

Stay tuned for more on this in the coming weeks!