Guest Blogger Series: Our Move to Hyperconvergence - Part 5/5
Jul 19, 2019
As we close out this guest blogger series with the final post from David Roden, Director of Technology at Goodell, DeVries, Leech & Dann, LLP, we want to convey a special thanks. Thousands of customers have made the move to hyperconvergence with Scale Computing but most have not chronicled their experience nor done it in such an eloquent way. We hope to hear more about David's adventures with hyperconvergence and Scale Computing in the future. Without further ado, here is David's fifth and final blog post in this series.
Our Move To Hyperconvergence, Part 5: Everything but the Kitchen Sink
Ok. We made it. All of our VMs, and all but 1 of our physical servers have been migrated to our new Scale Computing Hyperconverged Infrastructure. 34 VMs in all. That last physical server will eventually move too, but it has become the network junk drawer and we’ve decided to clean it out before we move it.
What We Expected
When we started on the path to HCI we heard lots promises from every vendor. When we selected Scale Computing, we believed we were gaining a higher performing virtualized platform with built-in replication and DR capability. We believed we would save money by eliminating the need for several backup and DR tools that supported our Hyper-V environment.
What We Got
Now that we have implemented Scale’s HCI solution, we have a higher performing virtualized platform with built-in replication and DR capability. Scale provided a stable, comprehensive system that addresses our needs for supporting a virtualized environment. We have a system that is easy to manage. And we are saving money.
We used 3 conversion methods during our move to Scale:
1. By far, the simplest and most obvious method was to export a VM from Hyper-V and import it into Scale. Uptime is certainly important to us, but over several nights and weekends we moved the majority of our VMs this way. Scale has a KB article that outlines a few simple steps to complete the process and update the required drivers to ensure full performance benefits to the newly converted VM.
2. Some of our VMs are quite large. Exporting, then importing that quantity of data would have required a little more downtime than we prefer. For these VMs we used the Move utility, a specialized version of DoubleTake (now a Carbonite product) optionally available from Scale. The process involves installing an agent on the source server and also on a newly built target server. This target server is nothing more than a matching OS version joined to the domain. Overall, this went well, but there were a few minor issues. On the physical systems we converted we had previously run Veritas’ System Recovery tool. Carbonite Move did not like the data files created by System Recovery. A quick call to Scale support identified the issue and once we excluded those, the migration completed without further issue.
3. New Builds. While migrating to Scale we also upgraded our Active Directory. We built new servers for this and promo’d them.
One of my favorite movie scenes is from the original Star Wars film. Shortly after rescuing the Princess from the prison on the Death Star, Luke, Han, Leia, and Chewy find themselves in a garbage chute full of discarded technology, among other things. When the trash-compactor walls start closing in, our heroes are battling to stay on top of, and ahead of, the junk tech. A fitting analogy of a career in technology.
With Scale Computing, perhaps we have finally found our R2D2 to turn off the compactor and let us escape the junk before it kills us. Scale seems to understand that we have other things to do.
We in tech tend to prefer best of breed products, but too often lose site of the goal. What good is the ‘best’ storage if it only solves one part of the problem? Instead of getting lost in benchmarks, maybe we should stop trying to impress ourselves with how ‘technical’ we are and instead focus on solving the business problem. Other solutions might have more checkboxes available, or offer a full 1.1% percent improvement under some condition you’ll never see, but does it actually do more? I’m sure there are solutions that do. But for us, this project is done and we couldn’t be more pleased.