Swapping EMC SAN for HPE Hyperconvergence: Better Backups, Storage, and Flexibility
Where I work, IT is expected to do more with less. That likely sounds painfully familiar. Nearly every IT professional I know looks to trim costs, and no one has a limitless budget. But it’s especially true when your business is paper. My company produces the paper you touch every day: your standard printing paper, the glossy paper you use at your booth on the convention floor, and maybe even the paper label on your beer bottle.
Our products are essential to your daily life, yet the paper market is going through a major overhaul. Several years ago, our company was acquired. Ever since then, my IT team has had to approach every infrastructure decision by leading with cost consideration.
Of course, you have to balance spend with quality. Being cost-conscious is not the same as being cheap. We approach purchasing decisions with a “good, better, best” mindset. What’s the best-case scenario, what’s the middle of the road, and then what’s the worst-case from a cost standpoint? That helps us put the price in perspective and helps us define where we want to go.
It’s a challenge to work under these constraints while still purchasing the best products, especially with changing market conditions. But I like a challenge.
EMC SAN vs Hyperconvergence: Weighing the True Cost
For a long time, our setup was largely SAN, a Brocade switch, and a traditional host. We had difficulties with the functionality of those devices and the speed of applications. But, no surprise, the factor that pushed us to consider new solutions was the cost.
Our gear was largely EMC, and the renewal came up for those SANs. Suddenly, we were looking at a relatively big cost to refresh. That’s when a member of our team came back from a trade show and told us about a product he discovered: HPE SimpliVity.
This was before the HPE acquisition, back when it was just SimpliVity. There were several things we liked about their offering. The biggest was that it would be easy to add co-provisioned compute and storage nodes and scale in small, 2U rackmount HPE ProLiant server nodes. This was vital, because even though we’re under tight budgets, our company is actually expanding. We grow at about 10% every year. With HPE SimpliVity, if we needed more capacity, we could just plug in another node and away we go. HPE SimpliVity would make growing easy, which was a key feature for us.
We did look at other products, but HPE SimpliVity seemed to check everything we wanted: backup, storage, a single pane of glass view for VMware, WAN optimization, deduplication and compression, and it was all in one box. Then there was the cost. When we compared apples to apples, HPE SimpliVity cost less.
It seemed like HPE SimpliVity was the perfect fit for us, but we wanted to make sure. So we ran a PoC to see what it could actually handle.
Up to 65% of our environment is dev and QA. That’s how we work: We throw together all of our build and dev in QA, test it rigorously, and then move it into production. Because of that, those were the easiest systems for us to move over, and were the best representation of our environments. This made it the obvious choice when we were designing our hyperconverged PoC.
For our PoC, we moved all our Dev/QA systems onto four nodes and began testing. We unplugged power cables, threw different workloads on there to test resiliency and performance, and we saw a performance increase right away. It was night and day. It was clear that HPE SimpliVity could handle our needs, so it was time to get to work.
Our Surprise Runaway Success
Once we decided to pull the trigger, we bought four more nodes for our production environment. It was a two-plus-two federation scenario, with two in production and two in DR. It accelerated from there: We bought two more nodes here, another four there, as we kept rolling into production sites. Then, in 2016, we started a technology refresh for our distributed mill sites, which were still on the old setup. We’d proved the value on our corporate data centers, so as part of that refresh we purchased 31 nodes from HPE SimpliVity.
We were able to reduce our costs by double digits. As for performance, I could talk about figures, but it was the feedback from the user community that said the most. We were running a lot of different applications on HPE SimpliVity: Honeywell’s OptiVision, Cognos, Microsoft SQL, Oracle, Citrix. Our end users were blown away by how much faster applications were. Some of their old processes took half the time.
There was an aspect of the implementation and expansion that I couldn’t have foreseen, which was our success at marketing HPE SimpliVity within the company. You would hear people outside of IT talking about how great the “cubes” were. We called them “cubes” because we’re from the old school of SimpliVity early adopters, and at that time it was called OmniCube. Today you would probably call them HPE SimpliVity “nodes” or just HPE SimpliVity. But at the time, when we made our case to the business owners for purchasing this new technology, we used the short-form “cubes” when describing what it could do. That’s the term that stuck and became part of our company’s language.
The SAP project I talked about was such a success, the “cubes” came up again and again as we rolled through all the application owners’ projects. It just snowballed. Our end users kept talking to each other and their bosses about the “cubes.”
The widespread use of that term helped when it finally came time to go back to the ownership about expanding to more nodes for different use cases. They’d already heard about our success, and their first question was, “Well, are we going to need more cubes?” It was the perfect lead-in to that conversation.
Where We Go from Here
Our corporate data center was the first to make the change, and I’d say it’s now 98% HPE SimpliVity. All our mill locations are also on HPE SimpliVity now. All told, we’re at about 75 nodes. Each of our bigger mills has four-node installations, with two nodes in our data center backing up with DR replication. We store two weeks’ worth of data on those two nodes for each mill.
It’s incredibly fast to restore and back up VMs. Cross our fingers, we don’t have to do a ton of restores, but we did recently experience a ransomware attack. We were back up and running within an hour by leveraging HPE SimpliVity rapid restores. With our old environment, that would have taken a day or two.
Faster backups, less downtime, and easy expansion mean that HPE SimpliVity makes almost every part of our jobs easier. We’ve cut costs, but we aren’t compromising. We’ve found a cost-effective way for my team to do more with less. In IT, that’s pretty well as good as it gets.