Building Infrastructure for the Future and Improving App Performance by 2X with HPE SimpliVity
My problem was a common one with a company whose server and storage infrastructure was reaching the end of its IT lifecycle. It was time to replace our servers and storage resources, and I was suddenly faced with a very real issue of trying to figure out exactly how that would happen.
Our existing IT infrastructure of blades and storage wasn't designed for fast disaster recovery. Offsite server and file recovery had a nine-hour data lag. Only our most critical servers were on a fast failover solution. The aging NetApp drives were beginning to fail more often. The blades were aging and our newer applications were running a bit slow.
As the systems were getting older, hardware failures were beginning to occur more often, especially the storage drives. We had ample failover capacity but the maintenance overhead was rising.
If we ever had a full loss of our server room, critical servers would be up fast, but the rest of the infrastructure would not recover quickly. This meant working with our off-site vendor to spin up VMs on their infrastructure using our remote storage with data that was nine hours old, and this required almost a week of rebuild time, plus nine hours of lost data. Not a pretty picture.
This simply wouldn’t work for us at ICMA. We're a membership organization for small- to medium-sized local government managers and the staff associated with them. We strive for local government excellence. But to get there, we needed IT excellence.
Getting Started: Our Dilemma in Detail
There wasn’t a shortage of choices to work with. In fact, I was probably facing the opposite challenge of too many choices, but each choice was a small piece of the overall solution. When you looked at the number of vendors involved (storage, blades, backup, cloud, failover), it showed the risk of a lot of finger pointing when problems arise. I was at the point where I wouldn’t know which was a correct fit until we actually took the deep dive with a commitment. And that could be a very big mistake if it turned out to be the wrong decision.
But we were also running out of time, so we needed to select something soon. I was carrying 13 TB of data embedded in a mesh of VM and file storage. To make sure we didn’t have a cataclysmic loss on a bad day, I had a duplicate NetApp as the backup in place to ensure continuity.
The NetApp resources we relied on for so long had served us well, but they were just beyond their full capability. The sync alone took something like nine hours and it failed repeatedly during runs. This got to be such a concern that we were working with two completely different vendors for something as simple as reliable storage. From a disaster recovery perspective, it would have taken us a week to get completely back online given the fragmentation of our resources and who had responsibility for them. It was not a good picture.
Hanging on the Cliff's Edge with Choices
The main choices for us circled like birds around either: 1) upgrading our NetApp and existing blade resources, or 2) moving entirely to a new choice with hyperconvergence. A third choice might have been cloud based. However, with some of the recent testing that we had with AWS and a cloud server approach the year before, I was skittish of the third option as a primary choice. With AWS, we had two outages within three weeks, so that didn’t seem like a reliable solution.
I had a specific shopping list of what was needed when I started our migration discussions. The servers and storage needed to be adequate for future growth for the next five years, instead of just our immediate capacity needs. I had to have all the requisite backup and recovery options available. And I also required the necessary resources to ensure business continuity if a server went down.
Previously, we had dealt with a hodge-podge of vendors to cobble this list together with our current system, and I wanted to get away from having so many hands in the pot going forward. After a lot of research, time spent studying options, and a lot of elbow grease, it was HPE SimpliVity that rose to the top of the hill with meeting all our needs in one service.
The Water Was Shallow Once We Got In
The shocking big step into a new IT paradigm for us started first with storage. With HPE SimpliVity’s package, we only needed to buy 9.9 TB of actual, real storage on physical resources. And that was after all the requisite overhead and behind the scenes set aside space. Right away, it stopped me in my tracks because of our previous storage demands, but HPE SimpliVity made it clear the virtual side of the picture was more than robust with a 20 TB logical capacity as the immediate offset. This result was due to the inline deduplication and compression enabled by HPE SimpliVity.
The implementation ran like a charm, and using the free version of VEEAM, we easily migrated our VMs from the old infrastructure to the new one. Now, on board with HPE SimpliVity, we are looking at 32.9 TB storage total. That is made up of 20% for virtual machine data, 80% for local VM backups, and 0 relocated to off-site resources as we had to rely on in the past. No more juggling with different vendors for one infrastructure.
HPE SimpliVity was a partner all the way through the conversion. They helped us with all the pre-planning, gap analysis, picking the right resources, and scheduling out all the phases. Instead of what would have taken months to put into motion, HPE SimpliVity got the real-time expectation down to one and half weeks of actual work and hands-on transition.
Once we were up and running, the results for our end-users were fantastic: Every application ran twice as fast. This meant their work would no longer be impeded by slow-running applications.
Even better, our new, all-flash system will easily carry our company for the next five years without a hiccup. We have plenty of storage and room to grow without any curtailing. We have 10-minute VM and data backups compared with the 9 hours it previously took. And we gained improved resiliency and agility with momentary continuity cutover in place versus what was the very real possibility of a complete week-long shutdown before.
HPE SimpliVity also provides us with more options including with choice in hypervisors. They support both VMware and Microsoft Hyper-V. As no single technology today is the one to choose in the future, as a user of VMware, I currently prefer it to Hyper-V. But, if VMware goes the way of Novell Netware, it's nice to have options.
Coming Out on the Other Side
Today, thanks to HPE SimpliVity, we are cohesively running 30 different servers. There is pure database management, accounting operations, sales functions, operations, and more. In fact, the core utilities in our portfolio are the big energy hogs, eating the most demand and requiring the fastest hosts available in our stable. And they are running smoothly with the new setup.
But here’s where HPE SimpliVity went above and beyond. In the middle of the install, it could have turned into a perfect storm for us in terms of going down the drain. Thanks to Murphy’s Law, my prime network engineer was taken out of commission with an injury. Instead of this derailing everything, HPE SimpliVity provided the key guidance and help, and we continued in a very timely migration to our new system without delays.
There is no question that moving onto a one-vendor platform was the smarter move. The choice costs us less, our system is far more cohesive and integrated, and our network is constructed holistically instead of patched together with lots of compatibility band-aids between modules.
We don't suffer from unexpected component conflicts. Instead, we enjoy greater synergies and performance leaps due to a finely running platform tailored to what we need now and for the future as we grow.
There are lots of experts out there who can consult how a company should migrate to a new platform. But they're not in the trenches like HPE SimpliVity was with our move to hyperconvergence. And that makes a world of difference. Put aside the talk and promises, when it comes down to real performance, a one vendor approach works exceptionally well.
From a technology standpoint, ICMA is in a vastly different place than we were a short time ago. And this isn’t just about today—it’s about the future. We’re prepared for whatever comes next. At ICMA, we’re helping create excellence for others, but we started with ourselves.