HPE SimpliVity vs Nutanix vs Scale Computing: The Winner May Surprise You
I’ve been in IT for 18 years, so I’ve seen a bit of everything. But after all these years, IT can still surprise me. Take virtualization for example.
It’s amazing what you can do these days with virtual machines. You can simulate ancient hardware and run obsolete operating systems on new equipment. You don’t have to worry about maintaining decades-old computers, and you don’t need to find ways to bridge old interfaces and new infrastructure. It’s pretty powerful stuff, and I thought I understood how much you could do with virtualization. Then I learned a new word: hyperconvergence.
Hyperconvergence was in its infancy when I started at Marketing Innovations (MI), and I thought I’d learned everything I needed to know by the time we started looking for a hyperconverged infrastructure (HCI) solution, but I was wrong. Like I said, IT has a way of surprising you, and HCI certainly surprised me.
Assessing the Situation
MI is a data-driven business. The company focuses on performance tracking and provides employee incentives and rewards, but for such a forward-thinking organization, their data center was in terrible shape. That, among other things, is why they brought me in.
I remember the first time I stepped foot into it. It was in downtown Chicago, and felt more like a dingy, half-forgotten basement than an enterprise datacenter. It was absolutely dismal, which is probably why it was being shut down.
But it wasn’t just the location that was sketchy, the whole setup was a mess. We had seven cabinets worth of ten-year-old gear: servers, switches, cabling, an HP EVA, and a tangled web of cabling and patch panels. It would have been an absolute logistical nightmare to move.
Asking the Right Questions
We needed new hardware and we needed to migrate, so we thought why not kill two birds with one stone. My boss at the time, the VP of IT, brought in a specialist who looked at our needs and researched the environment.
The specialist recommended a hyperconvergence solution from Nutanix, but my boss, being the kind of guy every IT professional hopes to work for during his career, stepped in and suggested that we widen our search a little bit. He was king of due diligence and he gave me the breakdown of the solution the specialist suggested.
I dug into it and started wrapping my head around this appliance that runs the VMware. It was super cool but holy God was it expensive. I had to find an alternative, but I also had to bring myself up to speed on this whole hyperconvergence thing.
I dug through online resources, and like every good techie, went to the best knowledge bank there is: user groups. Spiceworks is my go-to community forum for all things IT, so I started a thread to ask the community for input. It’s actually still up if you’d like to check it out: “Looking at Hyperconvergence – Thoughts on Vendors.”
It was a crash course on hyperconvergence. I learned a lot from Spiceworks and from our potential vendors, all of whom provided me with more than enough information to become an expert. I was an absolute beginner, but by the time we had completed the selection process, I could have probably passed CompTIA’s HCI certification exam with flying colors.
Finding the Right Vendor
We narrowed it down to three potential vendors: HPE SimpliVity (in full transparency, this was prior to the HPE acquisition and our cluster ran on Cisco UCS equipment), Nutanix, and Scale Computing. Scale withdrew on their own because they couldn’t meet our IOPS requirements - an unfortunate development for them, but I will always respect their integrity for it. We were a small to medium business, but we were so technology focused and our entire offering was heavily SQL dependent, so there were a lot of database reads and writes happening all the time. Scale said, "Your market segment is our target market, but your technology needs are just way, way beyond that."
Nutanix offered a compelling solution, but HPE SimpliVity came up with a hyperconverged single appliance that incorporated a data virtualization platform and a storage accelerator card at a better price. I’ll be honest; HPE SimpliVity’s actual sales pitch was kind of dry. It was pretty boring, but it was comprehensive. There was so much detail about what the hardware could do, and I thought, “If it does what they say it does, this is pretty fantastic. This is what we need.” Sometimes a slick sales approach is all bluster - SimpliVity let their technology speak for itself and the message was loud and clear: "OmniStack data virtualization delivers the magic."
Moving a Legacy Data Center
I went down to our data center with HPE SimpliVity’s integration engineer Richard Gay. This guy was a wizard. We racked and stacked the HPE SimpliVity hyperconverged nodes next to our legacy equipment, and he guided us through the process of cloning our VMs with VMotion and HPE SimpliVity’s backup functionality. Then he helped us move our setup.
We pulled out our four new nodes and drove them over to our new data center, where we racked them, powered them up, and then federated them across the LAN with a couple of loaners from HPE SimpliVity.
Over the next few weeks we ran repetitive backups to our cluster at the new site. When it came time to do the move for real, we duplicated our VMs onto the loaner nodes with VMotion. We then did a HPE SimpliVity move from the loaners to our own equipment.
It was incredible. Using the loaners to slingshot our VMs from our old DC into our new one, we were able to move a two-terabyte SQL database over a 100 Mbit/second EPL linein less than ten minutes. It was one of the coolest things I’ve ever seen, and the advantages were immediately obvious.
Let’s start with the cost savings. Our previous data center took up seven cabinets at a colocation facility. This one takes up half a cabinet. That’s for everything: network equipment, servers, back-up appliance—the works—meaning colocation leasing costs are now next to nothing. And since we have fewer power circuits and fewer cabinets we have to pay monthly maintenance on, those costs are way down, too.
Then there’s speed. We’re dealing with newer, faster processors and more RAM. The hot blocks on our servers are running on a flash array instead of hard drives, so our IOPS have substantially improved.
This also translates into significant gains in development speed. One of the greatest functionalities offered by HPE SimpliVity is the immediate cloning and deployment of virtual machines. Say we have a massive surge in server traffic, or we have to onboard a new client and start a new program, all we have to do is click on a template. Or if we need a new database server or an IIS server, all our developer has to do is go in and change host names and IP addresses. It’s that simple.
With our ten-year old equipment, cloning a VM template for a Windows Server 2012 web or SQL server meant taking the time to transfer 50 or 60 gigs of data from hard drives. Now, we just add markers to a data deduplication table, so there’s not a lot of disk I/O.
If we need additional capacity to boot up a test machine, we can spin up extra development VMs in no time. Even when our network team is swamped with requests today, we don’t have to wait until tomorrow before they have the time to kick off a clone. Our network team can replicate a VM from a template, tweak IPs, host names, firewall rules and load balancer settings, and we’re good to go. We can start developing right now.
Less Is Not Always More
I’ve said a lot about the less-is-more approach of HPE SimpliVity, but if you are going to switch, make sure you overprovision on storage, on bandwidth, and on subnets.
MI struggled with low-space alerts. Even with compression and deduplication, we had onboarded some big clients and we were struggling with storage. The ability to instantly bring online new VMs and databases means that growth can be explosive. Make sure you have enough storage to keep up with new business opportunities. OmniStack's design makes it easy to scale out by adding nodes to a federation, but it's better to have that capacity to begin with so you have plenty of room to grow while you get accustomed to the myriad benefits HPE SimpliVity brings to your IT presence.
When it comes to bandwidth, you need at least a 10-gigabit Ethernet. You cannot run an HPE SimpliVity federation on one gigabit. And don’t forget to provision additional subnets for federation management. You can’t just add additional IP addresses to an existing subnet. Ask your integration manager what you need, and make sure you have the right network configuration.
But when you get down to the nitty-gritty, it’s all about cost and value. HPE SimpliVity was the right solution for us, and fortunately, it wasn’t the most expensive. It was the perfect fit for our data migration, but it was also the backup tool that we needed. Plus, it simplified and streamlined our disaster recovery strategy.
You don’t want to be the shop that’s running your entire operation off a 20-year-old floppy disk. But you also don’t want to be the company that blew its IT budget for the next ten years on a cutting-edge solution that hasn’t ironed out all the bugs.
You can balance your needs and your budget, but it takes time to find the right solution—and the right IT partner. Don’t just jump at the first proposal that seems to fit your needs. Use it as a starting point to ask more questions and to explore all the options that are available to you. What you find might just surprise you.