Going Beyond the Cloud: Building Virtual Infrastructure at the Edge with HCI
For some time now, we’ve seen a heavy emphasis on cloud-based applications and systems. While this move to the cloud has opened up new ways of gathering, accessing, and sharing data, it has also begun to present some challenges, leading many professionals to look forward to what’s next. Edge computing is the latest frontier, and many businesses are exploring its possibilities with great success.
Edge computing allows data from the internet of things (IoT)to be analyzed at the edge of the network before being sent to the cloud or a data center. It presents many benefits when combined with the cloud, including reduced bandwidth requirements, enhanced processing abilities, and increased data storage. As companies move into this new phase of the virtual world, we are going to see a greater need for more virtualization.
At my company, we have seen this story play out in front of our eyes. We are one of the largest midstream energy companies in the United States. As director of IT server architecture, server delivery, and Citrix services, I’m responsible for all the server equipment we deploy remotely or within our data centers. Even our IT organization, while concentrated in Texas, is distributed across the state.
I’ve been here for over a decade, and a few years ago we geared up for a big change. We realized that if we didn’t find an appropriate edge-computing solution soon, we would be looking at a slower, costlier, and more vulnerable future.
Transforming Our Field Locations
The nature of the services we provide presented some unique challenges for edge computing. We have many applications that live in the field and have to be on-site to perform their function. With a centralized IT team, building infrastructure in the field and supporting it can be a real challenge.
It used to be easy with the standard rack-and-stack server and equipment, but as we started to move to a more virtual world in our field locations, our very traditional setup couldn’t support that—at least not without spending more money, using more resources, and buying more servers. Continuing on this path was going to be expensive in multiple ways.
The most pressing example of this was in our terminals, where operators load the trucks. Centralizing a terminal management system is very difficult. From a performance perspective, it’s important that the truck scale application is connected to the terminal operator’s console, but the availability of that terminal is also critical, and all those components have to work together.
We found that we have to house terminal management systems on-site in order to get the desired performance as well as align with the availability of that location.
When you have a system that requires such high availability in addition to all operational functions, you find yourself spending a lot of money and resources with the traditional methods: installing multiple servers, creating load balancing equipment, providing backup locations, and having the bandwidth to replicate data for centralized backups.
What we needed was a comprehensive solution to scale, deliver services, backup, and enhance our business continuity.
More Than Just Hyperconvergence
Five years ago, we first started looking at hyperconvergence as a solution to standardizing our equipment and tools. We’re an HPE shop, and loved the idea of standardizing our systems to simplify provisioning—and just overall management. HPE had come out with hyperconvergence tools in the past, but they lacked data management, backups, and replication capabilities.
That’s when I started believing that HPE SimpliVity could absolutely help us deliver edge virtualization services. But it was also so much more than that. It would protect data at the site as well as centralize it. We could collapse all our backup infrastructure and simplify our operations and finally improve our business continuity plan. Where historically we didn’t have the bandwidth to move that kind of data, with HPE SimpliVity it became possible.
A Record of Success During Rollout
With that demo in mind, we built a test for a single unit in one of our terminals. Our goal with this test was to ensure we could get those local and then replicated backups into our data center without impacting the network bandwidth. This specific terminal had a low-bandwidth circuit—around 5 Mbps—so any time we turned on backup services across the line, we could potentially impact the performance at that location or even at the office.
We set up the unit and built the application servers at the terminal, and then when we were ready, we had the network infrastructure folks monitor utilization while we pulled the application servers into our data center. We saw little to no impact to the network infrastructure during the test, and we also brought the VM back within a matter of minutes. Admittedly, we did not have a fully populated system at that early stage, but the idea that we could take that workload and replicate it into the data center without impacting the circuit told us HPE SimpliVity had us moving in the right direction.
That test took place two years ago. Shortly after that, we standardized the model of delivering terminal management systems on the HPE SimpliVity platform, including expanded and enhanced redundancy in those sites for the next terminal management rollout. Today, we are halfway through a four-year rollout project to standardize all our terminals with this infrastructure.
The simplicity of deploying the HPE SimpliVity units—spinning up and protecting the workloads, and knowing we had the availability plus the replication and backup—has made this such a straightforward implementation. HPE SimpliVity has not only delivered a virtual infrastructure at the edge for us, but it has simplified our operations in the field.
Historically, we did have a recovery plan, but we didn’t have the ability to fail over and continue operating in a seamless manner. That has changed with At the edge, we now have automated, policy-based backups of those workloads every 15 minutes, and every 30 minutes the VMs are efficiently backed up remotely to the data center. These are entire VM backups on top of the database backups taking place.
We have never needed those backups and I hope we never find ourselves in a situation where we do—but we have multiple copies of recovery, just in case.
Lots of Room to Grow (and Plenty of Room to Save)
When we began this journey, an initial concern was the potential for soaring costs that go along with an increase in devices. When we talk about savings with HPE SimpliVity, it is dependent on the extent to which the solution has been deployed, and we have to include not only hard savings, but soft cost savings in that equation.
Our terminals used to have multiple servers: the primary app servers, a backup server, and a tape library. The backup server was required to handle backing up the physical servers, and the tape library was to get data onto tapes.
It would be someone’s job to go to each terminal to verify the backup and take the tape off-site every single day. But now, we’ve removed that equipment because HPE SimpliVity handles the backups and the recovery into our data center. There are hard costs associated with removing those pieces of equipment at each location, but we’ve also completely streamlined those backups. We don’t need to have someone go to each location anymore. We can simply verify those backups from our central location through HPE SimpliVity.
Another area of soft savings comes from how we've grown this deployment. We were initially looking for something scalable, that wouldn’t require a large upfront investment, and that didn’t have to utilize a converged infrastructure in the data center. With HPE SimpliVity’s modular approach, it has meant that scaling is as easy as adding more nodes.
We started with a prototype of our VDI infrastructure to replace a c7000 enclosure with four HPE SimpliVity nodes. Now we’re on a second path of expansion from that model to deliver virtual desktops. That project would be difficult to justify if we had to come in with a larger system upfront, but being able to start with a few small nodes with HPE SimpliVity and then incrementally add more small nodes to spin up more virtual desktops as needed has been key for us.
From a day-to-day perspective, our customers are just happy. The solution works, it delivers what they want. By following guidelines from HPE regarding how to run, maintain, and operate infrastructure, we continue to simplify operations in the field, at the edge, and in the data center.