Accelerite, through a variety of products, provides a platform that helps enterprises transform themselves digitally. Think transformation from brick and mortar to digital, if that helps put some non-Silicon Valley context around the digital transformation unicorn that just pranced by.
Accelerite products include the following.
- Concert provides an application development platform, an insight creation platform, and then an ecosystem & monetization platform for IoT.
- ShareInsights is a self-service tool for data analysts—native analytics for at-rest Hadoop data as well as real-time data from Spark or Kafka.
- Sentient secures the data across all the other Accelerite products.
Meet Accelerite’s Rovius
Most interesting to readers of this site is likely to be Rovius. Rovius is a cloud platform for public and private clouds that came from an acquisition of CloudPlatform from Citrix. Accelerite has been adding to the product via contributions to the open source CloudStack project—bare-metal orchestration, service orchestration layer, Kubernetes (K8s) cluster orchestration, and hot upgrade features, for instance.
Rovius is addressing the reasons enterprises haven’t moved from traditional VMware to some sort of cloud. Technology skillsets are often a roadblock, as the people required to stand up a cloud are hard to come by. Accelerite claims that Rovius creates a ready-to-run cloud in hours, with no unusual technology skills required. That stands in stark contrast to the experience some enterprises have had with other private cloud solutions, notably OpenStack.
Rovius is the consumption layer on top of CloudStack that makes CloudStack easy to use. If you’re wondering why CloudStack underpins Rovius as opposed to OpenStack, Accelerite believes that CloudStack is a more stable platform, despite OpenStack being a larger community. Accelerite cares about customers and solving their problems, and went for stability. Another consideration is that consumers of Rovius are using…well…Rovius. Therefore, the fact that Rovius is CloudStack underneath isn’t all that interesting.
Rovius is scalable, with several customers running 4,000+ hosts, and launching/destroying as many as 1,600 VMs per day. While listening to Accelerite pitch at Tech Field Day’s Cloud Field Day 2, Justin Warren pointed out that, “The steady state is constant change.” Yes, exactly, which is where orchestrators like Rovius come in as IT scales well beyond the human ability to keep up.
Rovius is more than an orchestrator. Rovius is another entrant into the orchestrator of orchestrators space, as Rovius handles more than just CloudStack-based private clouds. Public cloud is handled as well, being just another zone in the Rovius system. The same scripts and same tools are used for a seamless experience, whether workloads are being managed in the public or private cloud. That makes Rovius an enterprise hybrid cloud management tool.
In fact, the ease of consuming public vs. private cloud resources was a highlight from the demos I saw. It’s as simple as choosing the proper zone to stand up your workloads in. There is no distinction between public cloud or private cloud zones. The provisioning experience is the same. Choose the public zone from the Rovius console. Want to double check that a new instance actually stood up in in public? Head on up to, say, your AWS EC2 console and check it out.
Even with all of that, Rovius offers more.
- Containers are supported.
- Kubernetes systems can be created on demand.
- Hadoop databases can be created on demand.
- CI/CD tooling is also available.
- Serverless capabilities are not offered yet, but are interesting to Accelerite.
In that sense, Rovius isn’t strictly an IaaS play. It’s also PaaS, at least for certain services it supports, meaning they are taking a page out of AWS’s, Azure’s, and GCP’s playbooks.
Rovius is a software stack only, meaning that it needs to be able to work with the hardware you bring to the hybrid cloud party. Let’s review several key aspects.
- You bring your own IPMI-capable bare-metal hardware or public cloud infrastructure. There is no difficulty for Rovius whether you bring bare metal, public cloud, or both. Rovius was designed from the ground up to be a management platform for hybrid cloud.
- The Rovius Controller is bare-metal. The controller tracks and manages everything in the hybrid cloud.
- The Rovius Configurator allows you to build blueprints via a YAML descriptor file and apply that hierarchy to the cloud (hosts, clusters, pods, etc.).
- The end-user console is actually “CloudPlatform.” It’s not new as such, but is part of the open source CloudStack. The Controller and Configurator are value-added Rovius magic.
- For disaster recovery, Rovius doesn’t integrate with anything specifically, but that doesn’t stop you from still using Zerto or VMware Site Recovery Manager. Rovius plans to integrate with DR solutions eventually.
- Rovius allows for a choice of hypervisors. KVM will be deployed by default, but if you’d rather use VMware, you can do that.
- Storage support is a little complicated.
- For standalone storage tiers, Rovius has many plugins that access most of the big names.
- VMware vSAN is supported.
- For other HCI-based storage, support is promised by the end of 2017. The trick is writing an abstraction layer that presents HCI storage as a consumable resource for compute just like a standalone storage platform would do. That’s going to take Accelerite a bit of time to get done, but they are working on it.
Rovius emphasizes that compared to MSP solutions, their platform is all self-contained, except for the management plane. The management plane is in the cloud, but the control plane is local. With MSPs, the management and control planes are both in the cloud, outside of the firewall.
Accelerite tried to make a big deal about the control plane being local as opposed to cloud-based, but to be frank, I saw this as merely interesting as opposed to important. In a world where the Internet is a reliable transport and it’s even possible to plumb yourself directly into the public cloud, it’s hard to get excited about the “control plane inside the firewall perimeter” feature. Accelerite confessed that they had to build the management- and control-plane separation anyway to support SPs who wanted to resell private cloud, so they just decided to sell it as a feature.
Viewing demos of Rovius showed a lovely, if typical, GUI, with workflow-driven pages through which you go clicky-clicky. However, Accelerite points out that the entire platform is API-driven, using CloudStack APIs as well as their own Configurator and Controller APIs. The GUI they offer is just a consumer of those APIs. End users can consume Rovius via those same APIs if they desire.
The GUI is quite adequate for low-volume operations, though. For example, provisioning of network links was a straightforward series of dropdown selectors to create virtual routers and instantiate an IPSEC tunnel between each cloud environment. You don’t have to worry about the specifics of standing up a tunnel whichever two cloud environments you are connecting. Rovius handles the details for you through the magic of abstraction.
Monitoring everything in real-time is touted as a feature, but note that Rovius is all about infrastructure management. Rovius is not for application management. That said, Rovius is conscious of enterprise concerns that are a bit higher in the stack. For instance, Rovius will integrate with LDAP.
Rovius monitoring offers history where log file entries are correlated via timestamps to other monitoring events such as CPU utilization. This correlation can be easily bundled into an exported text or CSV file. Future correlation will include not only timestamps, but also event matching by process ID throughout the CloudStack system and subsystems. Thankfully, marketing did not “machine learning” wash any monitoring slides, although I can’t help but wonder if there might be an interesting ML use case. After all, Rovius is in the middle of an awful lot of interesting infrastructure data.
The View From The Hot Aisle
As with Nirmata, the idea of an abstraction layer for cloud comes up with Accelerite’s Rovius. If you’re a developer writing to AWS APIs, your code is married to Amazon. Moving off of AWS becomes hard. Writing to a third party tool like Rovius means that you have public cloud portability if you’re writing to Rovius instead of to AWS. Useful.
Another major point is the worthiness of private clouds in general. Public cloud, especially at scale, is expensive unless carefully tuned to minimize waste and leverage serverless. Plus, some still have security and compliance obligations that compel them to house their data privately. Interestingly, even large cloud operators are seeing this as a significant market opportunity. For example, Microsoft Azure Stack is an answer to an on-premises flavor of the Azure public cloud. Microsoft doesn’t have to drop public cloud pricing in this case. Now, they eat both your public cake and private cake while maintaining margin, and that’s some pretty delicious cake.
Yet another thought is that artisanal management, where IT stack components are lovingly configured by
rocket scientists infrastructure engineers, is over. Managing infrastructure is officially the job of robots. Rovius is just another answer to the infrastructure stack management problem in an increasingly large group of choices. Listen to the Datanauts podcast episode 94 for a lengthy discussion where automated provisioning was a theme.
Bringing the focus back to Rovius, there are signs of immaturity. For example, high availability is a roadmap feature, when you might assume HA exists in a product aimed at enterprises. Therefore, as with any evaluation of technology, assume nothing. Rovius does an impressive number of things with a low ramp-up time and high value. On the other hand, you need to ask a lot of “what if” questions that map to your business processes to understand how Rovius fits in.
- What happens when your primary data center fails?
- Are the public cloud platforms you want to use supported?
- How long does it take for Accelerite to update Rovius when the public APIs they consume change?
And so on. I loved the ease of use of Rovius. I also loved the way Rovius made it easy to consume a variety of clouds. But then I was left pondering about those complex failure scenarios, trying to decide where Rovius’ role ended, and application design began. I’m increasingly less of a fan of solving terrible application design with infrastructure duct tape, but also realize that some infrastructure teams aren’t given a choice. Therefore, if an app is poorly designed and must survive despite an inability to tolerate infrastructure failures, does Rovius fit?
I don’t have a good answer to that question, but I know it’s on Accelerite’s mind, per their comment about integrating with DR solutions.
Yet another question is how many organizations reflect Justin’s comment, “The steady state is constant change.” Certainly, there are enterprises moving into the container world and experiencing ephemeral workloads. Rovius should be pretty interesting to those folks, especially if they are hybrid or multi-cloud. But how many companies are like that as opposed to companies who are rooted in VMware not just because they can’t move to cloud, but because they don’t feel a technical need to do so?
And that is the root of the trouble for those companies with leading edge technology trying to get a foothold in the broad swath of slow adopters comprising the enterprise market. Even the best product in the world can peak too early.