So, what’s important?
We’ve built a lot of features into Bright Cluster Manager over the years, and we wanted to know which ones our customers find to be most important. Is it saving time? Saving money? Something else? As it turns out, 89% of respondents reported that the most important feature of Bright software is that it makes clusters easy to manage. So it seems we’re living up to our company slogan which says, “Cluster management made easy.”
Next in order of importance, at 74%, is the fact that Bright saves cluster administrators time. After that we have, “gives a centralized view of the entire cluster” at 60%, “makes it easy to carry out upgrades” at 45%, and rounding out the top five we have, “carries out automated health checking” at 41%.
It’s interesting to see that, “saves money,” is ranked as important for only 18% of respondents. We will have to do a little digging into that one to find out why it didn’t rank higher, since everybody likes to save money in their IT budget. Less surprising is the fact that the ability to run different workloads on the same cluster ranked only 15%. We believe this is a growing trend in clustered infrastructure usage and will become increasingly important to more of our customers in the coming year.
Half of the users surveyed said they wanted to run containerized workloads on their clusters. So, it seems that the container trend continues at a brisk pace. Other cluster activity that our customers plan in the near term include, machine learning (31%), clusters on demand (28%), extending into public clouds (22%), and OpenStack private clouds (21%).
When it comes to container technologies, Docker is the clear winner with 36% using now, and another 23% considering using it in the future. Next up is Singularity with 23% using, and another 20% considering. This makes sense since many of our customers are HPC users, and Singularity seems particularly well-suited to using HPC environments. Somewhat surprising was the 6% usage reported for Mesos. This is quite a bit lower than we had expected, but container technology is still nascent so we will keep a close eye on things and see how the usage mix changes over time.
Most of the respondents that are using containers today, are running them on physical nodes (42%). It’s interesting to note that the majority (48%) aren’t using containers at all — at least not yet. We expect this to change in the coming year, since containerized workloads can bring so many benefits.
So what’s the takeaway?
There’s more of the survey to talk about, but I’ll save that for another post. For now, the biggest takeaway is that ease of use remains the most important feature Bright brings to system admins, and that the use of containerized workloads has taken root and started to grow.