Thoughts on KubeCon 2019
Jestin Woods recently attended KubeCon 2019, and he shares his thoughts about his experience
I’ve been attending KubeCon since the first one in 2015. Overall, I think this is my fourth time attending the conference. It has been highly relevant to the work I have done over the past several years, and I enjoy attending.
KubeCon 2019 was my first at working in a booth, and in fact, it was my first time working a booth at any conference in my career. I learned a lot about our product and what people’s initial impressions are, and I gained some insights into what we’re doing right and some things we can do better.
As I talked to people visiting our booth, I had some great conversations, and I got to answer a lot of questions about Humio. Here is a brief summary.
Humio does not use Elasticsearch
So many people I talked to nearly dismissed Humio because they assumed it’s backed by Elasticsearch. I can’t imagine how many people don’t consider a product like Humio because they may not realize there are modern, purpose-built products like Humio that approach data aggregation, search, and analysis in an entirely new way.
Elasticsearch itself has become sort-of a standard amongst organisations because it is “free.” We all know it starts out this way but can quickly become a management nightmare — for example, “The Elasticsearch cluster is YELLOW again, I guess someone left their logging set to debug mode….” This stigma may also extend to Kibana and other tools based on Elasticsearch.
Humio is not open source
Some people will simply dismiss Humio because it isn’t open source. Even though Humio isn't open-source software, users can easily customize it to fit their use case, and our support team loves working with customers to make Humio work for them. We find that many customers come back later when the open-source alternative either doesn’t meet their needs, or becomes difficult to manage at scale.
Humio and the three pillars of observability
The boundaries between the three pillars of observability become increasingly diminished in the context of using Humio. Because Humio doesn’t enforce structured data means Humio is a powerful data exploration tool, regardless of whether that data is considered “logs,” “metrics,” or “traces.” These categories no longer make sense from the observability perspective. We can mix, aggregate, and compare data from any source in any format.
When it is asked, “How does Humio compare to <insert logging tool>?”; “How does Humio compare to <insert metrics tool>?”; or “How does Humio compare to <insert tracing tool>?”, the answers will be different. But because Humio can aggregate, search, visualize, alert, and run scripts in real time, Humio can deliver the promise of any pillar of observability to organizations of any size.
Humio has features that are desperately sought-after, like Views
A question that has been asked by many — including myself — and at just about every company I’ve worked for is once we aggregate logs into a central tool, how do we ensure certain teams have access to their logs while other teams do not?
Humio has a solution for this with Views, which turns out to be a very simple concept: Views restrict access to a subset of data based on the user. I’m willing to bet nearly every large corporation has had a conversation about this at some point, yet they have not done anything about it because traditional tools make it nearly impossible.
Humio is fast even without the indexes
This is a more common question than I thought I would get. Not indexing on ingest makes the system behave differently and better. We have whitepapers explaining how we can ingest 100 TB/day with no problem. These may not adequately describe the speed at which data in the system can be observed, why it’s so powerful, and how we achieve that speed.
The logging space is crowded, but people are not satisfied
There are way too many “logging” tools and services, but this is because people are not happy with the previous solutions. I’ve heard complaints about just about every product out there, but perhaps more interesting is when people say that logging is not a solved problem.
One person asked what my personal thoughts are on a recent entry into the market that takes a unique approach to observability, though in very different ways than Humio. In this example, to achieve speed they require sampling at any significant scale. This individual knew this approach wouldn’t solve his organization’s problems, and was intrigued to know that Humio would let them log everything in real time.
Humio has flexible integrations
Humio integrates with other systems. To send data to Humio, there are a number of open-source tools. For example, it’s simple to point fluentd to a different endpoint to get existing data into Humio. And to get data out of Humio, we have webhooks.
Engineers don’t like to be emailed
The whole “gather leads and email them” thing at conference booths is quite undesirable for engineers. We know that if they want to use a product, they will do research, install it themselves, and then reach out if they have questions.
That being said, I understand we don’t want them to forget about Humio. We’ll probably continue to do this until there is a better alternative. In the meantime, we should remind them to join our public Slack to follow up with us. I find engineers are more inclined to join a casual place like Slack for a technical conversation, rather than a structured video call with a slide deck (that’s not really how we do things, but that is generally the impression people have when they share their email).
Overall, I really enjoyed the experience of meeting people and having conversations with them at the Humio booth. This was due to the people in attendance, and the highly similar work that they share with what we do on the Humio Infrastructure team. It is easy to talk about our product when many of the people in this space share similar experiences and knowledge.
This experience has also been encouraging, seeing how there is so much room for us to expand on the foundation of the solid product we have. I know that as a company, we’re all committed to open communication, sharing examples, creating helpful blog posts, and making our product easier to use with apps, documentation, or supporting tooling.