Chronograf, a Final Feature Review before GA
Chronograf is nearing the final stretch to GA! Tim Hall, VP of Products, will review the brand new capabilities of this highly anticipated visualization tool of the TICK Stack.
Watch the Webinar
Watch the webinar “Chronograf, a final feature review before GA” by clicking on the download button on the right. This will open the recording.
Here is an unedited transcript of the webinar “Chronograf, a final feature review before GA”. This is provided for those who prefer to read than watch the webinar. Please note that the transcript is raw. We apologize for any transcribing errors.
• Jack Zampolin: Developer Evangelist, InfluxData
• Tim Hall: VP of Products, InfluxData
Tim Hall 00:00.998 Okay. Good morning everybody. I hope you can hear me okay. This is Tim Hall. I’m head of products here at InfluxData. Joining me today is Jack Zampolin our developer evangelist, and we’re going to take you through all the great work that our engineering team and folks within the open source community have been working on related to the Chronograf. Before we get started, I’ll just walk you through the basic agenda. We’re going to cover for those of you not familiar with the TICK Stack and what we’ve been working on here at InfluxData, we’ll just give you a brief overview of what the components are and where they fit in the context of time series data, metrics and events in our modern platform that deals with all of that. I’m going to briefly discuss the goals with the Chronograf project as a whole and introduce you to our Chronograf mascot. Jack is then going to give a demonstration of where we are, and then we’ll talk through what’s left on the path to get us to GA, General Availability of this technology. And then we’ll wrap up with summary and Q&A. So without further ado, we’ll dive right in.
Tim Hall 01:04.592 So what’s the TICK Stack? So again, many of you are on the phone, I know, I see a couple of customer names out there, Peter and others, are really familiar with this. But for those who aren’t, I wanted to just provide a quick overview of what we have in terms of the technologies we work on here at InfluxData. We think about the problem of capturing metrics and events, both regular and irregular time-series data, anything with a timestamp, and we do that by collecting, analyzing it and monitoring and taking action on it. In terms of the way this is delivered, it comes in a series of parts, the TICK Stack. The “T” in the TICK Stack stands for Telegraf, and it’s backed by InfluxDB which is the time series database itself. So the collection portion is Telegraf, feeding that information into InfluxDB, allowing us to take action on it with Kapacitor. And what we’re introducing is the open source version of Chronograf, which is going to deliver a common user experience for administrative capabilities as well as an on-rep for visualization of the data itself. All four of those components are delivered in open source and available for download off our website through GitHub and of course, we love folks to participate with us in the community itself that working on the technology, is contributing your ideas, issues and even code if it’s capable of doing that.
Tim Hall 02:33.606 In addition, from a commercial perspective, we have the enterprise edition which delivers a high availability configuration, clustered scale up, and additional security features available for subscription. And that’s also available for customers in a hosted instance running in the InfluxDB Cloud. So hopefully that gives you a brief view of where the components fit, their relationship to each other and where we’re going. But today we’re going to focus on Chronograf. So in terms of the goals for Chronograf, what we were after was delivering a unified experience for the TICK Stack as a whole. And we started with, sort of the non-sexy things, which are the administrative capabilities of the Stack itself. So things like, I need a due database creation, inside of InfluxDB, I need to set up user accounts and what entitlements they have access to. Then as we get to the monitoring and action side, addressing things like configuring alerts and how do I quickly generate an alert configuration that can allow me to detect anomalies and decide where I want to send that information to, once detected. After learning those basics, we turned our attention to what other kinds of capabilities administrators were interested in. And one of the first things they get asked about is, “Hey, what data is within the database?” And so allowing folks to rapidly explore the measurements, fields, and tags that have been stored within the database and visualizing those results in a graphical way was the next thing that was put together.
Tim Hall 04:11.871 Now, of course, administrators of the platform also typically create custom dashboards based on available metrics and events. And we wanted to provide the ability for them to do that as well. In terms of deepening the user experience across the Stack, another thing that we’ve heard from the customers is, “Hey, I’m using Telegraf—” obviously one of the components for sending data in. “And if I’m doing that, can’t you pre-build the dashboard for me based on the metrics and the most commonly used metrics that are being provided for the various plug-ins and sources of those metrics? I mean, you guys are kind of experts at this. You’ve clearly created those dashboards before and, rather than providing some sort of a blank slate, where we could build visualizations on top, it would be nice if once a Telegraf provided a certain set of metrics that I could just click on a view of that and see the most interesting things available.” And of course, last but not least, we wanted to shift to deliver all of this in open source. So previous editions of Chronograf have been closed source. And we thought to drive this experience and offer it out to the entire open source community—was one of our goals.
Tim Hall 05:34.439 So with that, I want to dive right in and hand over to Jack for demonstration of where we are today. Where we currently shipped Chronograf Beta 9 last Friday. And we’ve got one more beta plan before the final GA push and we’ll highlight what remains to do from that point. But without further ado, let me hand Jack the controls of the presentation. And Jack, take it away.
Jack Zampolin 06:13.449 All right. Thank you very much, Tim. So let me start sharing here and we’ll get started. So one of the first things to know about Chronograf is login. So obviously you’re going to want some sort of access control if you’ll be exposing something to the internet, so we provided a number of different forms of authentication. Currently, we support GitHub, Heroku, Google, Oauth and we’ll be having a generic Oauth implementation that supports some other single sign-on providers. So let’s go ahead and jump in.
Jack Zampolin 06:52.883 So this demo that you see here is monitoring a Kubernetes cluster, so just a cluster of some compute resources on Google Computer. I have some various services that I maintain internally that are running on top of here, so we can kind of poke through. Here is your host view. If you’re running Telegraf on your hosts, we’ll give you some dashboards for free. So the stuff that you will build like memory, CPU, that kind of thing, we’ve already built that those graphs for you, and you get those out of the box if you’re running Telegraf. Also, canned dashboards for stuff like Kubernetes, so network ingress and egress, CPU and memory statistics, Docker as well. And then we’ve also got plugins for about 50% of those canned dashboards for about 50% of those Telegraf input plugins. So stuff like Memcached, MongoDB, PostgreSQL, RabbitMQ. Look at your Q depth to make sure that that part of your infrastructure, generally pretty critical, is healthy. So that’s our canned dashboard features here. Here’s the data explorer. So an easy way to drill down through your data. If you’ve used Grafana before, you know it’s kind of a little bit of pain in the neck. The query builder they have requires you to really know the query language quite well. It can be a little clunky. There’s a lot of clicking, and if you don’t save it, it will delete it for you. So you pick your database and then you click down through the measurements. Pretty easy to filter here.
Jack Zampolin 08:39.714 So if you’re looking for something specific, pretty easy to do there. Pick your field, apply your aggregations, pick a group by time, and very easy to do. So that’s the data explorer add multiple queries. Once you’ve figured out what you want to graph and want to make something more permanent, you can pop down into dashboards. This is our persistent dashboarding feature that would be similar to Grafana’s, a little bit easier to use. I’ve also found the editing pane a little bit more intuitive as well. You don’t have to pop up a whole different screen, you’re going to see the dashboards in the background. This is going to be the same type of query builder that you see over in the data explorer but will give you a few different visualization types, line, stacked line, step quad, single stat, and one-plus stat. So, for example, stuff like displaying any active Docker containers on my cluster, pretty easy to do with that single stat. And then you’re going to end up using those visualizations for a variety of other things. Stack lined, the first thing that comes to mind is, you’re trying to visualize all the different CPU’s on the machine, just stack them up so you can see the total.
Jack Zampolin 10:10.241 Then there’s the Kapacitor UI as well. So for generating alerts in Kapacitor, this is going to be very similar to the data explorer. So drilldown through your measurements, pick your field, apply aggregations and group by. The only thing that’s going to be different here, is that you can apply thresholding. So let’s say we want to alert when it gets to about 15%, just pick that spike up right there. We also have relatives, so if you want to alert when things get beyond—a couple standard deviations beyond the moving average, that’s what you would do there, and then dead-man. Obviously, if you’re familiar with Kapacitor, we do support many more functions, including some forecasting functions, in-TICK Script as well, but this is what we have exposed through the UI. And then, for your alert outputs, in this part of the UI, we support those right now. You kind of got your Swiss army knife HTTP, so if you wanted to generate some JSON down here, and get to some existing systems, very easy to do. And if you peek over here in the in the configuration section, you can configure the Kapacitor instances there. We also have an alert history here too, that you can filter by—so for filtering for an individual host, just seeing alerts for that, pretty easy to do. And then finally as Tim mentioned, there’s the administrative section as well. So this is where you would add retention policies to databases, create databases, user management, and then also, any long running queries are pretty easy to see in there. So creating the user. We’ll create “Tim”. His password might be Chronograf. Bam. And then setting permissions on him. Currently, we only have all, we’ll be adding many more features there soon. All right. With that, I’m going to go ahead and kick it back to Tim.
Tim Hall 12:49.180 Sorry about the little change over there. So the question is, “What’s left on our path to general availability?” As you saw from the demonstration, most of the dashboards that you can create are fairly static based on static queries. And one of the capabilities that we’d like to add, is this notion of template variables. So, in essence, it’s a mechanism which will allow the dashboards to be driven by a series of dynamic selection of information. In this case, contextual information that then, as you select through, will highlight and will essentially rerun the query and update the visual display. So for example, if you had a series of Kubernetes clusters like Jack was showing you, and they all have different instance IDs. Those instance IDs could be turned into template variables and you can use the same dashboard without leaving that particular screen to scroll through all of them at once. We’ll also have a final pass through of tuning some of the user experience and other minor known issues. As Jack was showing you, we have a whole range of permissions, things that we would like to land as part of the final push here so that you can sort of lock down certain aspects of user experience for administrators and potentially other users. So, for example, if you want some to be able to access the dashboards that are available but you don’t want them to be able to create databases or users, that’s a logical separation of concerns and task that we want to be able to deliver through Chronograf. And if you’re keeping up with the mascots, this is our Chrono giraffe in the corner. He’s the mascot for the team. He’s been helping us along the path here, and we hope that you guys have seen some of the stickers and other things that have been sent out supporting the Chronograf and the sort of branding of that as we go forward.
Tim Hall 14:49.237 So before we get to some Q&A, some final summary here. So the first thing is we’re trying to deliver that integrated experience for the TICK Stack across those basic administrative cases, including data exploration and alert definition. The dashboarding is coming along really nicely, including those out of the box dashboards when you use Telegraf as your means for ingesting data into InfluxDB. For open source users, we had previously announced the deprecation of the admin interface, and that will be removed as part of the 1.3 open source edition when it comes out. For our Enterprise edition customers, you can use Chronograf already as is, because it is Enterprise edition aware, meaning when you go and set up the connection to the database it will know that you are using the Enterprise edition, and it will ask for some additional metadata as part of the connection string. And then you can go ahead right away, and start using Chronograf directly. Once Chronograf does go generally available, we’ll announce the deprecation and subsequent retirement of the Enterprise web console, which is what many people are currently using with the enterprise edition. We just think this experience is much simpler to set up, easier to use, and frankly is our long-term direction related to the user experience for the entire TICK Stack. We’re excited about offering this out to all of our Enterprise customers, and our cloud customers as well. So if any of you are also using InfluxDB Cloud or thinking about using InfluxDB Cloud, our plan is to include Chronograf as part of the overall experience. Coming soon later this summer is the current plan, and we’re looking forward to providing that to all of our InfluxDB Cloud customers.
Tim Hall 16:38.558 So with that, we’re going to turn it over to some Q&A. If you’ve got questions, please place them in the Q&A section of the webcast, and we will go through and have a look. Okay, so first question is, “There’s no Mac installation, brew installation, for Chronograf, and I want to be able to use it on a Mac.” I think one of the limitations in brew was related to it being a beta, and we used the tilde character to restrict the beta. But once we hit GA, we’ll create the brew install package so that you can land it on your Mac. So apologies for that, we’ll make sure that gets taken care of. Next question is, “Will Chronograf UI and data explorer support TICK Script?” So you saw one example of the alerting TICK Scripts that we have out there now, and the idea as we continue down the roadmap here is to provide more expressiveness in terms of the types of TICK Script that you can write for Chronograf. And some of the things that we’re contemplating for the future, one of the best practices that we documented – and you can find examples of this in the documentation – are how to use Kapacitor for down-sampling. So essentially, if you want to offload the down-sampling activity, you create TICK Scripts for that and again, there’s a number of examples that you can draw from. That’s another sort of next step or low-hanging fruit that we’re looking at to providing for Kapacitor, and that’s sort of stopping short of a full-blown sort of TICK Script editor, if you will. I mean, one of the goals is—of course, any time you introduce a new technology, if there’s a barrier to understanding or a barrier to adoption because you have to learn a bunch of new things, that’s—that definitely can be a challenge for users. And so the idea behind Chronograf, at least initially, is to sort of lower that barrier of entry of creating those alert scripts and doing it through that sort of visual interface to allow people to access the capabilities of Kapacitor a little more quickly than they had.
Tim Hall 19:08.075 So the idea is to continue to broaden that capability over time. Another question is regarding the deprecation of the admin interface, “What is the takeaway for folks who are considering using the open source version if we remove the simple command line or the simple browser expectation? What’s my expectations about what users should do without it?” Let’s use Chronograf. It’s pretty simple. The idea is that we’re separating the parts into their core constituencies. So the database is the engine, if you will, and the CLI will continue to be there and available if you would like to use it. But as we move from the CLI into the world of the browser and how to access all these different components, all the things that you can do with the administrative interface is absolutely available. And so, yeah, we actually just think it’s a big step forward and a tremendously better experience in sort of the very simple UI that we’ve had there before.
Tim Hall 20:26.956 So we’re getting a question about the templating. So the first pass at templating should be available, not this Friday but next Friday in the final beta addition, so that’ll be beta 10, and you can look for the updated links to download. The team has been on a two-week sprint, every two weeks providing new capabilities and landing the updates, typically on late Friday afternoon. So you can look for that as something to play with over the weekend or picking up first thing Monday morning if you’re back in the office. And sort of that’s been the normal and regular cadence of the team. And we’ll expect to continue to land defect fixes and other updates post-GA as we prepare to launch Chronograf into our cloud environment. The last question is related to Chronograf showing Kapacitor tasks that maybe have been previously created in the CLI. I actually don’t know the answer to that question [laughter], to be perfectly honest. What we’ll do, Trevor, is we can follow up with you offline and maybe post the answer to that back. I think it’s a great question. Should definitely know if folks have gone deeper with Kapacitor, then where’s the intersection of these things to make sure that you can continue to use both mechanisms? So obviously there will be more sophisticated scripts that you may have already landed in Kapacitor and then is there any impact in terms of integration with Chronograf either now or as we continue to sort of build out other wizards or maybe even getting to the full-blown editing capability within Chronograf itself? I think for sure if we had a full-blown editor, the sort of import-export of all those scripts both existing should totally be possible. But since this one has a really restricted window in terms of what we offer from an alert definition perspective, there may be some constraints, but we’ll get that answer packaged up and sent out to you.
Tim Hall 22:42.815 Okay. Unless there are any other questions, I just want to thank everybody for joining today. We’re super excited about your participation. You can direct technical questions at any time to community.influxdata.com. For those of you that may not be aware, we’ve seen lots of different Q&A sort of sessions break out in terms of Google Groups or Stack overflow, and one of the things that we decided to do was curate our own community. And so if you haven’t visited community.influxdata.com, we’re getting lots of good interaction from folks—from users, both from the open source community and our—obviously our commercial customers there, and so we’d love you—for you to continue to participate with us there.
Tim Hall 23:37.522 Okay. Thanks again, everybody, and look for the webcast to be posted. Send it out to your friends. Tweet it and continue to look forward to the final beta push here to GA for Chronograf.