WEBVTT 00:00.000 --> 00:13.440 Hi, everyone. I'm Sam Jewel. I work at Grafana and this is about Aggrafana. We're trying to combine 00:13.440 --> 00:17.760 observability data with business data, trying to blend them together. We want to answer 00:17.760 --> 00:23.000 questions like how much lost revenue was there due to downtime, what's the latency impact 00:23.000 --> 00:31.080 on black Friday sales, how customers impacted by that recent incident. You might have 00:31.080 --> 00:36.520 run a service, maybe, and seen some error logs, and you can imagine, you can see that customers 00:36.520 --> 00:41.880 are affected. You can see some ideas for customers, but you don't know, are those customers 00:41.880 --> 00:47.480 paying you millions or are they just on the free tier and completely inactive? You don't 00:47.480 --> 00:53.240 know how important this is, whether to focus energy on this, and you need to blend in that 00:53.240 --> 01:01.000 revenue data to find out. That should be easy, right? What you want to do, in theory, you just 01:01.000 --> 01:06.520 want to join on a customer's table, and you want to select their names, their revenue, and 01:06.520 --> 01:13.160 so surely their solutions to do this. You'll look around, you'll find their solutions, you know, 01:13.160 --> 01:18.200 some logs back in, can speak SQL, some metrics back in, can speak SQL, and their solutions 01:18.200 --> 01:24.920 where you can put everything into a single SQL database, it all speaks SQL, and so you should be 01:24.920 --> 01:30.200 able to write a single query and join everything together and get your answer, right? And so 01:30.200 --> 01:37.320 your analytics, business analysts, and your engineers, your SREs, should be able to use this 01:37.320 --> 01:44.840 and you're away, fine, great. In reality, it's not that simple. For a few reasons, so your 01:44.840 --> 01:52.200 analysts probably are not writing SQL. They might be using Metabase, Superset, Tableau, 01:52.200 --> 01:58.440 Looker, they're creating queries using this abstraction layer, allows them to query with 01:58.440 --> 02:05.400 dimensions and measures, and the tool generates the SQL. And that is a problem for us in two ways. 02:05.480 --> 02:13.640 The first is that the things that have the most value, the things they're producing that 02:13.640 --> 02:18.520 have the most value are their reports, their dashboards, their queries, and those no longer live 02:18.520 --> 02:24.680 particularly close to the SQL database, they live within that tool, within Looker, within Metabase, 02:24.680 --> 02:30.760 and that's a problem because now your engineers, your analytics, your SRE, can't access 02:30.760 --> 02:34.840 those reports and those dashboards, and they can't access the queries that hold the meaning 02:34.840 --> 02:41.720 and the value around like how a monthly active user or annual recurring revenue has even been 02:41.720 --> 02:49.400 defined. So the other problem is, like tools like Metabase and Looker and Tableau don't even 02:49.400 --> 02:54.920 play nicely with log data and metrics data, it's just such a different shape that you can't 02:54.920 --> 03:00.760 make it kind of speak these dimensions and measures. So these guys are upset too because they 03:00.760 --> 03:05.400 can't access the log data in the metric data that they're trying to join in. And exactly 03:05.400 --> 03:10.600 the same thing is true from the other side. You'll see that SREs are not writing SQL either. 03:10.600 --> 03:17.000 They are using a UI UX in Data Dog in Grafana and Kibana, and those tools in turn are not 03:17.000 --> 03:23.640 speaking SQL either, often. Elastic can speak SQL, but when Kibana speaks to Elastic, 03:23.640 --> 03:30.040 for log data, it's using this query DSL because log data and metric data is a different shape, 03:30.120 --> 03:35.320 and there's a different access pattern, it's a different domain. And their queries and their 03:35.320 --> 03:40.920 reports and their dashboards live within Data Dog, within Kibana, within Grafana, where the 03:41.960 --> 03:47.000 analysts can't reach it. So we've got these divided worlds, they can't reach each other's 03:47.880 --> 03:53.160 queries and results, you know. So at Grafana we've fixed this. We've taken a new approach. 03:53.560 --> 04:03.720 We are connecting to Kib and I'll jump forward. And so it means you're no longer having to write 04:03.720 --> 04:10.200 SQL, you can write these dimensions and measure queries and you can click to you're just clicking 04:10.200 --> 04:14.760 on data to filter the whole dashboard and you keep clicking and filtering. This is what you get by 04:14.760 --> 04:22.440 using a semantic layer. And suddenly we're all working side-by-side in the same single pane of glass. 04:23.000 --> 04:29.000 And learn the whole, you've got your logs, query your analytics query and you join them together 04:29.000 --> 04:36.840 with a SQL engine that we've built into Grafana. And that simple SQL query at the very end just 04:36.840 --> 04:44.280 joins the data together and you can see your customers. And there you go. We've released that 04:44.280 --> 04:48.360 this week. For first time and we'd love for you to try it out and I'd love to hear your feedback. 04:48.360 --> 04:58.360 So thank you very much.