SolarWinds’ Database Observability – Kevin Kline, SolarWinds

This is techstrong tv. Hi everyone. Welcome back here to techstrong tv.

Our next guest today is Kevin Klein. Kevin is a technology evangelist for databases over at SolarWinds. And, uh, excuse me, it's Kevin's first time on Techstrong tv.

Kevin, welcome. How are you? I'm great, and it's a pleasure to be here.

Thank you so much. It's a pleasure to have you on, Kevin. Um, so technology evangelists for databases over at SolarWinds.

Tell us, you know, how, how did you wind up here, Kevin? Well, it's, it's, uh, like all of us who have some gray hair, um, more hair at all. Any hair at all, exactly right.

A circuitous route. Um, but I started my professional career in the mid eighties, uh, in a city called Huntsville, Alabama, which is Sure, uh, actually very, there you go. And, uh, my dad was a rocket scientist.

I later started to work for NASA in, in the eighties, and I was a good FORTRAN programmer, and I also knew what a select statement was, and that was pretty extraordinary at that time. So, um, mm-hmm. That's where I started my career working on very early versions of Oracle.

And, uh, I was also working towards a master's degree. And, uh, I saw in the back of a magazine I liked a lot of the time called Data Nation Magazine. They had a call for authors.

And so I sent my, uh, social, uh, my, uh, master's degree, uh, abstract to them, and they said, does it make a great book? So, uh, soon thereafter in the early nineties, I wrote my first book about Oracle and, uh, continued to write books over the years at each of the different jobs I held throughout, throughout the many years. So, after I left nasa, I moved to Nashville, Tennessee, where I live today, and, uh, was part of the database administration led, uh, the database administration team, uh, during the nineties, uh, when we first converted from mainframes to relational databases.

So, worked a lot with Oracle and SQL Server. And then, uh, because I used so many SQL Server and Oracle tools from the big tools vendors of the day, I actually jumped ship and went to one of the, uh, the big, uh, software manufacturers for a product called Towed, if you've ever heard of Towed for, uh, a very popular Oracle tool. And so that's what set me on the path to be, uh, somebody who specializes in databases, but rather than being a D B A or, uh, that kind of thing, I use those skills, uh, as, uh, a company making tools for DBAs.

Fascinating, fascinating. And of course, you've authored a, a number of books now. Yeah.

You know, uh, this is my bestseller, uh, sequel in a nutshell. If you can see that one. Wait, wait, turn that the other way.

How thick is that book? Yes, this, I like to call this one. That's a nutshell.

It's a, it's a coconut shell. Yeah, yeah, exactly right. So most of the nutshells are like a, a walnut shell, right.

To 300 pages. But this one, because we cover the anus is, sorry, I said It's an opus. It is.

Well, and what it is, is nowadays you can get so much information from vendor websites and documentation or even, uh, conversation sites like Stack Overflow. We need a little different value add for this particular book. So what it does is it covers not just the anz uh, SQL Standard, but it also covers all the implementations for Oracle SQL Server, the two biggest, uh, commercial databases, and the two biggest open source databases.

So that would be Postgres and my SQL slash Maria db. So you actually get five books in one with this. Wow.

So it's a great deal. Alright. A lot of work.

Nice work. Absolutely. Kevin SolarWinds is not a stranger to our audience, but there might be a few people out here maybe who either have not heard of SolarWinds or not sure.

But how would you describe SolarWinds to them? SolarWinds is a, a company that is dedicated to making the life of administrators and IT professionals better. And so much of that is dependent on awareness of how your different infrastructural components are operating.

So we focus a lot on monitoring and observability, as well as other tools that enable you to accelerate, uh, development processes, uh, to, uh, keep track of your infrastructure, uh, to alert on issues and respond to them. So we have a number of product lines, not just in database, but for networking, for example, we have that for other kinds of, uh, application monitoring. And so there's a whole variety of products, and there's a little bit of, uh, uh, a little bit of territory that comes with my title of, uh, database technology evangelists, since we have three other major product lines.

Sure. Excellent. Thank you for the, for the update.

com is the website if people want to go find out more. All right. So Kevin, uh, thanks for all the background.

I wanna jump in today to the topic of our discussion, which is the SolarWinds database observability solution. And I don't know, that's a mouthful. Is there a particular name for this solution, or is that the actual name and tell us about it.

Yeah, so SolarWinds Observability, um, SWO, s w o is, is our new generation of observability and monitoring tools. So when you think about monitoring in some ways that's, uh, a 20th century, early 21st century, a way to approach awareness of your systems and how well they're performing. Uh, however, you know, the big transition between early two thousands and today is that so much is cloud-based, uh, cloud first.

Uh, we have so many different SaaS products, s a a s products that we need to understand how they're performing and how they're interoperating with their other systems. So many of our our products are traditional on-premises monitoring products, whereas this is our entree into observability, which straddles both on premises and the cloud, um, heterogeneous environments, multi-cloud environments, as well as, um, the concept, one of the key concepts that is different for observability than for monitoring is monitoring is kind of, I know what I need to look out for. So I set thresholds C P U over 80%.

I want a warning, um, IOPS over a certain amount. I want a email warning. Observability differs from that in that, uh, what we are examining and keeping an eye on is potentially a black box because it's a service we get from a different provider somewhere out in the, out in the cloud.

And so we look at it in a different way and we determine its performance, uh, or lack of performance based on known outputs. We may not know the internal, uh, metrics of a given service because it's provided by someone else, uh, but we can tell by its outputs how things are performing. And so observability is, uh, a little bit different than monitoring.

It's kind of a superset of monitoring, if you will. I love it. Um, look, observability of course, is all the rich.

One of the interesting things about observability, Kevin, is I, I think why we've seen this, I don't wanna call it hype cycle around it, but why we've seen it gather a lot of attention is, is it's open source kind of, uh, roots not open source. You know, I mean, this, let me be plain. There's a couple of open source projects that are huge that are really powering this whole observability revolution.

What is, is, you know, is SolarWinds U utilizing things like Prometheus and Open Telemetry and so forth as part of this? Or is this outside of that? Um, well, that's a, that's a great way to kind of stage the question because we see open source, uh, in so many ways of becoming part of, part of the inf um, enterprise infrastructure rather than just kind of the, the, you know, one time it seemed like, uh, my sq the lamp stack, you know, Linux, Apache, my sql, P H B, um, those early open source products were kind of for startups or for really lean organizations, you know, three developers and a dream that they wanted to bring something to market.

But now we see open source, um, really encroaching on what was only available to the biggest commercial vendors, you know, so, uh, we see Postgres doing, uh, growing extremely quickly in the relational database market at the cost of those who are selling their database products like Oracle or DB two, or some of those old school relational database platforms. Um, so yes, we see that encroaching everywhere, uh, in a positive way. On the other hand, like you said, there's certain sorts of products and, um, and projects, if you will, like Prometheus, or we have things like, um, uh, Nagios and other kinds of tools Sure.

That, you know, let us keep a, an eye on our enterprise. And what we see is that, uh, we personally are not currently using those. Um, and I really can't speak too much about futures of where we're going, although there are plans and analysis being conducted right now to incorporate certain, uh, certain elements of the open source, uh, world.

However, um, the, the threat, I shouldn't say a threat, the, the new coming, um, architectures that we're seeing really do use a lot of that. So we're aware of it. Yeah.

And we're building that out to meet our customers needs. Sure. One other kind of, I feel like Dustin Hoffman in the graduate, two words, ai, well, two letters.

How, how's AI playing out in all of this stuff? Right? I'm thinking plastics, right, right.

Well, well, you're of an age. We're dating ourselves, but go ahead, aren't we? Yeah.

Yeah. So, uh, great point. And that is, um, one of the key elements that is so different about, uh, what we offer, uh, in our product lines as opposed to what you would get with an open source monitoring kind of product, like you said, Prometheus or, uh, some of the other ones.

So we do have built in right now, uh, our own, uh, proprietary ais. These are for things like anomaly detection, um, and sometimes the differences, uh, between today's performance and yesterday's performance is subtle enough that we know it's slow, but it's hard to identify if we're looking at a chart full of numbers. So the, uh, the ais that we have constructed are, um, uh, primarily directed at that sort of activity.

Hey, how's this this week different than last week? Uh, can you help me characterize, uh, you know, do we have a difference in io? Do we have a difference in network latencies?

Um, but reading the lines a little bit about your question, I think you're asking about generative ai like we see with chat G T P, uh, G P T and, and the Yeah. And the other types, um, stable diffusion and what have you. So that's really new, you know, it just, it just hit the, um, the streets a few weeks ago.

And so, uh, we haven't incorporated that yet into our products, but boy oh boy, are we spending a lot of brain cycles, uh, thinking about what would be the best way to, to build that in. So at some point in the not too distant future, we'll, we'll be seeing that. I, I agree with you.

Agree with you. You know, one of the interesting things about the SolarWinds observability solution is it has capabilities for both cloud and on-prem. Right?

I, I think, you know, in talking to the amount of organizations we talk to here, no one raises their hand and says, well, I want one set of solutions for my cloud, and I want another set of solutions for my on-prem infrastructure. Cause oh, you know, and then maybe something else for when they mix, right? Because I don't have enough to manage Everybody, Everybody, everybody would like to move to a single plane of glass as much as possible and, and wherever possible.

Let's talk a little bit about the SolarWinds products capability around that. Well, um, you're absolutely right. And one of the other things that we see, uh, in communicate, you know, and I'm talking to, um, to my peers, DBAs and data analysts and so forth, people who are, uh, you know, hard at work in, in this line of business on a regular basis.

And we knew that there were long ago, we knew that this was, uh, an issue. Uh, for example, uh, in serving many of our thousands of our bigger customers, um, about 70, 75% of the customers, even in the early to mid two thousands had multiple heterogeneous databases, right? So maybe they had Oracle, they had SQL Server, uh, the people in finance went out and bought a DB two product.

So now we gotta keep track of, so your point about a single pane of glass is very valid. Um, even in the preh hyperscaler pre aws, pre, uh, Microsoft Azure days. Now, um, we have additional nuances to that.

We, we have containers instead of just virtual machines, right? So we have VMs, we have, um, you know, real iron, big iron, uh, or racks of in our, um, data center. Uh, and then we also have all these, um, different multi-cloud scenarios too, where maybe the company has decided, uh, for, for certain situations we'll buy from one vendor and others will buy from a different cloud vendor.

So that is a massive challenge bringing all of that data and information, uh, and particularly because you have to have some subject matter expertise to know, you know, what a big query problem might be on Oracle versus SQL Server. And so we have a couple solutions to that. Um, right now, one of our, uh, main products for heterogeneous DBAs is called dpa, uh, database Performance Analyzer.

And so since all databases, uh, today holding common, the idea of a weight statistic, uh, internal to the relational database, and even some non-relational like, uh, MongoDB, they, they keep track of what is making you wait on the completion of the query. So it could be network latency, it could be io you know, maybe you've got a pure storage device and got beautiful IO for one set of systems, but you have an older standard kind of san how's the IO different? And so, um, our systems will look at all of those different, uh, components, including the different cloud vendors, and bring those together for you in D P A based on the weight statistics, and then you can, uh, troubleshoot or debug from there.

And, uh, as you brought up earlier, there's the AI that'll also tell you there's a big anomaly of some kind or a subtle anomaly that you can also troubleshoot. And then, uh, I'm Sorry. No, keep going.

No, no, keep going. Yeah, so on our database side, we actually have a couple solutions. So, um, then we ha, so that is D P A, which our heterogeneous product.

Then we also have SQL Century, which is our very deep SQL Server product. So it goes way beyond what you would normally get, uh, from any other vendor or any other set of, uh, tooling out there on the market. And so it gives you extremely deep information, but it's specific to SQL Server and Azure SQL in the cloud.

Excellent. com earlier, but for people who want to, you know, get specific information here on observability, the SolarWind's observability, wo as you called it, well, where on the website can they go? Uh, and I should point out that SWO is our third product, our newest product.

It is the one that is entirely, uh, sas. It's primarily, uh, built for SAS applications and databases and so forth. And it also includes all of our, uh, no SQL monitoring for Cassandra and MongoDB and other popular no SQL databases as well.

So, Kevin, thanks. Thanks for making us smart today. A little bit about Swo and all of the different observability solutions at SolarWinds.

It was fascinating listening and, and learning. Come back another time and keep us posted. Thank you so much, Alan.

I really appreciate it, and I look forward to talking to you again sometime. All righty. Kevin Klein, technology evangelist for databases at SolarWinds here on techstrong tv.

We're gonna take a break. We'll be back in a moment.