DevOps is coming to databases. While there are several companies who have been dedicated to this over the last few year, by all accounts DevOps for databases has been catching stride as more DBAs and organizations realize the benefits of DevOps for databases.
In this DevOps Chat we sit down with Yaniv Yehuda, co-founder, CTO of DBmaestro, to discuss what is driving the increased awareness and what is coming to the market. Yaniv is always a great interview with lots of good insight. Enjoy.
As usual, the audio of our discussion is immediately below, followed by the transcript of our conversation.
Alan Shimel: Hey, everyone. It’s Alan Shimel, DevOps dot com, and you’re listening to another DevOps Chat. Welcome to DevOps Chat today to our – a repeat guest. We’ve – he’s been a guest on DevOps Chat several times. Yaniv Yehuda, co-founder, CTO at DBmaestro. Yaniv, welcome to DevOps Chat.
Yaniv Yehuda: Hey, Alan. A pleasure being here again – pleasure being here today.
Shimel: Yes, thank you. Always a pleasure to have you, my friend. Yaniv, what we’re going to spend a little time talking about today is, you know, as DevOps continues to mature and we see it moving beyond, you know, the orthodox Dev and Ops only to encompass, you know, more pieces of the IT stack, it’s becoming, you know, increasingly obvious to people that having the database and the DBA as part of the DevOps team, as part of the – you know, under the DevOps umbrella – is a must have.
And I guess this is something that is kind of elemental, if you will, to the DBmaestro folks. It’s what you’ve been preaching now for, how long?
Yehuda: For several years. Several years already, yes. Absolutely. Absolutely. No, so I’m in perfect agreement here, you know. It – if you’re trying to start and practice the new or modern process as – may that be DevOps, variations of agility, trying to practice CI/CD – at the end of the day, you need to take care of everything. You can’t just do two thirds or half of the equation. You – it’s like having a car that, you know, you have three wheels spinning at one speed and one spinning slower. So it doesn’t make sense. You really need to cater for everything in order to be efficient, in order to be repeatable.
And recently, you know, talking about DevOps, it’s really expanding. People now really want to cater for security at the same – at the same rate. Because you’re trying to – to shorten feedback loops, you don’t want to deal with stuff too late, so it’s – it’s mainly DevSecOps and this is what we’re seeing from our customers, the need to take care of security, they need to take care of the database, they need to take care of – of agility and productivity together. So with security.
Shimel: Yes. Yep. Now, but Yaniv, when we start talking about, you know, DevOps for the databases, DBAs, and so forth, one of the things – right, the database market is a fairly mature market with more than several players in it. Right? You have your Oracle, DB2, Microsoft SQL, then you have sort of the open source crowd, like the MySQL and the MariaDB, you know, and so forth, and then of course you have the NoSQL group or the so-called NoSQL group which have come on, let’s say, in the last six, seven years.
I would – is there – from DBmaestro’s point of view, let’s say, is there a difference which database you’re using or they – they’re relatively the same across the board?
Yehuda: So discussing, like, from a customer challenge point of view, it’s all of the same. It’s all variations of the same challenges. For Oracle you might be pushing, I don’t know, structural changes and – and index creation or even DMLs just to tune the data and for NoSQL you might be deploying an update or a patch that needs to be implemented in each one of the databases but you have a thread of them. It’s – it’s like shards and – and, you know, replications that – and you actually need to take care of a lot of instances.
So in all cases, you need to have a process to implement changes whether, you know, it’s one type of a database or another. And, you know, it’s just the same problem from different angles.
Shimel: Yep. Is there anything – let’s say NoSQL. Right? Is there anything specific to NoSQL from a DBmaestro point of view?
Yehuda: So first of all, we’re about to ship a MongoDB edition for our software.
Yehuda: And I think, talking about specific databases, you tend to see different aspects of – as I said, different aspects of the same thing. I’ll give you an example, a quick one. So many times, when you talk to DBAs and database developers and DevOps people, they want to move fast. And, you know, people say, move fast, break things, and be effective, be productive, but at the same time you need to be careful what you break because some things are just harder to fix.
So, you know, if you drop a table from production then all break – all hell breaks loose but if you do things that would just take too much time to fix, even if it’s reversible, it’s not a good thing. So you need to come up with a way to move fast, break whatever is breakable, but not break Humpty Dumpty because it’s going to be hard to fix.
And this is where we introduce a lot of safety nets into the process. So these safety nets might be more database flavored and – and I’ll give you an example. So let’s say you have a team of developers who wants to be agile and they want to run quick, they have their datalines, they have their tasks that they need to provide, and they need to go to the DBA and ask him to do stuff. If he actually has to go to the database and log in and run strips and do stuff, it’s – it takes time.
You know, I’ve talked to DBAs who had – you know, a few days of tickets to go through and execute what people want to do. And it’s nothing that really benefits anyone because it’s all about just having the right credentials to the correct place and knowing who asked for what and should he be able to run that.
So we’re automating all of these things in order to make sure that you can run fast, but about the breaking stuff, you break what should be broken. And I’ll give now the example about, like, Mongo versus Oracle.
So for instance, we have what we call organizational policies and this is really the – helps balance the amount of work with the need to control what goes forward and not destroy things. So for instance, you have policies for Oracle or SQL or DB2 that says you cannot drop a table. Why? Because, you know, you might lose the data. Or you cannot rebuild an index on production in the middle of the day. Obviously why.
But when you go to something like Mongo, then it’s all about fine tuning these policies. So if you’re trying to, like, create an index with background faults, that means that when you’re going to try and rebuild it, it will be a nightmare. So it’s all about stopping it at the first time when they – when the person puts in the command and saying, “Listen, you have an attribute missing from what you’re trying to do,” in order to live with that safely later on.
So it’s all about catering for the, like, problem areas of each database and providing the benefit so people can run and if there’s something that they’re doing wrong, they will get very quick feedback, so that’s, you know, all about trolling the feedback loop just like DevOps and Agile are trying to achieve, and this is very much database-specific – so to your question.
Shimel: Excellent, excellent. Yaniv, let me – let me turn a little – well, before we do, I wanted to just get maybe a – give our audience a quick update on DBmaestro. You mentioned a new MongoDB version coming out, but what else do we have?
Yehuda: So actually we’ve – we’ve released a DB2 edition – I’m saying edition, it’s actually everything within the same product. It’s DB2 Sport, so DB2 for the mainframe z/OS and for the open systems – Linux, UNIX, Windows – we’re about to ship Postgre support and MariaDB so this is on top of Oracle and – and Microsoft SQLs that are already out there in the market for quite some time. And, you know, Mongo, as we’ve mentioned before.
So what we’re doing is we’re not, like, sitting in a lab and dreaming, you know, we should do this at – we’re actually, in daily conversations with our customers, getting a lot of feedback, you know, working with the people who need to crack their own tests and trying to make sure that DBmaestro actually helps them in the most efficient way.
So, you know, I’m proud to say that – you know, we have quite a big vision of where we’re going but a lot of it is fine-tuned with what’s happening in the market so it’s not just lab dreaming and wishful thinking.
Shimel: Absolutely. So I – I don’t know if your being here told us everything though. I think I heard somewhere about an award or something.
Yehuda: Oh yes. That is – this is quite, quite new. It’s actually today. We’ve just been announced that we won the Stevie Award for – we won a Bronze award for the best new product. So of course we’re very proud of this and probably will be boasting about this for quite some time now.
But actually, with – you know, it’s – it’s very encouraging because we’re getting a lot of great feedback. The feedback that I like the most actually, more than awards and, you know, the stuff that you put on the website, is just customer feedback. You know, when you – when you see a person telling you, you know, “I used to wait three days for the tickets to go forward and now the developer is pushing changes five times a day, 10 times a day.”
This is all I need. This is, like, what we dreamt when we wanted to – to do what we’re doing, to make sure that the database can play the modern processes, that people can practice, you know, the database stuff safely because, you know, if you’ve been talking to, like, DBAs two, three years ago. I think talking to them about automation would be a big, scary thing because, you know – you need to know what you’re doing, you need to make sure that you don’t introduce any risks. And at the end of the day, this is something that you just can’t do manually. You need to be repeatable and the only way to be truly repeatable is to automate stuff and to keep improving it.
If you do something manually, at the end of the day you’ll make mistakes and, you know, as – as a certain big bird once says, you know, everybody makes mistakes. Oh yes they do. And this is it. You need to make sure that DevOps plays to the values of what you’re doing and – and this is what we’re putting for ourselves as targets. You know? Help people be more agile, be more productive, but not at the expense of safety. So more safe at the same time.
And that, you know, if it can give you the ability to be also compliance-worthy at the same breath, yeah. Sure, why not?
Shimel: I agree – you’re preaching to the choir over here, Yaniv. So let me – I wanted to bring up another topic, Yaniv, and this may be a little close to home for you, a little too sensitive, but, you know what, obviously I follow the DevOps market very closely now for a number of years and, you know, I think back to three or four years ago when, you know, the biggest question we got on DevOps dot com was always, “What’s the definition of DevOps? What exactly is DevOps? Is it just a marketing word?”
And then when we – you know, when it was just Dev and Ops and the idea, when you would talk to someone about, you know, database for DevOps or DevOps for database, excuse me, DevOps for databases, there were a lot of – I don’t know what you want to call them, cynics or just – they were a little bit – a little stiff, who said, “Oh, that has nothing to do with DevOps.” Where do – within the DevOps community itself, how much – how has the attitude changed where DBmaestro and the whole, hey, DevOps is for databases too, has become more accepted within the DevOps community?
Yehuda: That’s actually – you’re just describing my history here. So for a long time I have been trying to educate and tell the story as I think it should be told and, at the beginning, you know, when we were just bootstrapping and didn’t have the product that we’re so proud of today, cynics were even more extreme and it was, like, DBAs on one side, developers on the other side, a big wall in between, people tossing things over that wall.
And actually, if you go back two years, two years back I think, this is no longer the case. People are implementing DevOps not just to be modern. They don’t have a choice. Gene Kim once said, you know, it’s DevOps or die. You don’t have to do it but you don’t also have to survive.
So today – in today’s business, everybody’s trying to run fast. If you’re not running fast, that means that you’re running slower than your competition and nobody wants to afford that.
You know, you had a webinar, I think, with ABN AMRO a couple of months ago. I was just delighted to sit there and listen to how they described what they built. It’s just – it’s just amazing. So – so – you know, it’s a carefully thought out process that takes out all of the junk time out of the process and makes everything efficient, whatever can be automated is automated, and, you know, it saves a lot of time. It makes the whole company more efficient.
And then you reached a point where, you know, you’re looking at what you’re doing and that is working – I’m not talking about specifically them, but in general – everything is working but you’re not as efficient as – as you could be because every time that you go to the database side, you go back – you fall back to the manual processes – the single steps, God forbid, the DBA that goes – and side commands in production to make the change go forward.
You know, the funny thing is when I’m recruiting people to the company, I’m telling them what we’re doing and they said, “This is amazing. This is what we’ve been doing all our lives wrong and this has always been a headache and you’re just solving it.” So that’s – that is certainly something that has drastically changed and I think the last two years has been an accelerated – I would say, accelerated approach to the whole thing that – it cannot be left aside. You cannot just do application, you cannot just do database, by the way, as well.
You have to introduce a testing automation. You have to build with your infrastructure automation. So may that be containers or virtualization depending on what – where you’re coming from. It all has to be combined in order to reach that Holy Grail of efficient automation.
Shimel: Got it. Excellent. You know, it – I think it’s a good thing for the DevOps community as a whole but it’s interesting, Yaniv, you almost have to be an evangelist to the DevOps community at the same time you’re evangelizing to the database community because I don’t know if they raise their hand and say, I want to be part of that DevOps thing too. Right? They – they need to be shown the way and evangelized and converted, if you will.
Yehuda: So on this we do see a changing approach. It used to be, like, trying to convince why they should care. Today it’s not – I’m happy to say, it’s no longer the case. Some customers we have, you know, DevOps champions that push us forward. In other organizations, it’s the DBA or the DBA manager because they see what’s happening. They see the projects around them in the company.
They don’t want to be left out and there’s – you know, there’s really no gain in – in doing a mundane job. Let’s do the fun stuff. Let’s – let’s improve logic, let’s fine-tune the database. Let’s do the stuff where, you know, computers can’t – can’t still, at the moment, solve the problems for us and leave that, you know, the mundane – mundane back-work or whatever to automated process.
Yehuda: And it’s not just DBA. You know, today you talk to – to, you know, all of our European or the ones who – who need to cater for GDPR. They have a set of new problems they need to cater for. It’s not just, you know, automation. It’s not just DevOps, they need to now make sure that they manage their roles and responsibilities and auditing and obfuscation of data, if you’re taking, like, data from production, now there’s no private data inside and how you do that process. So you want to have holes.
You have automation for many reasons now. It’s not just efficiency. It’s creating the good process and sticking to it because if you have a great process but it’s just a manual process, each time you do it, you’re doing it slightly different. And this is exactly where I think the whole ARA market is spot on. This is why we’ve – I think we’re seeing the changes we’re seeing in the market in the last couple of years.
Shimel: Agreed. Excellent. Yaniv, you know what, I apologize, we’re way over our time here. So we’ll wrap it up. Before we wrap up, any last thoughts – anything you want to share with the audience?
Yehuda: No, I think – I think, as you said, we did – took a bit more time so I think we can wrap it up.
Shimel: Okay. Well, look, I wanted to mention that DBmaestro will be at Gene Kim’s DevOps Enterprise Summit in London, I think June 22nd to the 24th, maybe it’s the 23rd to the 25th. Something like that, towards the end of June. So if you were planning on attending DevOps Enterprise Summit, London, please stop by and say hello to the DBmaestro folks. And then Yaniv, you’ll also be at Cloud and DevOps World in London in June as well. Correct?
Yehuda: Correct, correct. Yep.
Yehuda: So, as I said, going – meeting our peers, meeting our partners, meeting our customers – just talking.
Shimel: Excellent. All right, Yaniv Yehuda, CTO, co-founder, DBmaestro, frequent DevOps Chat guest, thanks for being here today with us and continued success.
Yehuda: A pleasure being here again. Thank you. Bye-bye.
Shimel: All right. This is Alan Shimel for DevOps dot com and you’ve just listened to another DevOps Chat. Have a great day, everyone.