Nineteen thirty two dot org. The information economy as a ride. The world is teeming with innovation as new business models reinvent every industry industry. Inside Analysis is your source of information and insight about how to make the most of this exciting new era. Learn more at inside analysis dot com, Insideanalysis dot com And now here's your host, Eric Kavanaugh. Well, ladies and gentlemen, Hello, and welcome back once again to the only coast to coast radio show
all about the information economy. It's called Inside Analysis. Your host here, Eric Kavanaugh. Very special show today, folks. In fact, we just recorded a nice session with some slides. Hop online to inside analysis dot com and go to the webinar section to find where you can watch that show. Later on we'll put it on YouTube. But it's all about streaming graph at a company called that dot which has invented this technology. It came out of
DARPA and we have Paige Roberts with us. Will also be hearing from Sanjif Mohan of Sanjmel formerly of Gartner. Now he's out independent so we can do cool stuff like this with us, which is awesome and a long time listener, first time caller h Alexander Husky is out there from Exxon Mobile. He is a real aficionado of technology and architectures and he always asks great questions and he's always challenging me on some of the things that we say in the show.
So it's good to have him here in the mix as well. And with that, let's dive right in with Paige Roberts. So, Paige, first of all, congratulations on lining up with this company. Is absolutely fascinating. I think there are a whole host of use cases that simply cannot be reasonably addressed by existing technology, which you are uniquely positioned to tackle. So with that, give us a quick overview of that Dot and streaming graph.
Sure, so that dot was, as you said, the technology of quine open so was developed by DARPA and then that DOT was developed as the as the company the commercial side of that. So that Dot streaming Graph and that dot Novelty are products, and they both are super useful for uh, finding really advanced threats in cybersecurity, and Novelty is especially useful for finding unknown unknowns,
the things that you don't even know to look for. So you can just feed it your data and it has an AI built in to find whatever is an omalis in your data set as it flows in. Eric, I think you're talking about I'm not hereious, I'm I was on mute. Sorry there rarely do that, but it's okay. One of the very interesting use cases here, of course, is the fact that typically analytics is fixed, you have a structured query on set unfixed data. Well, streaming comes along
and you could do persistent queries on the streaming data. Now, they've been streaming analytics solutions for a while, but maybe you could talk just per a minute about how that dot achieves this. So you have this slite up for our studio audience, but talk about how you have a persistent query on streaming data. But then that graph can alter over time, so new nodes can be sort of dynamically created as new information comes in. Can you talk about
that a little bit? Sure? So the concept is you have data streaming in from multiple data sources, different Kafka topics or kinesis or pulse are Plus you're also probably got some context data coming in from some sort of batch capability like a file or whatever. And we have two things inside of that dot or colin that you do and that is first one is called an jest query.
Now, in jest queries turn data into graphs. The idea in our where is that all nodes already exist, and what you were doing as the data comes in, you are using an extension of the cipher graph query language, which most people know that it worked with graph to tell it which pieces of data need to be, nodes need to be, properties need to be, edges, relationships, and as that data flows in, the graph becomes more and more rich, and it's a dynamic graph, so it changes over
time as the information comes in. Now, the other thing that we do is called standing queries, and these are again just cipher query language, So it's anybody who's used NEO for j or any other graph capability is probably familiar with this language. But the concept is that you can query the past, the present, and even the future all at once. There is no time
window, there's no definition of what time you're querying. So if a piece of data came in a year ago, and a piece of data comes in a microsecond ago, and then the final piece of the pattern you're looking for comes in next week, as soon as it comes in, it finds it. It finds that pattern within less than a couple of microseconds and immediately gives the answer and puts that into a new data flow like another Cafka topic.
Or it can trigger actions within an application, which is really common. It can also go into a real time monitoring system and say, hey, you know, flag it. This is something important, this is what you were looking for. It can help with predictions, It can do all that sort of thing. So and of course the data is persisted in it's pluggable.
You can decide what you would like anything that's Cassandra or Cassandra API compatible like rockstv SCYLIDB things like that, or ClickHouse or a couple of other things. So you can persist that what you need to recreate the graph in case you lose power or something, because it's all in memory or you know. It also persists the cool new insights that you've gotten, the places where you've actually found what you were looking for. So that's the cool idea, and you
and you preserve state too. Write In other words, as the draph changes, you can still roll back in time and see what the world looked like at a particular point in time. Right, because if you change a schema typically you're changing the view of everything, but you can roll back and see what it was like to do some sort of comparison. Is that right? You can? There's also a really a big value to that because a lot
of graph algorithms are designed to look at static data. So because you can take a snap shot from whatever time you wish, you can then use that to train your graph algorithm if you'd like, and then apply it to predict what you're what you're looking for. Yeah, this is just fascinating. Well, I'm going to be quiet for a while and bring Sanjiv Mohan into the picture so he can ask some questions. Look him up online, by the way, Sangmo Sanjeev. I'm fascinated by this stuff. What do you think?
What are you seeing? Yeah? Eric, me too. In fact, it is ironic when you invited me to come up this show. The two topics streaming and graph were part of my podcast and blogs just in the last two weeks, and it's fascinating. In fact, in my document, without knowing too much about that Dot, I'd already mentioned this product, so
it was quite quite a I guess this is this is that dot. This is like Paige and I've already connected in this one hundreds of millions of dots in a graph, but with this event our relation, the edges certainly came alive and those dots are now like, you know, the focus is on that, and I think that's why you call it that dot page. I know you're happy. There's actually there's actually like our our founder is a big
geek. So essentially when you when the data flows in, it goes into a variable called that and every time you tell it, you know, say the the IP address is now the node I d it has in the code, it'll say that dot node address equals node ID or something along those lines. So the code is full of that dot something. So that's pretty much I think where they where the original name came from. That is great. The other thing that that struck me is that the note already created, they
just coming into the context. Yes, right, all notes exist. That is one of the fundamental process so that that's quite a deviation from a typical graph database. The other thing that's also a deviation, and that's why I find this to be fascinating, is in streaming stream processing event stream processing, there's always this concept of a window. That window is like either bounded for a certain time and then it slides or it tumbles. So there's all these
things. And what I'm learning is that in this scenario there are no windows because windows sort of constrain you, and you may have a later arriving transaction, maybe hours later, weeks later. So the standing queries, that's a common concept. I think we call it continuous queries. You're calling it standing queries, correct, Yes, So the query is always running, streaming data is coming in. There is no window, which means I can identify patterns
even a week later, yes, even a year later. Yeah, We've had we've had tests up to three years of data and you can still less than five microseconds, and I'm being generous with that. It's usually less than one, and less than three is the most we've ever seen. So page give us an example. What was this use case where the data just showed up a year later? And then what insight was generated? Not sure if you know, but I think the one that the first one that always comes
to my mind is the one that we were invented for. I mean, we've we've done a lot of things since then, and there's we're great in financial fraud and things like that. But the big problem that we see the most often is the advanced persistent threat thing. That's that's why DARPA developed us because bad actors are as I think Eric mentioned earlier. They're not a single
guy in a hoodie in a basement somewhere. It's a government, it's a bad you know, it's a whole bunch of people, or a criminal organization. It's a lot of people. And they're very patient and very smart. And what they do is find a way to get into a secure system. And once they're there, they just sit there and wait for those time windows that you're talking about that everybody else depends on these time windows, and as soon as they slide past them, them getting in, that first part of
the pattern is completely invisible. It's gone. They might they look like just like any other legitimate user of the system. So if they then start accessing data or maybe even finding a way to copy it to a temp file, and then somebody else just magically comes in and copies that ten file externally and boom, you have stolen something that's very important, or in some cases they go in and do damage. I mean, we've had instances lately of people
doing this sort of thing. It's like Microsoft and water utilities and lots of problems with this particular difficulty. And that's why DARPA developed the software is because we just can't have that continue to be a problem. It's like we need something to address it. And once we invented the technology, then we realized, oh, this is really good for a lot of other things as well.
But that's what it started as it's to summarize, So go ahead, Alex, Hey, Alex timing, Go ahead, Well, I was just going to say to get my mind around streaming versus static and a year old
or two year old data. It doesn't seem streaming to me, though, So how do you So the data streams in and is capped, it's persistent, so you can stream data in on a cluster and have millions of data points your graph as it as it gets new, data will constantly grow and it will become more and more elaborate and more and more capable, and it might have three years of data if that's if that's what it takes to find what you need. But it a data continuing to stream in, so each
time it right more information, it becomes more elaborated. We have a reposit. So you have a repository and you want to look back and analyze it stream it into your system. That the no because because the majority of our capabilities are in memory. Is so actually all of this is in memory and it doesn't take a lot of resources. You might think, oh my god, this is gonna this is going to require massive beefy round nodes with that are expensive and all of that. We have actually done a demo on the
smallest Raspberry Pie they make. Our technology usually takes less than it usually takes between It usually takes about one hundred and fifty mics of RAM, and at the max when it's really just going strong and doing like I said, three years of data whatever you need, it uses about three hundred megabytes of RAM for machine not really so the scaving. I'm starting to understand that the analysis is taking place of pattern plaining, pattern recognition. All that's taking place continually.
You're not storing all this data. It's coming in being processed and either alerts are showing up or some kind of analysis is happening, right exactly that that's correct. We do persist everything, but it's just in case you have a power outage or or you need to go audit it or something like that. It's it's not persistent for the for the technology to function. It's persistent in case something bad happens. Oh yeah, I totally, I totally get
that. So persistent that would be I'm just trying to get the synonyms right here, data persistent, you know, Cassandra, that's really just a database, right. Wouldn't you be able to use Snowflake for that too, No, just because we're not set up to support it. It's very It's not because there's anything wrong with snowfl like, it's just because it's not not designed for that. We recently had a large customer request that we add ClickHouse as
a capability. We didn't have it before, and now we do. It's pretty much just a matter of adding a new persistor to the technology, Cassandra. One that's most common. I'll have to look into the other. I just want to jump in and ask one more thing. But maybe it's the novelty module. But I was. I read your website and instead that the okay, this this technology does this pattern recognition, finds fines of things, but it does mention in there somehow a person has to set up the pattern
that you want to find so it'll find it streaming and all that. But I got the idea that somehow you have to define a pattern that you're looking for. So there's two technologies that we're talking about. One is the that dot streaming graph, in which case you need to define what it is you're looking for. And the other is that dot novelty, which has a built in pattern learning AI which figures out what what is a normalist in your data
set. It doesn't require any training, it doesn't require you label the data or anything or even know what pattern you're looking for. It will just look you link them together and figured out would you link them? Sorry, yeah, it is kind of crazy. So would you link those together? Like would a company set up a novelty first and then pipe two your streaming graph.
Yes. In fact, we've talked a lot about advanced persistent threats, but there's another really tough one to find, which is an insider threat, which is when you have an actual person who is a legitimate user who is nonetheless doing something they shouldn't. And there was a big contest recently to try and find you know, what's the most efficient way to find insider threats,
and we pretty much found a far more efficient way to do that. We can do it in a few seconds using first streaming graph to find a particular pattern and then novelty to find anomalist behavior. And you can you can absolutely stream them together and make them do do their thing separately and put it together and come up with something exceptional. Yeah, like in a workflow. Basically, this is cool stuff. We're coming up on our first break, folks.
But we've been sitting here talking about streaming graph. It's fast technology. With Paige Roberts of that dot. We've got Sanji Mohona Sangmo and that was Alex Husky chiming in from Mexican Mobile. Will be right back. Don't touch that down. We're listening to Inside Analysis. Welcome back to Inside Analysis.
Here's your host, Eric Tabanac page. I have a question without talking about the data persist for Faull Tolgrand persisting the data in case of our failure, if I just want to be I just want to talk about that in memory piece, what do you by any chance have an ability to replicate from one cluster to cluster, memory to memory for maybe data residency reasons or through port
or something like that. I honestly don't know the answer to that. I think at this point it's like, Brian, are you who invented this offer to be on the show? And I'm like, oh, oh, I wish I was here. Yeah, I'm not. I'm not really certain. I'd have to get back to you on that one, Okay, no worries, Okay, because I know, like some company like Redit's, you know, they do that, you know, cross replication across clusters. Because three
years in memory sounds like a long time. It sounds like a long time. But we've done it, We've tested it, you know it works. We have I mean, we have some of the more advanced uh cybersecurity folks doing it now, so we know it can be done. The I think probably the answer to your first question is if someone needs that, if we have a We're a small enough company that if a if a large customer comes in and says, we want this, but we need this capability, We'll
build it. That's been It's kind of the difference between a big company and a large and a small company is the small company goes okay and builds it. The large company goes I got five hundred other people and need something else, you can just wait, right. We actually have an interesting question from an attendee in our virtual studio audience here, and folks, if you want to be in the virtual studio audience, just top online to inside analysis dot
com and register for one of these events. But one attendee's asking, what about competition? Would X A Beam be a competitor? I mean, I haven't seen anyone doing exactly what you're doing, so I don't personally know of any competitors. But who do you run into other companies when you're out there? Tell us about the competitive landscape page. There aren't a lot of competitors because it's such a new technology. I think some people are trying to do
something similar. Memograph is one that I've seen that's sort of approaching something like this. I think the LinkedIn liquid uh, where they're their graph capability that they're doing behind the scenes on LinkedIn is probably pretty similar. That's it. I mean, that's all I can think of a few of I think stream graph, there's stream stream set. I think there was stream sets. There's a lot of event stream processors, uh fleet yeah, stream stream sets.
Guys like that, but very few of them are doing I don't think any of them are doing the graph paradigm alerts. But is that that's centering on the alerts that you is that what we want to we want to Gardner, right, is alerts quick warning to shut something down or whatever. They're not doing that, what you guys are you know, they're just not They're not doing it as a graph, and they're not they're doing it with time windows.
I think, Sanji, you've mentioned pretty much all of the other events stream processors are limited by their time windows, and they have to be. It's just the way they're designed. And we were designed from the ground up to not have that limitation. So it's very it's different in that way. And the other difference is that all of the other events stream processors sort of perceive data as almost a rowan column, as if it were about to go
into a normal database, whereas we perceive it as a graph. And one of the advantages of that is the categorical thing. So if you want to look at IDs, IP, addresses, names, you know, all the things that are categorical as opposed to you, you know, temperature, which
is you know, up and down it's got a numeric range. If it's not a numeric data, most of the state of the art is to take that data and convert it into numeric and it's huge and bloated and sparse and really hard and takes a long time to annaly lies, whereas graph data is
already in a category and we just analyze it immediately as is. And this makes it much better for finding things that are for answering certain questions, and in particular questions that are categorical in nature, like who did this, Where did you? Where did you do it? What what you know? When? When was that? When did that happen? What is the relationship between this person and that person? All of these are our categorical questions, and
right right as we do the graph should it makes sense. I see some maybe inspiration from Mark logic. It's still out there under a different name. Now what did they get? But they were big on the the semi structured, not having to be wrong on column, not having to be sequel friendly. But they can still make associations the way you are. But they weren't dwelling on the streaming aspect. Well. In the the Mark logicts of the
world also do. Uh. I mean they pretty much we expect our data to come in in that semi structured form like Jason or something along those lines, log formats that sort of thing, but when and then we immediately turn it into a graph. So the semi structured formats are not really ideally suited to find those relationships and patterns across times and individuals and you know, IP
addresses and networks and all that sort of thing. So they're they're designed to be sort of self contained so that I can send a single message and it has all the information for you to understand that message. So they're really good at that, but they're not so good at finding the relationships between multiple things. So it's interesting. So what you think is that the input is because the input is cough cop also kin so so payload is Jason. But you
are then ingesting that and converting it into a graph model. That's correct. That's that's what the ingest query is for. It's pretty much as the Jason comes in, which parts of it do you want to be a node? Which parts of it do you want to be a relationship? What should be a property? How do I take this data and turn it into a graph. That's that's what is a graph model? Is it a property graph or is it an IDF. I think the answer to that is IDF. But
I don't vote me. I'm not one hundred percent certain. Well, Sanji, I think you. I think we need to preview, we need to get all of our techies on there and ive deep on that. Yeah, because cipher was bolted by NEO four J and cipher is a property graph. So yes, yes, And Sparkle if you said you sports sparkle, then that sounds like it maybe an rd EF. Yeah, and that again you're you're asking the wrong person on that. But cipher is our main language that
we support. We're looking at supporting some others, but we're looking more at like graph q L and some others. I don't think we're looking at sparkle because and I think that answers your question actually, because it's not the right kind of graph. So I knew that, but I did not know the technical terms. Yeah, we for financial financial fraud. I'm sorry, Eric, go ahead, No, you go ahead. That's good. That's good. Well, all right, let's say let's say a big company like the
representative here, but I'm an advocate for honesty. So so if they're interested in, let's say, finding fraud within or the company were outside, you know, the people we interact with invoices and all that, how does h your screaming graph help find a financial fraud? You've kind of like a banquet.
But I just wondered, if I don't know about Benfort's law, the incorporate that sort of algorithm in there already, or are you able to without that kind of thing in some kind of innately find that these patterns of cheating going on? Well, I think that's it. There's two things going on
there. One is, if you're trying to look for a particular pattern and you already know what fraud looks like, it's like, you know what a fraudulent actor is likely to do first, second, third, fourth, fifth, And if if you then find somebody doing all of those, you can immediately go, oh, that's fraud. And the thing we do there is we're using pretty much the same patterns as everybody else, except we don't have to take the categorical data, change it to numeric, then train something,
then come back and predict. You can actually do it in line. You can do the analysis on much more quickly. So a lot of what we do is not necessarily different fraud detection. It's faster fraud detection. And the other aspect of that is because we have novelty. A lot of times, if you don't know what fraud looks like, if you don't know what you're looking for, you can feed it into novelty, and novelty will find the
things that are unusual that are different. Like here's a good example. If I am have just paid for a hotel room and dinner in Paris and then lunch, I pay for lunch work, You're gonna the anomalist thing is the
New York and you're going that's probably a fraudulent transaction. On the other hand, if I know that my pattern is, you know, multiple purchases in a single location and then purchases in a different location, then I might be looking for that pattern, but I'll probably get a lot of false pots. You know, I got on a plane and troubled somewhere else. You know that sort of thing. I travel a lot, so I used to get false positives a lot, and we've gotten better over time about not catching that
sort of thing. So I don't you don't get my card turned down when I'm in you know, some other country doing I mean there's some other city doing a conference or something like that. The novelty is really good at figuring out unique is not necessarily novel and finding something that's when lay novel is very
different. RUD be great to see if you have templates, let's say, by know, you might have been working with big businesses that have tons of invoices and that sort of thing, and maybe have a template where where not everything has to be custom made. We do. We have what we call recipes in the koin open source capability so that you can just pretty much pull
it up and it's used. It uses publicly available data. There's been enough like fraud contests, cybler security contrastests, things like that out there that there's a lot of publicly available data which we know that data has the information to find the problem, and it's just a matter of how good are you at finding it. So we can just pull in that data and we have a recipe which is essentially a self packaged code bundle that you can operate yourself on
your own laptop and see how it works. And those are those are those are really cool for learning how the tech works and its capabilities. Yeah, Alex was a musing earlier before we hit the record button that it's hard to find time to play with things, but this looks like something that's a lot of fun to play with because you do have to kind of play around and I found an interesting concept here as I'm just trying to wrap my head around
it myself. Here's a question that I bet that dot is very good at and very other solutions are not good at, which would be something like find all the people who interacted with this potentially fraudulent account in the last hour. Right. So, a traditional solution, you're not going to get there. You can't get you could run reports, it's going to take hours, it's
going to take longcount you just won't get there. Whereas this, because it is graph oriented and you're noticing all these nodes and edges and you're constantly streaming, that's the kind of question you're get to answer to quickly. And you know, from my experience, once you understand the kinds of questions you can ask from an analytical tool, that's when the possibilities open up, right, Because I think there's a lot of sort of built in lack of enthusiasm about
asking questions that can't be answered. And so if you know a question could be answered, you start to ask it. But if you think it's going to be a long running query or you're just going to get in trouble for doing it, you just don't go down that road. You don't know that it's possible once you see that these things are years people absolutely right. I mean, I think that we've trained our analysts to think that certain things can
be asked and certain things cannot. And this is one of the technologies that you have to kind of change your mind and go, I can ask that. I can ask not just you know, how many people has this person interacted within an hour, but how many how many has it interacted in the last year. It's like, I've found this person. Now I know they're they're a frauductor. It's like, oh, but you know in the last last year, I didn't know that there's you know, fifty other people they've
interacted with since then. And I can get that answer in seconds. Yeah, I think this is the streaming part is super exciting. That graph piece. In fact, the recent graph databases became quite very known many years ago
was because of this massive scandle if you remember, called Panama Papers. So somebody uh exposed a treasure trove of documents called Panama Papers, and these journalists got hold of those documents, but they contell memos and all that, but forever to actually go to forever until some journalists discovered Draft database and said, let me just ingest it into graft database, and the whole story unraveled.
But it took a long time. And now comes that dot and quine, which is saying that what if this data was in real time streaming, we can find all the relationships and patterns, uh even go back into the history and these insights. Yeah, that's right. Well, folks were up on our next break here, but we're going to pick that up right after the break. This is really interesting stuff. Don't touch out how folks, you're listening to Inside Analysis. Welcome back to Inside Analysis. Here's your host,
Eric Tavanaugh. All right, folks, back here on Inside Analysis talking to an expert panel. We've got Paige Roberts of that Dot. We've got Sonjif Mohan Sangmo, and our good buddy h Alexander Husky from Exon Mobile representing himself. Just asking some fun questions, and Paige, I'm gonna throw this one back to you because it just it finally opped in my mind. The real value here is getting through to people, to the analysts that they really can
ask these very interesting, gritty questions. And you know, the unveiling of this, the timing is actually pretty interesting because there's something else happening in the AI world called large language models, which have taken the world by storm, quite frankly and for good reason. And I've spent a lot of time researching
these things to understand exactly how they work. And they use probabilistic math, and they convert text into vectors basically, so you have vector databases and you do all these similarity searches, and that's how they're doing what they do. But then you have these rag models that kind of spin around on very interesting
stuff. The point being you can ask all sorts of interesting questions and get interesting answers, not always accurate with large language models, but at least you can start asking these very interesting questions and getting prosaic responses, and that's somewhat analogs analogous to what you folks are doing, and that you can ask these really detailed questions that traditional analytical systems would choke on for hours or days or
would just fall over sideways. And that is the exciting part. So I think it was sanchiev you mentioned at the end of the last segment. Yes you can find this thing. Now, Oh look at this actor, and now you can build a whole new set of queries around that actor. Well, who else, who else interacts with this person on a regular basis? How many times has this person come in? How many times do people like him come in? You can really start to explore what's happening once you get
that foothold. And as you've said, with these two different products, one dynamically comes up with new stuff, the other you have to tell it. But my point is that once you use this technology to find the actor, whether it's a bad or good actor, then you can ask amazingly rich questions, very detailed that other systems are never going to be able to handle.
That's a huge game changer. What do you think, Paige, Yeah, I think it the power of being able to ask I call it, you know, query the future, being able to query the past, the present, and the future simultaneously. So if I just discovered something interesting, I can then relate it to things that happened a year ago and things that haven't quite happened it you know that I kind of expect to happen within the next
week or two. So I can see say this is a financial broadstraer, and rather than you know, maybe I don't know where he's coming from. I can then look at everybody that he's related to over the last year and then maybe keep that up and as the time goes on, see who he
acts with next and figure out where he is. You know. You know what this is reminding me of is you think about all these TV shows and movies where they have the detectives, right, and they're all in the room in the conference room, and they have this big graph on the wall that has pictures and information and they're like looking at the graph and like, huh, like this guy's connected to that guy and all that stuff. But you have but you have it in an engine that can run on a Raspberry Pie.
That's how lean it is, and you can absorb absolutely gobsmacking amounts of data in order to do it. So it is like that investigative graph in the conference room at the Downtown office basically. But it's a technology which can pull in massive amounts from all sorts of sources. Right, you can set up COFFA topics on whatever the heck you want. You're not limited to the types of data or even the volumes of data. Right, You're not even
limited just to streaming data. I mean, we have a lot of people who pull in of files or whatever that they had setting around that might be related to the data that's streaming in, and you can build your graph with that data so that you have the context, you have everything that's around the data that's streaming in, and you can see how everything relates to everything.
It's really handy for like things like optimizing networks and finding problems there because you can analyze the CDM data, the data coming in from all of your different networks, all of these different IP addresses, maybe the data that you had sitting around already, and you can put that all together and find new information that you didn't even think about looking for before. Maybe. Yeah, you're able to build out a case on whatever it is that comes across the trans
And that's the key. That's kind of what I'm getting at here, is that even human beings how we perceive the world. We have a set of assumptions that we've made that we've learned. What if Mark Twain said, biases that set of prejudices you've you've developed over the first eighteen years of your life or something. You have a certain way of looking at the world, and that's what you accept as normal. But when and I'll give you another example,
and then maybe we'll bring Sanjeepe back into this. I've noticed this myself. When you hear a new word, and if you're fifty plus years old, it's not that often you hear a new word or a new term or something and you learn that we're like, oh wow, I never learned that word before, Like kwine like a quine, right, so you learn this wine is actually quine has actually been around. It's it's named after a guy, right. I looked at up that the guy. There's also another thing
named after the same guy. That's an application that generates its own source code. We're not, but we're named after the same Yeah. Well, at the point being you hear you learn this new word, and all of a sudden, you hear that word like seven times the next ten days. You're like, what is going on I'd never heard this word before, and now I've heard it seven times. Now you know, now your model is trained on that word, and so you recognize that as a pattern, you recognize
it as something noteworthy. That's true, right, So this is why that's actually let's go ahead out. It's actually the reticulate reticulated, reticulating mechanism in your brain that does that. It's networking. In other words, it associates all that sort of thing, probably streaming and articulating time. Yeah, in fact real time. So I want to actually talk about this. You know, how these tones come into existence and then in our mind we we've made
these these hard connections, but they may not always be correct. Like when we think streaming, we always think real time. And I let's say you are bringing this up and page. You mentioned that it may not be real time. It could be batch upload of documents for instance, or it may be real time. But if the real time comes into a cat topic and does not get consumed for a certain period of time, it's no longer real time. It's streaming data. So real time in streaming, I kind of
use analogis sleep. But there may not be a data that is at rest in a database is obviously not streaming, but the moment I started using it, then the data is in motion. Yeah right, I actually I think I did. I don't know. Years back, I did a thing called the Really Real Meaning of real time because it seems like everybody I talked to had a different definition. I asked a prominent analyst not too long ago, and he said, well, within fifteen minutes, and I was just like,
okay. I think at some point we had a customer that was like, I want it to check data every hour in real time, and I was just like, I'm not sure you we know what this word means. Yeah, because the types of real times, the two when people say real time, there's engineering real time and then there's a business real time. Engineering real time is actually only possible buyomachines, and it's microseconds or maybe maini seconds.
Humans humans are just too complicated to make such fast decisions. The business real time could be under fifteen minutes because it depends on business use case. Yeah, yeah, and I think it depends a lot. But I think the idea of being able to find a fraudster, find a cybersecurity risk, find that person and stop them before they have done something bad is important. I mean, you've gone from like the whole the whole idea of fraud detection
versus broad prevention is time. That's the only difference. It's like they're identical things. It's just one you're finding after the fact and the other one you're finding fast enough that you can do something about it. And I think that's the real power of real time is being able to go, hey, that I caught that advanced persistent threat who's been sitting here for a year the moment they tried to steal something. It's like that is that is really powerful,
that ability to go from I detected it. It's a little late, but I figured it out to I stopped it the moment they tried to do something wrong. And I think that's the real to me, that's a definition of real time is doing things fast enough that you can make a difference. Mm hmmm. Yeah, Well you're also motivating the analysts. I mean, this is one of my standard soapbox issues that morale is the most important characteristic of
any organization because when morale is high, good things happen. When morell's low, bad things happen. It doesn't matter how much money, you have, how many resources, what you're doing? None of that stuff matters and more ale, it's very difficult to quantify, to understand or to qualify. How do you know? I mean you can tell by looking at various metrics. But the point is this gets people excited. And look, let's face it,
cybersecurity is a very very challenging environment. I mean, we'll talk about it all the time, because you know, things are happening and if you have no mechanism for being able to get to them, it's very frustrating and people kind of go into road processes and just sort of give up, and that is not an acceptable answer. Well, let's pick this up. I have one more question from the studio audiens. We'll do for the podcast bonus
segment coming up in two seconds. We'll be right back. You're listening to Inside Analysis. All right, folks, Time for the podcast bonus segment here and a fascinating show. Hats off to all of our guests today. Paige Robertson at that dot look them up online, Sanjiv mohon Off, Sangmo and Alex Husky calling in from the back end fields way out there in North Dakota. That's why he's got some connectivity problems down again, he's way out there, no go to. It is really big and there's a lot going on
out there. But Page we had a great question strike it's the Moon. Great question from someone saying this sounds like it's an awesome solution for what's called user and entity behavior analytics, which again speaks to the fact that human beings as users, for example, can do all kinds of things. I mean, it's not like a list of five things I could do as a human. There's an infinite list of things I can do, and there's an infinite
array of behavioral patterns that could be identified. And traditional databases are not good at that kind of thing, certainly, not relational databases or object stores or all these different things. They hold stuff and that's their job, and they're designed to hold things and then deliver them upon demand. But even the analytical engines are are more relational in nature, and they're not graph in nature. And graph is fascinating because again it's got nodes and edges. The node could
be any entity. You could have a million nodes, you could have more than that. And beavis also are so variant, and I think that's what's so exciting about this streaming graph. From that thought is that it is so malleable. Nothing is unwieldy to it. What do you think about that? I think, yes, I heard, But I think when I talk about it's one of the challenges when I talk about categorical data, it's such an esoteric term that a lot of people are like, Yeah, what's the big
deal. What we're talking about is users, entities IP addresses. It's talking out about user and entity behavior. That's exactly what it is. And being able to analyze that in real time as the data appears right as opposed to excuse me having to transform that data into something numeric and then run an algorithm across the numeric data, which is huge, and it takes forever and then and then I can eventually figure it out, maybe a few hours later,
a few days later. The user and entity behavior that happens, now I can immediately analyze it. And that that is the power of streaming graph. Is the relationships the entities, the actions the people doing a thing. Is what a graph is designed to represent, is entities behaving. If that makes sense, I have alreadys heard it as iob as opposed to the acronym that the user used. But it's the same thing. It's that idea of being
to analyze categorical data without transforming it. First. Well, now that I could speak to some of that, maybe that if you can hear me, all right, But the behavior, the user behavior, all that stuff is also known as heuristics, I think. And they've been doing that, let's say, the world's fastest database in memory. What they do a lot of that. That's how they figure out how to customize ads for you based on what you lingered on with your microspike. I'm writing a technical book with Riley
on that with the air so aerospike exactly. So. I mean they're one of the first in memory of flash kind of outfits. But that whole idea of Okay, the guy is mousing over this, we're going to feed him this ad. Well, that applies also to hate so and so is trying to charge so and so's card for something that so and so doesn't you know there are four times in California using my credit card or something, and I know that's been done, but you guys are making it more accessible. I
think that's I think I don't know about accessibility. I think speed is really our secret secret sauce. I think it's because it's a young technology. It doesn't have a beautiful ter interface. It has one and you can actually see what's going on, which I think is I love being able to see what's happening, and that's marvelous. But I think the the real, the real key is speed. It's it's the same I mean Aerospike's advantages that they can
do speed at scale. That is what we can do. We can drew graph analysis at speed at scale, so that pretty much gives you a different kind of capability of event streaming and you know, you might stream the event into aerospike and then do something interesting with it there. It's like that this is the same kind of technology. I think progression that's happening a lot. Is it used to be. You know, it's might to you days to
to do a query. It might take you, you know, once a week to update your data, and then you're you're you're lucky if you're you know, you were really uh snazzy if your dashboards were only a day old. And it's like now if it's becoming it's like that took you ten seconds down. That's slow. I know, it's like there's there's very This is a very different world and we're trying to accomplish some things that that require that upper level, that that next gen capability, and I think streaming graph is
a really good example of that. We've just taken it to the next level. Yeah. Well, you know Aerospike a spike was handling streaming use cases in key value and then now divided graph. So I can see that how this technology is becoming the de facto standard. Yeah. I think I'm going to have to tell the Aerospike guys that they got to mention on the ship all right, So I was going to say one more thing is fine because it's speedy and it's smart. Is that under the hood at Talentier or how
do they tell us about bad guys? You know, starting a Chinese leaving the dog and the same for Hawaii, you know, like, yeah, I don't I can't talk all about most of them. But we're public with CrowdStrike. I mean, at this point, we've got one of the largest corporations in the world is embedding our software. Uh, we just got some funding from a US military branch. There's a lot of things going on that a lot of them are allowed to talk about. I mean pretty much all
I can tell you, folks. Look all these folks up online that dot. I love the story behind that dot, by the way, connecting all the dots. That dot is an IP address, that dot is a user. Look these guys up online that dot streaming graph of course, Sanjif Mohan of Sangmo and h Alex Husky thank you for your time for wanting to be in the show. Send me an email info at inside Analysis dot com. This does conclude our program for the day. What a fantastic show. We'll
talk you next time. Folks. Take care by miss something today, yesterday, last week. Check out our podcasts at WWWKCAA radio dot com. We leave no listener behind. Redlands. Ranch Market is a unique, full surface international grocery store that specializes in authentic food items from Mexico, India, and
from many Mediterranean and Asian countries, including popular items from the US. They offer fresh baked items from their in house bakery, housemade tortillas from their tortilla area, a delicious array of prepared Mexican foods, a terrific fresh food and juice bar, and a large selection of meats, seafoods and deli sandwiches, salads and halal meats. Their produce department is stocked full with fresh, local and hard to find international fruits and vegetables that you cannot find anywhere else.
Don't forget to step into the massive Beer Cave and experience the largest selection of domestic, artisan and imported beers in the IE. They can also cater your next event with one of the delicious takeout catering trays of food. Visit them at Redlands Ranch Market. That's Redlands Ranch Market dot com. Redlands Ranch Market a unique and fun shopping destination. Hi, I'm Lanniswardloe and I'm back on KCAA ten fifty am and Express one oh six point five FM every Tuesday at
eight pm. My show is Beyond Common Sense. It's Lanni Sense featuring me Lanni Swardlow, kcaa's resident gay, Jewish liberal, potsmoking, race mixing, left handed atheist, an evangelical, fundamentalist, Christian nationalist, worst nightmare with subjects that no one else will touch in quite the same way. Every Tuesday at apm on Express one oh six point five FM. The Legacy ten fifty
AM and live streaming on KCA Radio dot com. T Hebot Club's original pure powder RCO Superta comes from the only tree in the world that fungus does not grow on. As a result, it naturally has anti fu uncle, anti infection, anti viral, antibacterial, anti inflammation, and anti parasite properties. So the tea is great for healthy people because it helps build the immune system, and it can be truly miraculous for someone fighting a potentially life threatening disease
due to an infection, diabetes, or cancer. The tea is also organic and naturally caffeine free. A one pound package of tea is forty nine ninety five, which includes shipping. To order, please visit to Hebota club dot com. T hebo is spelled tea like tom a, h ee b like boy o. Then continue with the word t and then the word club. The complete website is to Hebot Club dot com or call us at eight one eight six one zero eight zero eight eight Monday through Saturday nine am to five
pm California time. That's eight one eight six one zero eight zero eight eight to Hebot club dot com. Need insurance help The Insurance Hour with Carl Susman is your answer from policy queries to coverage concerns. Carl an acclaimed expert in media guests, offers a join us for a show that simplifies insurance, making it understandable for everyone with sixty years of fascinating facts. This is the man from yesterday and back in time. We go to this time in nineteen ninety.
Michael Keaton, just coming off his biggest movie so far with Batman, has another movie coming out. In it, he plays a tenant from Hell. Look for Pacific Heights later this summer year in nineteen ninety. It also stars Melanie Griffith. He's not like anyone else. He does whatever he wants. He's changed the lots. I don't like this try this is a sick
individual, Miss Palmer. And from this time in nineteen fifty six, Walt Disney's Wednesday Night Disneyland series on ABCTV is preparing for another TV season, it's third, beginning in September. Disneyland is a top ratings show on ABCTV. Walt Disney's Disneyland Will You Wish? Show, honest Uff Like More Different, and from around this time in nineteen eighty five, a top movie Beverly Hills
cop starring Eddie Murphy, whose career is hot. The movie's theme song axel f will become a huge hit with more at man from yesterday dot Com. The Village Mud wants to remind pet owners of the importance of spang and neutering. Shelters overflow with unwanted pets. Spain and neutering helps prevent this and as many health benefits to that message. Courtesy of the Village Mud at sixty sixty five East Foothill Boulevard and Claremont for self serve peta washtubs and high end food
and treats for dogs and cats featuring natural and rock. Call the Village Mudd nine oh nine six two four three zero two zero and like them on Facebook. Do you want to learn and get answers to questions not addressed in the mass media, then you want to hear my show business Game Changers with me, Sarah Westall. I have conversations with thought leaders who have the courage to address off limit topics so you and your family have the tools to make the
right decisions. Join me Wednesdays at four pm on ten fifty AM and one oh six point five FM. Right here at CACAA, that station that leaves no listener behind. NBC News Radio, I'm Chris Garragio. The latest round of talks aimed at a ceasefire in Gaza stalled today. Correspondent Megan Fitzgerald has more on developments in Cairo. We heard from Prime Minister Benjamin Netanyahu today doubling
down, essentially saying that ending this war is off the table. As we also know that he said that deal or no deal, that planned ground incursion into Rafa will be happening. Both Israel and Hamas trade a blame over the impass, which came despite word that negotiators were close to a breakthrough. CIA Director Bill Burns is reportedly traveling to cutter as the US continues to push for
progress. Negotiators were said to have been working on a new framework involving the release of hostages kidnapped from Israel in exchange for a pause in the fighting. UCLA is resuming regular operations starting tomorrow. The university announced today faculty and staff will be able to resume in person classes, but did leave the option to conduct them online until May tenth. Officials also said the law school would still
hold exams as regularly scheduled. The announcement recommended people avoid a roy squad given the cleanup efforts that are underway there. The university will maintain a police presence on campus. South Dakota Governor Christy Nome continues to face criticism over details of her upcoming memoir. Appearing on CBS's Face the Nation, Nome faced a grilling
over controversial claims that include having met North Korean leader Kim Jong un. As soon as it was brought to my attention, we went forward and have made some edits. So I'm glad that this book is being released in a couple of days and that those edits will be in place and that people will will have the updated version. The Republican governor refused to say directly whether she met
with the dictator. She also defended a passage in the book where she details how she shot and killed her dog, calling it a dangerous animal that was killing livestock. Nome was considered to be on Donald Trump's shortlist for vice president. I'm Chris Karragio, NBC News Radio, NBC News on CACAA Lowel sponsored by Teamsters Local nineteen thirty two, protecting the Future of Working Families Teamsters nineteen thirty two dot org. You're listening to an encore presentation of this program k
C A A The Inland Talk Express. Thank you for tuning in for this edition of Justice Watch with a Turn