MyPage is a personalized page based on your interests.The page is customized to help you to find content that matters you the most.


I'm not curious

How UK data solutions developer Systems Mechanics uses HP Vertica for BI, streaming and data analysis

Published on 23 July 14
957
0
0

Three years ago, Systems Mechanics Limited used relational databases to assemble and analyze some 20 different data sources in near real-time. But most relational database appliances used 1980s technical approaches, and the ability to connect more data and manage more events capped off. The runway for their business expansion just ended.


So Systems Mechanics looked for a platform that scales well and provides real-time data analysis, too. At the volumes and price they needed, HP Vertica has since scaled without limit ... an endless runway.


Listen to the podcast. Find it on iTunes. Read a full transcript or download a copy.


To learn more about how Systems Mechanics improved how their products best deliver business intelligence (BI), analytics streaming, and data analysis, BriefingsDirect spoke with Andy Stubley, Vice President of Sales and Marketing at Systems Mechanics, based in London. The discussion, at the HP Discover conference in Barcelona, is moderated by me, Dana Gardner, Principal Analyst at Interarbor Solutions.


Here are some excerpts:


Gardner: You've been doing a lot with data analysis at Systems Mechanics, and monetizing that in some very compelling ways.


Stubley: Yes, indeed. System Mechanics is principally a consultancy and a software developer. Weâve been working in the telco space for the last 10-15 years. We also have a history in retail and financial services.


How UK data solutions developer Systems Mechanics uses HP Vertica for BI, streaming and data analysis - Image 1

Stubley

The focus we've had recently and the products weâve developed into our Zen family are based on big data, particularly in telcos, as they evolve from principally old analog conversations into devices where people have smartphone applications -- and data becomes ever more important.


All that data and all those people connected to the network cause a lot more events that need to be managed, and that data is both a cost to the business and an opportunity to optimize the business. So we have a cost reduction we apply and a revenue upside we apply as well.


Quick example


Gardner: Whatâs a typical way telcos use Zen, and that analysis?


Stubley: Letâs take a scenario where youâre looking in network and you canât make a phone call. Two major systems are catching that information. One is a fault-management system thatâs telling you there is a fault on the network and it reports that back to the telecom itself.


The second one is the performance management system. That doesnât specify faults basically, but it tells you if youâre having things like thresholds being affected, which may have an impact on performance every time. Either of those can have an impact on your customer, and from a customerâs perspective, you might also be having a problem with the network that isnât reported by either of the systems.


Weâre finding that social media is getting a bigger play in this space. Why is that? Now, particular the younger populations with consumer-based telcos, mobile telcos particularly, if they canât get a signal or they canât make a phone call, they get onto social media and they are trashing the brand.


Theyâre making noise. A trend is combining fault management and performance management, which are logical partners with social media. All of a sudden, rather than having a couple of systems, you have three.


In our world, we can put 25 or 30 different data sources on to a single Zen platform. In fact, there is no theoretical limit to the number we could, but 20 to 30 is quite typical now. That enables us to manage all the different network elements, different types of mobile technologies, LTE, 3G, and 2G. It could be Ericsson, Nokia, Huawei, ZTE, or Alcatel-Lucent. There is an amazing range of equipment, all currently managed through separate entities. Weâre offering a platform to pull it all together in one unit.


The other way I tend to look at it is that weâre trying to turn the telcos into how you might view a human. We take the humans as the best decision-making platforms in the world and we probably still could claim that. As humans, we have conscious and unconscious processes running. We donât think about breathing or pumping our blood around our system, but itâs happening all the time.

We use a solution with visualization, because in the world of big data, you canât understand data in numbers.


We have senses that are pulling in massive amount of information from the outside world. Youâre listening to me now. Youâre probably doing a bunch of other things while you are tapping away on a table as well. Theyâre getting senses of information there and you are seeing, and hearing, and feeling, and touching, and tasting.


Those all contain information thatâs coming into the body, but most of the activity is subconscious. In the world of big data, this is the Zen goal, and what weâre delivering in a number of places is to make as many actions as possible in a telco environment, as in a network environment, come to that automatic, subconscious state.


Suppose I have a problem on a network. I relate it back to the people who need to know, but I donât require human intervention. Weâre looking a position where the human intervention is looking at patterns in that information to decide what they can do intellectually to make the business better.


That probably speaks to another point here. We use a solution with visualization, because in the world of big data, you canât understand data in numbers. Your human brain isnât capable of processing enough, but it is capable of identifying patterns of pictures, and thatâs where we go with our visualization technology.


Gather and use data


We have a customer who is one of the largest telcos in EMEA. Theyâre basically taking in 90,000 alarms from the network a day, and thatâs their subsidiary companies, all into one environment. But 90,000 alarms needing manual intervention is a very big number.

Using the Zen technology, weâve been able to reduce that to 10,000 alarms. Weâve effectively taken 90 percent of the manual processing out of that environment. Now, 10,000 is still a lot of alarms to deal with, but itâs a lot less frightening than 90,000, and thatâs a real impact in human terms.


Gardner: Now that we understand what you do, letâs get into how you do it. Whatâs beneath the covers in your Zen system that allows you to confidently say you can take any volume of data you want?

If we need more processing power, we can add more services to scale transparently. That enables us to get any amount of data, which we can then process.


Stubley: Fundamentally, that comes down to the architecture we built for Zen. The first element is our data-integration layer. We have a technology that we developed over the last 10 years specifically to capture data in telco networks. Itâs real-time and rugged and it can deal with any volume. That enables us to take anything from the network and push it into our real-time database, which is HPâs Vertica solution, part of the HP HAVEn family.


Vertica analysis is to basically record any amount of data in real time and scale automatically on the HP hardware platform we also use. If we need more processing power, we can add more services to scale transparently. That enables us to get any amount of data, which we can then process.


We have two processing layers. Referring to our earlier discussion about conscious and subconscious activity, our conscious activity is visualizing that data, and thatâs done with Tableau.


We have a number of Tableau reports and dashboards with each of our product solutions. That enables us to envision whatâs happening and allows the organization, the guys running the network, and the guys looking at different elements in the data to make their own decisions and identify what they might do.


We also have a streaming analytics engine that listens to the data as it comes into the system before it goes to Vertica. If we spot the patterns weâve identified earlier subconsciously, weâll then act on that data, which may be reducing an alarm count. It may be "actioning" something.


It may be sending someone an email. It may be creating a trouble ticket on a different system. Those all happen transparently and automatically. Itâs four layers simplifying the solution: data capture, data integration, visualization, and automatic analytics.


Developing high value


Gardner: And when you have the confidence to scale your underlying architecture and infrastructure, when you are able to visualize and develop high value to a vertical industry like a telco, this allows you to then expand into more lines of business in terms of products and services and also expand into move vertical. Where have you taken this in terms of the Zen family and then where do you take this now in terms of your market opportunity?


Stubley: We focus on mobile telcos. Thatâs our heritage. We can take any data source from a telco, but we can actually take any data source from anywhere, in any platform and any company. That ranges from binary to HTML. You name it, and if youâve got data, we could load it.


That means we can build our processing accordingly. What we do is position what we call solution packs, and a solution pack is a connector to the outside world, to the network, and it grabs the data. Weâve got an element of data modeling there, so we can load the data into Vertica. Then, we have already built reports in Tableau that allows us to interrogate automatically. Thatâs at a component level.


Once you go to a number of components, we can then look horizontally across those different items and look at the behaviors that interact with each other. If you are looking at pure telco terms, we would be looking at different network devices, the end-to-end performance of the network, but the same would apply to a fraud scenario or could apply to someone who is running cable TV.

The very highest level is finding what problem youâre going to solve and then using the data to solve it.


So multi-play players are interesting because they want to monitor whatâs happening with TV as well and that will fit in exactly in the same category. Realistically, anybody with high-volume, real-time data can take benefit from Vertica.


Another interesting play in this scenario is social gaming and online advertising. They all have similar data characteristics, very high volume and fixed data that needs to be analyzed and processed automatically.


Why Vertica?


Gardner: How long have you been using Vertica, and what is it that drove you to using it vis-Ã-vis alternatives?


Stubley: As far as the Zen family goes, we have used other technologies in the past, other relational databases, but weâve used Vertica now for more than two-and-a-half years. We were looking for a platform that can scale and would give us real-time data. At the volumes we were looking at nothing could compete with Vertica at a sensible price. You can build yourself any solid solution with enough money, but we havenât got too many customers who are prepared to make that investment.


So Vertica fits in with the technology of the 21st century. A lot of the relational database appliances are using 1980 thought processes. Whatâs happened with processing in the last few years is that nobody shares memory anymore, and our environment requires a non-shared memory solution. Vertica has been built on that basis. It was scaled without limit.


One of the areas weâre looking at that I mentioned earlier was social media. Social media is a very natural play for Hadoop, and Hadoop is clearly a very cost-effective platform for vast volumes of data at real-time data load, but very slow to analyze.

So the combination with a high-volume, low-cost platform for the bulk of data and a very high performing real-time analytics engine is very compelling. The challenge is going to be moving the data between the two environments. That isnât going to go away. Thatâs not simple, and there is a number of approaches. HP Vertica is taking some.


There is Flex Zone, and there are any number of other players in that space. The reality is that you probably reach an environment where people are parallel loading the Hadoop and the Vertica. Thatâs what we probably plan to do. That gives you much more resilience. So for a lot of the data weâre putting into our system, weâre actually planning to put the raw data files into Hadoop, so we can reload them as necessary to improve the resilience of the overall system, too.


Listen to the podcast. Find it on iTunes. Read a full transcript or download a copy. Sponsor: HP.


You may also be interested in:













































Three years ago, Systems Mechanics Limited used relational databases to assemble and analyze some 20 different data sources in near real-time. But most relational database appliances used 1980s technical approaches, and the ability to connect more data and manage more events capped off. The runway for their business expansion just ended.

So Systems Mechanics looked for a platform that scales well and provides real-time data analysis, too. At the volumes and price they needed, HP Vertica has since scaled without limit ... an endless runway.

Listen to the podcast. Find it on iTunes. Read a full transcript or download a copy.

To learn more about how Systems Mechanics improved how their products best deliver business intelligence (BI), analytics streaming, and data analysis, BriefingsDirect spoke with Andy Stubley, Vice President of Sales and Marketing at Systems Mechanics, based in London. The discussion, at the HP Discover conference in Barcelona, is moderated by me, Dana Gardner, Principal Analyst at Interarbor Solutions.

Here are some excerpts:

Gardner: You've been doing a lot with data analysis at Systems Mechanics, and monetizing that in some very compelling ways.

Stubley: Yes, indeed. System Mechanics is principally a consultancy and a software developer. Weâve been working in the telco space for the last 10-15 years. We also have a history in retail and financial services.

How UK data solutions developer Systems Mechanics uses HP Vertica for BI, streaming and data analysis - Image 1


Stubley

The focus we've had recently and the products weâve developed into our Zen family are based on big data, particularly in telcos, as they evolve from principally old analog conversations into devices where people have smartphone applications -- and data becomes ever more important.

All that data and all those people connected to the network cause a lot more events that need to be managed, and that data is both a cost to the business and an opportunity to optimize the business. So we have a cost reduction we apply and a revenue upside we apply as well.

Quick example

Gardner: Whatâs a typical way telcos use Zen, and that analysis?

Stubley: Letâs take a scenario where youâre looking in network and you canât make a phone call. Two major systems are catching that information. One is a fault-management system thatâs telling you there is a fault on the network and it reports that back to the telecom itself.

The second one is the performance management system. That doesnât specify faults basically, but it tells you if youâre having things like thresholds being affected, which may have an impact on performance every time. Either of those can have an impact on your customer, and from a customerâs perspective, you might also be having a problem with the network that isnât reported by either of the systems.

Weâre finding that social media is getting a bigger play in this space. Why is that? Now, particular the younger populations with consumer-based telcos, mobile telcos particularly, if they canât get a signal or they canât make a phone call, they get onto social media and they are trashing the brand.

Theyâre making noise. A trend is combining fault management and performance management, which are logical partners with social media. All of a sudden, rather than having a couple of systems, you have three.

In our world, we can put 25 or 30 different data sources on to a single Zen platform. In fact, there is no theoretical limit to the number we could, but 20 to 30 is quite typical now. That enables us to manage all the different network elements, different types of mobile technologies, LTE, 3G, and 2G. It could be Ericsson, Nokia, Huawei, ZTE, or Alcatel-Lucent. There is an amazing range of equipment, all currently managed through separate entities. Weâre offering a platform to pull it all together in one unit.

The other way I tend to look at it is that weâre trying to turn the telcos into how you might view a human. We take the humans as the best decision-making platforms in the world and we probably still could claim that. As humans, we have conscious and unconscious processes running. We donât think about breathing or pumping our blood around our system, but itâs happening all the time.

We use a solution with visualization, because in the world of big data, you canât understand data in numbers.

We have senses that are pulling in massive amount of information from the outside world. Youâre listening to me now. Youâre probably doing a bunch of other things while you are tapping away on a table as well. Theyâre getting senses of information there and you are seeing, and hearing, and feeling, and touching, and tasting.

Those all contain information thatâs coming into the body, but most of the activity is subconscious. In the world of big data, this is the Zen goal, and what weâre delivering in a number of places is to make as many actions as possible in a telco environment, as in a network environment, come to that automatic, subconscious state.

Suppose I have a problem on a network. I relate it back to the people who need to know, but I donât require human intervention. Weâre looking a position where the human intervention is looking at patterns in that information to decide what they can do intellectually to make the business better.

That probably speaks to another point here. We use a solution with visualization, because in the world of big data, you canât understand data in numbers. Your human brain isnât capable of processing enough, but it is capable of identifying patterns of pictures, and thatâs where we go with our visualization technology.

Gather and use data

We have a customer who is one of the largest telcos in EMEA. Theyâre basically taking in 90,000 alarms from the network a day, and thatâs their subsidiary companies, all into one environment. But 90,000 alarms needing manual intervention is a very big number.

Using the Zen technology, weâve been able to reduce that to 10,000 alarms. Weâve effectively taken 90 percent of the manual processing out of that environment. Now, 10,000 is still a lot of alarms to deal with, but itâs a lot less frightening than 90,000, and thatâs a real impact in human terms.

Gardner: Now that we understand what you do, letâs get into how you do it. Whatâs beneath the covers in your Zen system that allows you to confidently say you can take any volume of data you want?

If we need more processing power, we can add more services to scale transparently. That enables us to get any amount of data, which we can then process.

Stubley: Fundamentally, that comes down to the architecture we built for Zen. The first element is our data-integration layer. We have a technology that we developed over the last 10 years specifically to capture data in telco networks. Itâs real-time and rugged and it can deal with any volume. That enables us to take anything from the network and push it into our real-time database, which is HPâs Vertica solution, part of the HP HAVEn family.

Vertica analysis is to basically record any amount of data in real time and scale automatically on the HP hardware platform we also use. If we need more processing power, we can add more services to scale transparently. That enables us to get any amount of data, which we can then process.

We have two processing layers. Referring to our earlier discussion about conscious and subconscious activity, our conscious activity is visualizing that data, and thatâs done with Tableau.

We have a number of Tableau reports and dashboards with each of our product solutions. That enables us to envision whatâs happening and allows the organization, the guys running the network, and the guys looking at different elements in the data to make their own decisions and identify what they might do.

We also have a streaming analytics engine that listens to the data as it comes into the system before it goes to Vertica. If we spot the patterns weâve identified earlier subconsciously, weâll then act on that data, which may be reducing an alarm count. It may be "actioning" something.

It may be sending someone an email. It may be creating a trouble ticket on a different system. Those all happen transparently and automatically. Itâs four layers simplifying the solution: data capture, data integration, visualization, and automatic analytics.

Developing high value

Gardner: And when you have the confidence to scale your underlying architecture and infrastructure, when you are able to visualize and develop high value to a vertical industry like a telco, this allows you to then expand into more lines of business in terms of products and services and also expand into move vertical. Where have you taken this in terms of the Zen family and then where do you take this now in terms of your market opportunity?

Stubley: We focus on mobile telcos. Thatâs our heritage. We can take any data source from a telco, but we can actually take any data source from anywhere, in any platform and any company. That ranges from binary to HTML. You name it, and if youâve got data, we could load it.

That means we can build our processing accordingly. What we do is position what we call solution packs, and a solution pack is a connector to the outside world, to the network, and it grabs the data. Weâve got an element of data modeling there, so we can load the data into Vertica. Then, we have already built reports in Tableau that allows us to interrogate automatically. Thatâs at a component level.

Once you go to a number of components, we can then look horizontally across those different items and look at the behaviors that interact with each other. If you are looking at pure telco terms, we would be looking at different network devices, the end-to-end performance of the network, but the same would apply to a fraud scenario or could apply to someone who is running cable TV.

The very highest level is finding what problem youâre going to solve and then using the data to solve it.

So multi-play players are interesting because they want to monitor whatâs happening with TV as well and that will fit in exactly in the same category. Realistically, anybody with high-volume, real-time data can take benefit from Vertica.

Another interesting play in this scenario is social gaming and online advertising. They all have similar data characteristics, very high volume and fixed data that needs to be analyzed and processed automatically.

Why Vertica?

Gardner: How long have you been using Vertica, and what is it that drove you to using it vis-Ã-vis alternatives?

Stubley: As far as the Zen family goes, we have used other technologies in the past, other relational databases, but weâve used Vertica now for more than two-and-a-half years. We were looking for a platform that can scale and would give us real-time data. At the volumes we were looking at nothing could compete with Vertica at a sensible price. You can build yourself any solid solution with enough money, but we havenât got too many customers who are prepared to make that investment.

So Vertica fits in with the technology of the 21st century. A lot of the relational database appliances are using 1980 thought processes. Whatâs happened with processing in the last few years is that nobody shares memory anymore, and our environment requires a non-shared memory solution. Vertica has been built on that basis. It was scaled without limit.

One of the areas weâre looking at that I mentioned earlier was social media. Social media is a very natural play for Hadoop, and Hadoop is clearly a very cost-effective platform for vast volumes of data at real-time data load, but very slow to analyze.

So the combination with a high-volume, low-cost platform for the bulk of data and a very high performing real-time analytics engine is very compelling. The challenge is going to be moving the data between the two environments. That isnât going to go away. Thatâs not simple, and there is a number of approaches. HP Vertica is taking some.

There is Flex Zone, and there are any number of other players in that space. The reality is that you probably reach an environment where people are parallel loading the Hadoop and the Vertica. Thatâs what we probably plan to do. That gives you much more resilience. So for a lot of the data weâre putting into our system, weâre actually planning to put the raw data files into Hadoop, so we can reload them as necessary to improve the resilience of the overall system, too.

Listen to the podcast. Find it on iTunes. Read a full transcript or download a copy. Sponsor: HP.

You may also be interested in:



This blog is listed under Enterprise Applications and Data & Information Management Community

Post a Comment

Please notify me the replies via email.

Important:
  • We hope the conversations that take place on MyTechLogy.com will be constructive and thought-provoking.
  • To ensure the quality of the discussion, our moderators may review/edit the comments for clarity and relevance.
  • Comments that are promotional, mean-spirited, or off-topic may be deleted per the moderators' judgment.
You may also be interested in
Awards & Accolades for MyTechLogy
Winner of
REDHERRING
Top 100 Asia
Finalist at SiTF Awards 2014 under the category Best Social & Community Product
Finalist at HR Vendor of the Year 2015 Awards under the category Best Learning Management System
Finalist at HR Vendor of the Year 2015 Awards under the category Best Talent Management Software
Hidden Image Url

Back to Top