InetSoft Webinar: High Speed Parallel Processing to Connect to Data

This is the continuation of the transcript of the DM Radio show "Avoiding Bottlenecks and Hurdles in Data Delivery." InetSoft's Principal Technologist, Byron Igoe, joined industry analysts and other data management software vendors for a discussion about current issues and solutions for information management.

Eric Kavanagh: Yes, may be let’s bring Ian back in. We haven’t heard from him in a while. Ian, what do you think about that?

Ian Pestell: Yeah, I mean, definitely the area we are seeing that like a local audience around the appliance area, the people putting in very high speed parallel processing to connect to data very, very quickly, but I think the performance that you want to output certain data fast, and you want to put other data on systems that needs to process slower than others in terms of their size.

And again the ability to bring all that data together and place things that you have to repair the data, I think what is going to emerge out of the next year in terms of the business, the purpose is where they have designed but targeting the user where they came. Where did they come from?

Richard Walker: Yeah. it's a good point, and I am going to another hardware piece. Last week I was at Teradata Partners Conference, and I saw an appliance there from Teradata that has solid state drives in it. So we are all kind of waiting around to see if solid state was going to kind of replace memory. What about flash memory? Is that coming on in the enterprise systems, that kind of thing?

#1 Ranking: Read how InetSoft was rated #1 for user adoption in G2's user survey-based index Read More

So it ties into what Ian was just saying about you have to make decisions about where to put data if you have some faster systems for storage than that kind of slower shelf-system, you want to put your hotter data up there. So I think for a lot of people, bottlenecks in terms of moving data can involve a kind of data temperature assessment how hot is this data? How often is it accessed and putting in the faster stuff. And I think in the future for the faster stuff, it's going to be solid state drives.

Eric Kavanagh: Yeah, that’s a good point, and lots of different changes in hardware and software, and it's making us all look at this whole situation differently. I get back to one of my soap box issues these days, and that’s the whole concept of information architecture. And if you have that strategic view of your information architecture, that’s when you can start seeing which systems are slower, to Richard’s point there in the last minute, which are faster and where you need to move things and all that kind of fun stuff, right Phil?

Philip Russom: Yeah, absolutely, and that ties into something David Inbar was saying earlier. You asked him what kind of mistakes is he is seeing people make with bottlenecks, and David was talking about, a lot of times we have these major bottleneck, and the solution is just to throw more hardware at it, more server hardware. And it's tempting because server hardware is cheap, and now another technology invention, the multicore CPUs, and those things are fast and they are built to speed up processing, right?

But there is a limit to throwing on hardware the problem. I think a lot of people could solve a lot of their bottlenecks if they just revisit their architecture like you are talking about, Eric. And number one is to have an architecture. I am telling you, how many people are perplexed by saying that data integration is not an architecture. They look at me blankly, like what do you mean? It superimposes on architecture, is that what you mean? No. You have to take charge of the solution you design and make sure it is performing.

view demo icon
View a 2-minute demonstration of InetSoft's easy, agile, and robust BI software.

I reported earlier this year on next generation data integration. For a lot of companies, the next generation is to rewrite the old generation because so many people wrote their data integration solution with serial programming. They need to go back and rewrite it to make it fully leverage parallel processing in the design, you know what I am saying. That’s actually an architectural issue and that can help get rid of bottlenecks.

Eric Kavanagh: That’s right and in fact, you reminded that multicore is something that we definitely need to talk about next year or so we will figure out where we could fit that in. But folks time flies when you’re having fun we have burned right through an hour and change. A big thank you to all of our guests including my good friend Philip Russom from TDWI taking time out of his day and thanks to all of you out there for all your tweets and for tuning into the show today we do appreciate it. We will catch up with you next week, folks. Intil then take care. You have been listening to DM Radio.

Previous: Cube of Data in Server Memory