i would guess that much of the more incendiary stuff jimrutt has scribbled, as someone offered in the 'niggardly' example, would have to do with race. jim and i have regularly traded barbs on the subject, but i have always found him to have the integrity of a dedicated interlocutor.
now as i go back to the well, the context of many of our discussions will be destroyed. people will read a conversation in which one of my personnas is yelling at a non-existant wall. it makes us both look loonier than we are. so what's interesting is what speculation people might draw from these disconnected comments. i would have not put it past myself (although i haven't engaged jim on the well or anywhere else for a while) to have told him that any of his admittedly provocative comments 'was the most idiotic and reactionary racist tripe i've ever heard'. one can only guess what people will guess.
when all is said and done, i would make it my duty to see that he's not slammed in that regard. i've found him a fine foil for instructive purposes, and i would hate to see him catch heat for being provocative. i have little doubt that plenty folks would like to give him heat he doesn't deserve.
my advice to myself has been to write in lowercase to distinguish my online communications from those which i type up for other purposes. as well, i have anonymized myself in various ways. the chill factor for me is already built in, yet i understand that my speech is subversive, so that gives me courage of a different sort. i hope that it can live on in this context, and i lament the loss of my virtual sparring partner's words.
faster isn't always better. palm wins based on usability. it's simple, yet expandable. that's just good design - a design with the end user in mind, not the settopbox development market.
I can save you a lot of time and trouble offline. yes i sell the stuff, but i have also done implementations going back several years. i have seen what folks are doing all around the country, and i understand the strengths and weaknesses of the major products.
it's very exasperating for me to see people thrashing around on this issue, especially when they start talking up the minor vendors.
if you absolutely are bent on not spending 25grand for your basic essbase server license and want to build your own stuff, then i really can't help you. that's just beneath the radar. if you want to go for the cheap stuff, my best recommendation would be tm/1.
but if you want a first class system. mail me or discuss it here. mbowen@panix.com
the market is saturated with front-end tool vendors, and all of the ones who are making any money have already ported to the essbase api and/or the microsoft api. the spec got completed but by that time the market-share game was already over. therefore, there is no interest in doing anything new and untested.
some folks have complained about the need for a full objecty kind of api for essbase. hyperion is scratching its head about api stuff. the confusion is easy to understand: why do more when there are so many successful vendors out there using the current api?
if you don't mind crappy performance and like doing everything by hand, then holos might be a bargain. i don't mean to flame, but if you knew how many customers buy holos based on its elevated rhetoric and replaced it as soon as they could afford to...
my sources say that seagate doesn't even do its own olap on holos. and the holos front end tools allow you to use essbase as a back end through the essbase api. explain that.
on the other hand, if you'd rather program a 4gl than use an api, i can sympathize with a choice of holos.
by the way, if you want java beans, try painted word. if you want java component based web, try alphablox.
ibm santa teresa has done something like this. they actually have the ability to select out a 'square' of data within a fact table. this square has it's own security, logging and it is optimized for incremental update and read.
by 'square' i mean only the area defined by the some contiguous columns intersecting contiguous rows WITHOUT the rest of the record. it's basically cell level retrieval, with security. --
if tm1 went open source, that would be an amazing thing.
powerplay is essentially a front-end tool. it makes little cache files called 'powercubes' which reside on the client after you vaccuum up a big query from a source database. the main presumption of this design is that you intend to navigate around in a tiny subset of multidimensional space. the second presumption of this is that your server can't handle real-time queries. some people call that olap. it's not really.
i have never met anyone under the age of 40 who had the patience to work with SAS in any olap or D/W project. on the other hand, people i meet get to understand the genius of essbase.
sas is a brilliant company for leveraging legacy apps. if you did a masters anywhere doing statistical research or clinical trials, you've tasted sas. lots of folks learn it in postgrad studies. but using sas in today's market is a little like mba's using their same hp calculators to run the company finances. it just doesn't make sense except for the fact that you are extremely comfortable with it. but what can you say to people who are in love?
hyperion as a company has a problem in reaching slashdot-type folks. that's probably because it came from the finapps space. not a lot of bsd in corporate finance. be that as it may, don't let that convince you to try and build olap from scratch. i've met with very bright guys from citibank who built their own. (believe me it was a seriously hostile, new york style meeting). but when i explained the internals of essbase to them, they came around. i mean the stuff is patented, and the inventor understands everything about sparse matrix math and all that eigenstuff.
there are a significant number of computational problems that solid olap technology addresses. the fact that ibm, acknowledged master of sql optimization, has opted to oem the hyperion technology rather than build their own is all the proof anyone should need.
there are two kinds of voodoo. chicken dancing voodoo and patent office voodoo. if everybody could do this, there would be 10 olap servers on the market. as it stands, there are about 4 worth mentioning.
olap engines also figure out things like, practically speaking, how would i tune the profitability of international container shipping using activity based costing methods?
data warehousing and data marting have completely different dynamics. data warehousing is all about storing everything in a consistant manner such that it can be replicated for any reason without losing its contextual meaning.
data marts are specifically designed to answer specific questions about some process of the business.
datamarts built with relational technology are quick and easy to design but offer limited functionality, performance and flexibility. datamarts built with true olap engines do all that work better, but they take a bit more brains to setup. (mostly the kind of brains that tell you relational technology doesn't cut it.)
this is the closest thing to an intelligent post i have bothered to read. the answer, my dears, is essbase. http://www.hyperion.com. essbase is the world's fastest and most capable multidimensional database engine. it has been ported to every unix and the linux port is in beta/qa.
essbase has had an open api (both in c, and vb) for years, and dozens of partners who have written front-ends like cognos, business objects, brio and a host of others which plug and play to the database.
olap is not olap in the way that relational is relational. the fact of the matter is that arbor software recruited *the* dr. codd to sum up the weaknesses of relational database for realtime access as in the way relational dweeb dba's have been attempting for years in datawarehousing. only two vendors have implemented the full scope of codd's rules. they are arbor (now merged with hyperion) who own essbase, and oracle who purchased iri's express multidimensional engine. microsoft, on the late freight, bought an israeli joint called panorama, scrounged up some of the engineers from oracle and are now trying to undercut the market with a deally called microsoft olap services now bundled with sql server 7.0. (codenamed 'plato'). for those of you with enough sense to resist microsoft, i needn't comment further on this nt only wizard-based thingy.
oracle has decommitted and begun to defund the express division. all of their marketing people have been canned. they are hoping that people will continue to remain clueless and assume that extentions to sql will handle everything 'olap'. of course unless you pay attention to the full spec of things codd said many years ago, then you are likely to buy all the marketing crap and actually believe that relational databases are going to do everything.
the momentum remains with dweebBAs with big iron who hack their way through datawarehouse projects without real server-based olap. to them it's all 'slice & dice' and other PHB front-end sillyness. in the meantime, people who know are scarce. the rest of us are making buttloads of cash in the consulting biz, or are running the market-leading hyperion.
the people who really know this stuff live at centrobe, beacon analytics, mcanalyst, data into action, pinnacle solutions, answerspace, navigator systems, ranzal & associates, symmetry and eds.
bottom line is this. there is one company who is going to crunch the uncrunchable, and i'm talking about weblogs from yahoo, and that will be hyperion. it won't be done with relational crap.
Slashdot Mirror
Definitely has my vote. It's a basic legend - half the stories about hacking begin with 2600.
wow. that's what i got paid in 1984 for a xerox internship.
i would guess that much of the more incendiary stuff jimrutt has scribbled, as someone offered in the 'niggardly' example, would have to do with race. jim and i have regularly traded barbs on the subject, but i have always found him to have the integrity of a dedicated interlocutor.
now as i go back to the well, the context of many of our discussions will be destroyed. people will read a conversation in which one of my personnas is yelling at a non-existant wall. it makes us both look loonier than we are. so what's interesting is what speculation people might draw from these disconnected comments. i would have not put it past myself (although i haven't engaged jim on the well or anywhere else for a while) to have told him that any of his admittedly provocative comments 'was the most idiotic and reactionary racist tripe i've ever heard'. one can only guess what people will guess.
when all is said and done, i would make it my duty to see that he's not slammed in that regard. i've found him a fine foil for instructive purposes, and i would hate to see him catch heat for being provocative. i have little doubt that plenty folks would like to give him heat he doesn't deserve.
my advice to myself has been to write in lowercase to distinguish my online communications from those which i type up for other purposes. as well, i have anonymized myself in various ways. the chill factor for me is already built in, yet i understand that my speech is subversive, so that gives me courage of a different sort. i hope that it can live on in this context, and i lament the loss of my virtual sparring partner's words.
faster isn't always better. palm wins based on usability. it's simple, yet expandable. that's just good design - a design with the end user in mind, not the settopbox development market.
I can save you a lot of time and trouble offline. yes i sell the stuff, but i have also done implementations going back several years. i have seen what folks are doing all around the country, and i understand the strengths and weaknesses of the major products.
it's very exasperating for me to see people thrashing around on this issue, especially when they start talking up the minor vendors.
if you absolutely are bent on not spending 25grand for your basic essbase server license and want to build your own stuff, then i really can't help you. that's just beneath the radar. if you want to go for the cheap stuff, my best recommendation would be tm/1.
but if you want a first class system. mail me or discuss it here. mbowen@panix.com
say "doornail".
the market is saturated with front-end tool vendors, and all of the ones who are making any money have already ported to the essbase api and/or the microsoft api. the spec got completed but by that time the market-share game was already over. therefore, there is no interest in doing anything new and untested.
some folks have complained about the need for a full objecty kind of api for essbase. hyperion is scratching its head about api stuff. the confusion is easy to understand: why do more when there are so many successful vendors out there using the current api?
my sources say that seagate doesn't even do its own olap on holos. and the holos front end tools allow you to use essbase as a back end through the essbase api. explain that.
on the other hand, if you'd rather program a 4gl than use an api, i can sympathize with a choice of holos.
by the way, if you want java beans, try painted word. if you want java component based web, try alphablox.
ibm santa teresa has done something like this. they actually have the ability to select out a 'square' of data within a fact table. this square has it's own security, logging and it is optimized for incremental update and read.
by 'square' i mean only the area defined by the some contiguous columns intersecting contiguous rows WITHOUT the rest of the record. it's basically cell level retrieval, with security.
--
if tm1 went open source, that would be an amazing thing.
this is a little bit dated but a decent enough overview.
powerplay is essentially a front-end tool. it makes little cache files called 'powercubes' which reside on the client after you vaccuum up a big query from a source database. the main presumption of this design is that you intend to navigate around in a tiny subset of multidimensional space. the second presumption of this is that your server can't handle real-time queries. some people call that olap. it's not really.
sas is a brilliant company for leveraging legacy apps. if you did a masters anywhere doing statistical research or clinical trials, you've tasted sas. lots of folks learn it in postgrad studies. but using sas in today's market is a little like mba's using their same hp calculators to run the company finances. it just doesn't make sense except for the fact that you are extremely comfortable with it. but what can you say to people who are in love?
hyperion as a company has a problem in reaching slashdot-type folks. that's probably because it came from the finapps space. not a lot of bsd in corporate finance. be that as it may, don't let that convince you to try and build olap from scratch. i've met with very bright guys from citibank who built their own. (believe me it was a seriously hostile, new york style meeting). but when i explained the internals of essbase to them, they came around. i mean the stuff is patented, and the inventor understands everything about sparse matrix math and all that eigenstuff.
there are a significant number of computational problems that solid olap technology addresses. the fact that ibm, acknowledged master of sql optimization, has opted to oem the hyperion technology rather than build their own is all the proof anyone should need.
olap engines also figure out things like, practically speaking, how would i tune the profitability of international container shipping using activity based costing methods?
data warehousing and data marting have completely different dynamics. data warehousing is all about storing everything in a consistant manner such that it can be replicated for any reason without losing its contextual meaning.
data marts are specifically designed to answer specific questions about some process of the business.
datamarts built with relational technology are quick and easy to design but offer limited functionality, performance and flexibility. datamarts built with true olap engines do all that work better, but they take a bit more brains to setup. (mostly the kind of brains that tell you relational technology doesn't cut it.)
this is the closest thing to an intelligent post i have bothered to read. the answer, my dears, is essbase. http://www.hyperion.com. essbase is the world's fastest and most capable multidimensional database engine. it has been ported to every unix and the linux port is in beta/qa.
essbase has had an open api (both in c, and vb) for years, and dozens of partners who have written front-ends like cognos, business objects, brio and a host of others which plug and play to the database.
olap is not olap in the way that relational is relational. the fact of the matter is that arbor software recruited *the* dr. codd to sum up the weaknesses of relational database for realtime access as in the way relational dweeb dba's have been attempting for years in datawarehousing. only two vendors have implemented the full scope of codd's rules. they are arbor (now merged with hyperion) who own essbase, and oracle who purchased iri's express multidimensional engine. microsoft, on the late freight, bought an israeli joint called panorama, scrounged up some of the engineers from oracle and are now trying to undercut the market with a deally called microsoft olap services now bundled with sql server 7.0. (codenamed 'plato'). for those of you with enough sense to resist microsoft, i needn't comment further on this nt only wizard-based thingy.
oracle has decommitted and begun to defund the express division. all of their marketing people have been canned. they are hoping that people will continue to remain clueless and assume that extentions to sql will handle everything 'olap'. of course unless you pay attention to the full spec of things codd said many years ago, then you are likely to buy all the marketing crap and actually believe that relational databases are going to do everything.
the momentum remains with dweebBAs with big iron who hack their way through datawarehouse projects without real server-based olap. to them it's all 'slice & dice' and other PHB front-end sillyness. in the meantime, people who know are scarce. the rest of us are making buttloads of cash in the consulting biz, or are running the market-leading hyperion.
the people who really know this stuff live at centrobe, beacon analytics, mcanalyst, data into action, pinnacle solutions, answerspace, navigator systems, ranzal & associates, symmetry and eds.
bottom line is this. there is one company who is going to crunch the uncrunchable, and i'm talking about weblogs from yahoo, and that will be hyperion. it won't be done with relational crap.