Re: XPS

Fred Nurk wrote: > "Neil Truby" <neil.truby@ardenta.com> wrote in message > news:3fo1v1F8h8niU1@individual.net... > >>"Data Goob" <datagoob@netscape.net> wrote in message >>news:Utvle.5$p5.0@fe82.usenetserver.com... >> >> >>>So the DB2 guy is pushing a big sale based >>>on a lot of CPU's, but the Power5 guy is pushing fewer CPUs to do the >>>work. Nice little quandary they got themselves into. >> >>But this is a "quandry" - I prefer the term "balancing act" - that applies >>even more to Informix resellers, because of Informix's pricing. The list >>price per cpu for an IDS 9 or 10 licence is about '34,000. The temptation >>is there then to skimp on processors, not for the hardware cost itself, >>which is relatively insignificant, but to keep down the database software >>cost. Since becoming p-Series sales accredited I have to say though that >>I am hugely impressed: we also sell HP9000 series but these Power 5 boxes >>really are the dog's bollocks, so maybe people are getting away with it >>because of the excellent hardware performance. >> >>I don't know what the answer is though, except maybe to cut the IDS price >>to at least reduce the software cost component ... >> > > With most projects, the single biggest cost is the software. Fortunately > here in OZ I can get DB2 cheaper than SQL Server (standard edition CPU) and > Oracle (Standard Edition CPU) for a Workgroup Unlimited license. Though I > am having disucssions on new Informix licenses so I can't say. > When it comes to comparison shopping "db-vendor vs MS-SQL-Server", it is important to compare the "Workgroup" version of your product with SQL-Server and not others. For example DB2 Workgroup is comparable to SQL-Server, but DB2 EEE has functionality that SQL-Server does not, such as shared-nothing clustering, and a lot of other goodies. Not many SQL-Server shops are even aware of this, not to mention what a pain-in-the-neck SQL-Server clustering is compared to a real database engine. You'll often hear SQL-Server people quickly pipe up that SQL-Server can do HA, which is so rigid and delicate as to be laughable. ( You could get Windows people into Sybase and do them a better service. :-) Anyway the point is, DB2 is competing into SQL-Server space on the Workgroup level. The license price is quite dramatic though ( 26K USD ) when considering DB2 EEE, which is what you would really want in most if not all cases in any kind of real enterprise. It is baffling why IBM is charging so much for EEE though in light of how cheap hardware is, and how prolific SQL-Server is, at least here in the USA. It might be simply that IBM knows they have a UNIX market that is still used to paying big money on databases, and so they figure keep the price up until they lose enough sales to other products--or they finally get it that they are losing on the per-CPU pricing model. Linux is the market they really want but they have to drop the price if they really want to penetrate Linux and get into MySQL's space. DB2 needs to get to that sub 2K price/CPU on the EEE to really take over the Linux market. Clustering is where it's at, not to mention how fast single-CPU or 2-CPU systems proliferate in web server environments. But this is the big problem for that IBM DB2 sales guy, who is salivating to get multiple CPU sales, and now Power5-6-7 comes out, and blade servers. The hardware people are pushing systems with fewer CPUs, virtualization, consolidation, totally ruining big ticket DB2 sales. Over in Windows-land a lot of the braindead decisions are still settling into SQL-Server because it is more of a known quantity, whereas DB2 is not--despite a lot of market share propaganda. Maybe IBM is comfortable with their Windows market which is what I've been told is their real interest, and where they've made the most sales. As far as Informix, they should just open-source the product completely. Now they have to compete with Postgres, and MySQL, both of which have made dramatic improvements. Postgres now has tablespaces, and MySQL 5 is coming out with partitioning, not to mention their in-memory shared-nothing NDB clustering. How much longer do you think your favorite "commercial" database has a chance in this market? Not long unless it goes open systems/open source. > >>>>Incidentally I saw the back page of a recent Eweek magazine with an IBM >>>>ad that used the language "DB2 or Informix". Shocking to say the least. >> > Haven't seen that one. > > >>Here in Blightly we have an ad every week in Computer Weekly, which I >>expect is, as most IBM press and TV advertising here, a re-run of a US ad, >>which is without any shadow of a doubt, or any attempt to disguise, an >>advert not for the "DB2 Information Management" brand, but for DB2 UDB v8. >>However this actually mentions Informix a couple of times too. The point >>was made several times yesterday at the UK Infobahn that awareness of >>Informix is fatally compromised by being advertising in the shadow of DB2 >>(brand or product). The IBM US speakers both pointed out, reasonably >>enough, that the situation is much better than it was, say, a year ago >>(when Informix would not get a mention at all), and that they hoped >>progress would continue to be made. >> >> >> > > >

"Data Goob" <datagoob@netscape.net> wrote in message news:58kme.3661$g6.3487@fe65.usenetserver.com... <quite a lot about DB2 UDB> Yes, your analysis is pretty good. And it's interesting that, as a US-based correspondent, you seem to recognise SQL Server as the true competition: you'll recall that the original debate was sparked by the American speakers at the recent UK Infobahn being convinced, in defiance of all UK partner experience, that Oracle is the main competition to Informix. But I'm very interested that you agree with the IBM line that DB2 is the product with which to compete against SQL Server (you go on to suggest that Informix should be made open source). My experience as a partner - and I fully accept that this battle is lost - is that IN THE MARKETS IN WHICH INFORMIX WAS STRONG - DB2 is nowhere; it is (inaccurately in my view, but my view doesn't matter) overwhelmingly perceived as a mainframe product. IBM's inabiliy to grasp this point is the true tragedy of Informix IMHO.

Data Goob wrote: > The license price is quite dramatic though ( 26K USD ) when considering > DB2 EEE, which is what you would really want in most if not all cases in > any > kind of real enterprise. It is baffling why IBM is charging so much for > EEE though in light of how cheap hardware is, and how prolific SQL-Server > is, at least here in the USA. It might be simply that IBM knows they > have a > UNIX market that is still used to paying big money on databases, and so > they figure keep the price up until they lose enough sales to other > products--or they finally get it that they are losing on the per-CPU > pricing model. EEE is a DB2 V7 release. In DB2 V8 that would match DB2 ESE (formaly DB2 V7 EE) + DPF (Data partitioning feature). You are kind of contradicting yourself if you state that any serious enterprise needs DPF (EEE). That is definitely not true and you fall into the same pit you just discussed above. DB2 ESE + DPF comepetes only with Teradata and to a small extend with Oracle RAC. DB2 is selling just fine into the BI market with DB2 ICE (integrated cluster environments) for Linux. Also let's not forget that in this profession the game changer for teh game changer is always just around the next corner. Take a look at http://linux.sys-con.com/read/87960.htm > Linux is the market they really want but they have to drop the price if > they > really want to penetrate Linux and get into MySQL's space. DB2 needs to get > to that sub 2K price/CPU on the EEE to really take over the Linux market. > Clustering is where it's at, not to mention how fast single-CPU or 2-CPU > systems proliferate in web server environments. So where does DB2 ESE + DPF compete with mySQL? This makes no sense to me.... AFAIK, mySQL isn't playing in BI.... > But this is the big > problem > for that IBM DB2 sales guy, who is salivating to get multiple CPU sales, > and > now Power5-6-7 comes out, and blade servers. The hardware people are > pushing > systems with fewer CPUs, virtualization, consolidation, totally ruining big > ticket DB2 sales. "Tel Aviv Mondamin" c'est la guerre.... Same is true for anyone selling per CPU, not just DB2. > > Over in Windows-land a lot of the braindead decisions are still settling > into > SQL-Server because it is more of a known quantity, whereas DB2 is > not--despite > a lot of market share propaganda. Maybe IBM is comfortable with their > Windows > market which is what I've been told is their real interest, and where > they've > made the most sales. You need to switch your radio station. Yes Windows is important. But it's not IBM's and it's not DB2's "real interest". I question that DB2 makes the most money on Windows. > As far as Informix, they should just open-source the product completely. Which one? XPS, IDS, RedBrick, ...? > Now they have to compete with Postgres, and MySQL, both of which have made > dramatic improvements. Postgres now has tablespaces, and MySQL 5 is coming > out with partitioning, not to mention their in-memory shared-nothing NDB > clustering. How much longer do you think your favorite "commercial" > database has a chance in this market? Not long unless it goes open systems/open > source. At the end of the day, this isn't about having a codebase winning. This is about making money. If IBM thought that there is more money to be made by open sourcing then they might do it. But what for? Cloudscape is an entry point for DB2. It makes economic sense to have an open source teaser. (A bit like Access for SQL Server). If IBM wants to make money on an open source database for its own sake, then they could equally well decide to play with mySQL. This is by the same token as IBM isn't open sourcing AIX. Instead pusing Linux.... -- Serge Rielau DB2 SQL Compiler Development IBM Toronto Lab

"Serge Rielau" <srielau@ca.ibm.com> wrote in message news:3fuepdF94om2U1@individual.net... > Also let's not forget that in this profession the game changer for teh > game changer is always just around the next corner. Could you re-phrase this, as I'd like to understand the point you are making? thanks Neil

Data Goob wrote: > How much longer do you think your favorite "commercial" > database > has a chance in this market? Not long unless it goes open systems/open > source. Let me ask a question. How many people value open source databases because 1) They can have access to the source code 2) There is a large development and supporting community 3) The product can be used for free I would contend that most people equate Open Source with 3. Very few companies can take advantage of 1, and for a lot of the OS vendors, 2 is just not happening as fast as it should (except for Linux, which seems to be the execption for succesful OS, not the rule). This seems to be especially true for the OS databases. And unfortunately it's not clear how the 'free software' approach allows a survivable business model to be built - see McVoy and the whole Bit Keeper discussion. So in the end, the business case on OS becomes one of Support. And there is nothing in the OS model that allows the OS stake holders to support their own products anymore cheaply or efficiently that anyone else. In fact, it's often easier, cheaper and more efficient for companies like IBM and even Oracle, with fully built support infrastructure in place, to provide this level of support. This is what happened with Linux, and it is what has made Linux successful. And it's support that is the annuity stream for all these companies. So my predicition is that instead of seeing commercial companies open source their own products, you will see the open source companies turning to 1 or 2 of the bigger players for indeminification and the provision of a viable support based business, perhaps with some level of royality payments going to the initial OS stakeholders.

Mark Townsend wrote: > > And unfortunately it's not clear how the 'free software' approach allows > a survivable business model to be built - see McVoy and the whole Bit > Keeper discussion. > Here's the article I was obliquely referring to http://www.forbes.com/technology/2005/05/26/cz_dl_0526linux.html?partner=rss

"Mark Townsend" <markbtownsend@comcast.net> wrote in message news:429A0D7F.9090809@comcast.net... > Data Goob wrote: > > > How much longer do you think your favorite "commercial" > > database > > has a chance in this market? Not long unless it goes open systems/open > > source. > > Let me ask a question. How many people value open source databases because > 1) They can have access to the source code > 2) There is a large development and supporting community > 3) The product can be used for free > > I would contend that most people equate Open Source with 3. Very few > companies can take advantage of 1, and for a lot of the OS vendors, 2 is > just not happening as fast as it should (except for Linux, which seems > to be the execption for succesful OS, not the rule). This seems to be > especially true for the OS databases. The greatest contribution of open source databases has been that established players like Oracle,IBM have reduced their licensing cost considerably. Why do u think Oracle's share is below $20 for 4 straight years. 20 is considered as somewhat as faultline for the share price, below which a stock is considered really low priced or junk. Point to be noted is that Oracle share has been below 20 since Jul 2001 (before 9/11, if someone wants to blame 9/11 for it). That's four years of worthless price. I think those days have gone forever when oracle cud have made money from licensing cost. They still hold market share, but it is not translating to much revenue for them.

rkusenet wrote: > > The greatest contribution of open source databases has been that established > players like Oracle,IBM have reduced their licensing cost considerably. Bollocks. Tell me exactly how much IBM or Oracle have reduced their price by in the last 2, 5 or even 10 years and I might be prepared to discuss this with you, but if you look, you will find that IBM, for instance, have actually raised their prices. > > Why do u think Oracle's share is below $20 for 4 straight years. 20 is considered > as somewhat as faultline for the share price, below which a stock is considered really > low priced or junk. Point to be noted is that Oracle share has been below 20 > since Jul 2001 (before 9/11, if someone wants to blame 9/11 for it). That's > four years of worthless price. You do know that the price you currently identify as useless is the result of over 8 splits, right ? Heres a chart that compares Oracle with Microsoft - not too shabby http://finance.yahoo.com/q/bc?t=my&s=ORCL&l=on&z=m&q=l&c=msft Here's a chart that compares Oracle with IBM - definately not shabby at all. http://finance.yahoo.com/q/bc?t=my&s=ORCL&l=on&z=m&q=l&c=ibm Now - why don't you show me a chart that compares any of the OS vendors with any of these commercial players ? > > I think those days have gone forever when oracle cud have made money > from licensing cost. They still hold market share, but it is not > translating to much revenue for them. > We are in a quiet period right now, so I can't possibly comment on the current situation, but strangely enough Gartner published some market shares figures just last week on the relational database market as a whole (revenue based). They said that in 2004 the market as a whole grew 10.3%, from 7 to 7.7 billion, of which they said Oracle garnered over 2.6 billion. Hardly a small stream of revenue. Note that they also showed Oracle growing at 14.3%, 4 percentage points faster than the market, and actually gaining marketshare. They also gave Oracle 80% of the Linux marketshare, with a very healthy .5 billion dollars in revenue coming to Oracle on this platform alone. They also said that IBM only grew 5.6% in total (i.e actually slower than the overall market) and even managed to lose marketshare on Windows. They also noted that the Informix growth on Unix over the last year was actually down nearly 20%, that their growth on Linux was 1/14 of Oracle's, and that indeed "much of that growth [IBM's] was generated by DB2 on the zSeries platform". So what you say about OS pricing affecting a commercial vendor's revenue stream only seems to be partly right.

"Mark Townsend" <markbtownsend@comcast.net> wrote in message news:429A3392.90502@comcast.net... > Bollocks. Tell me exactly how much IBM or Oracle have reduced their > price by in the last 2, 5 or even 10 years and I might be prepared to > discuss this with you, but if you look, you will find that IBM, for > instance, have actually raised their prices. what about the introduction of workgroup edition etc, aka cheap edition. I am talking about CPU based licensing model. I am quite positive that CPU based WG model wasn't available until few years ago. Correct me if I am wrong. > You do know that the price you currently identify as useless is the > result of over 8 splits, right ? > Heres a chart that compares Oracle with Microsoft - not too shabby > http://finance.yahoo.com/q/bc?t=my&s=ORCL&l=on&z=m&q=l&c=msft > Here's a chart that compares Oracle with IBM - definately not shabby at > all. > http://finance.yahoo.com/q/bc?t=my&s=ORCL&l=on&z=m&q=l&c=ibm impressive because of phenomenal performance in the good ol days. What about those who purchased shares after jun 2001 and are holding on to it like a toilet paper. Don't u think four yrs running of < $20 tells something. Why do companies split in the first place. So that it keeps sitting in the basement. > Now - why don't you show me a chart that compares any of the OS vendors > with any of these commercial players ? > They said that in 2004 the market as a whole grew 10.3%, from 7 to 7.7 > billion, of which they said Oracle garnered over 2.6 billion. Hardly a > small stream of revenue. > > Note that they also showed Oracle growing at 14.3%, 4 percentage points > faster than the market, and actually gaining marketshare. They also gave > Oracle 80% of the Linux marketshare, with a very healthy .5 billion > dollars in revenue coming to Oracle on this platform alone. I have serious doubts of gaining market share resulting in increasing revenue in the same proportion. > They also said that IBM only grew 5.6% in total (i.e actually slower > than the overall market) and even managed to lose marketshare on > Windows. They also noted that the Informix growth on Unix over the last > year was actually down nearly 20%, that their growth on Linux was 1/14 > of Oracle's, and that indeed "much of that growth [IBM's] was generated > by DB2 on the zSeries platform". > > So what you say about OS pricing affecting a commercial vendor's revenue > stream only seems to be partly right. how is that soothing to anyone who wants to invest in Oracle shares or is already regretting after jun 2001. OK as a senior oracle exec, can u tell me why Oracle share is in the basement for the last 4 yrs, despite these impressive market share. After every Quartly result when I see more claims of Oracle growth, I also see flat share price. For what market is punishing oracle? Something doesn't jell Mark. U are hiding something :-) Do you know that in the last 4 yrs, more than 80% of the time Sybase stock has been traded higher than Oracle. Isn't that a shocker. And we all thought Wall street is smart to punish bad companies and reward good companies.

rkusenet wrote: > "Mark Townsend" <markbtownsend@comcast.net> wrote in message > news:429A3392.90502@comcast.net... > > >>Bollocks. Tell me exactly how much IBM or Oracle have reduced their >>price by in the last 2, 5 or even 10 years and I might be prepared to >>discuss this with you, but if you look, you will find that IBM, for >>instance, have actually raised their prices. > > > what about the introduction of workgroup edition etc, aka cheap edition. > I am talking about CPU based licensing model. I am quite positive that > CPU based WG model wasn't available until few years ago. Correct me > if I am wrong. > You mean IBM's Workgroup Unlimited pricing ? Your are correct - they did take a huge cut - around 45% I believe. From memory they lowered this at about the same time they raised the pricing on the high end product - towards the end of 2002 ? I think they always had a Workgroup edition, however (didn't they go from $15k to $7.5K). I'm pretty certain IBM didn't lower their price in the Workgroup because they thought MySQL or OS was a threat. They seemed to just want to come in under MS's pricing - at least their user based pricing in the mid ground always seemed to be very close to MS's CAL based pricing. To me the Unlimited stuff always seemed to be a little bit of an intelligence test - why would anybody pay per CPU when they could get the same product per user for a LOT cheaper. >> >>So what you say about OS pricing affecting a commercial vendor's revenue >>stream only seems to be partly right. > > > how is that soothing to anyone who wants to invest in Oracle shares or > is already regretting after jun 2001. > OK as a senior oracle exec, can u tell me why Oracle share is in the basement > for the last 4 yrs, despite these impressive market share. After every Quartly > result when I see more claims of Oracle growth, I also see flat share price. > For what market is punishing oracle? Something doesn't jell Mark. U are hiding > something :-) Nope. I still argue that your assertion that Oracle is a junk stock is wrong. It's true that the shares have been flat for the last few years, as have pretty much everybodies in the tech industry. There are a couple of reasons for this, not the least being the phenomenal run up during the internet boom, which has made period on period comparisons very hard. Personally I think share prices are largely cyclic - they have their ups and downs. If you can find a company that can ride the ups to their fullest extent, and then turn a down into 'flat', then long term (i.e kids college fund) I think you will come out OK. For the short term (i.e tonight), I prefer poker :-)

"Mark Townsend" <markbtownsend@comcast.net> wrote in message news:429A47B3.9020806@comcast.net... > You mean IBM's Workgroup Unlimited pricing ? Your are correct - they did > take a huge cut - around 45% I believe. From memory they lowered this at > about the same time they raised the pricing on the high end product - > towards the end of 2002 ? I think they always had a Workgroup edition, > however (didn't they go from $15k to $7.5K). I did mention IBM also. I am not taking on Oracle alone. > To me the Unlimited stuff always seemed to be a little bit of an > intelligence test - why would anybody pay per CPU when they could get > the same product per user for a LOT cheaper. OK, please prove it me that unlimited user license for Oracle RDBMS on a 4CPU box was available at $22k per processor in 2000, regardless of any version. Today it is. I was in Minneapolis at that time and I still remember someone mentioning to me. "oracle is obscene. 50K per processor". IIRC, at that time WG edition was there but on a user count basis - thus making it unsuitable for a web application. > Nope. I still argue that your assertion that Oracle is a junk stock is > wrong. correction. Oracle stock has become worthless for quite some time. Only a fool will deny that O stock has been one of the most rewarding tech stock till 2001. But then there is the golden rule of stock market: Past performance is no guarantee of future. 4 yrs is a lot of time to remain < 20. I can imagine the frustration of those who are sitting on it for years. and btw the yahoo chart you showed earlier was powered by MySQL. Remember the ad "Yahoo uses oracle" :-)

rkusenet wrote: > OK, please prove it me that unlimited user license for Oracle RDBMS on a 4CPU box > was available at $22k per processor in 2000, regardless of any version. Oracle was using three pricing metrics in those days - unlimited user pricing based on the MHz of the CPUs, named users and concurrent user. In 2000 the unlimited user pricing for Oracle Standard Edititon was $15 per MHz. Most machines at the low end were around 600-700 Mhz in those days - so around $10K per CPU for Standard Edition. FYI - People hated the MHz pricing, and as the speed of the CPUs increased the pricing model simply wasn't sustainable. So for this reason the fixed $15K per processor pricing for unlimited users Standard Edition came out with Oracle9i in 2001. I'm not sure where the $22K figure comes from ? Or the $50K figure ? > > and btw the yahoo chart you showed earlier was powered by MySQL. Remember > the ad "Yahoo uses oracle" :-) > There is an implication here that Yahoo Finance migrated from Oracle to MySQL. They didn't. They were using BTrieve (and some flat file stuff they had written themselves) and they moved to MySQL - for what is actually a low throughput application supporting not much more then 60 concurrent user connections per database. See http://www.mysql.com/news-and-events/success-stories/yahoo_finance.html There are a small number of highly visible websites that serve content out of highly replicated MySQL farms, using 10's or even 100's of copies of the same database. This is a _great_ use of the MySQL technology, given it's current capabilities and pricing, and a much better approach than using a flat file, or a non relational data management system. However, if you think MySQL is replacing Oracle, DB2, Microsoft or even Informix in the OLTP or Data Warehousing space, or even under packaged applications, then you are completely mistaken.

"Mark Townsend" <markbtownsend@comcast.net> wrote > Oracle was using three pricing metrics in those days - unlimited user > pricing based on the MHz of the CPUs, named users and concurrent user. > > In 2000 the unlimited user pricing for Oracle Standard Edititon was $15 > per MHz. Most machines at the low end were around 600-700 Mhz in those > days - so around $10K per CPU for Standard Edition. > > FYI - People hated the MHz pricing, and as the speed of the CPUs > increased the pricing model simply wasn't sustainable. So for this > reason the fixed $15K per processor pricing for unlimited users Standard > Edition came out with Oracle9i in 2001. > > I'm not sure where the $22K figure comes from ? Or the $50K figure ? $22k per processor is what I saw last year when trying to prepare a feasibility study of migrating out of Informix (I no longer work for that company). $22k is Canadian Dollars. $50K (in 2000) was I think referred to some high end Solaris/HP box running 8 or more CPUS. I don't recollect anything more at this time. I wasn't working with Oracle, not even remotely concerned. The guy who told me was in a O project. Are you suggesting that $50K per CPU isn't just possible for a high end machine. > There is an implication here that Yahoo Finance migrated from Oracle to > MySQL. They didn't. there is no such implication. I posted that bcos I know how easily non IT folks misunderstand an ad like this "YAHOO USES ORACLE" or "ORACLE POWERS YAHOO". I bet the first thing they think of is a site like yahoo/finance. That's why I clarified from my friend working at Visa (San Mateo) on which DB they use for their most intensive OLTP application. It was DB2. Jonathan: Wasn't Visa using Informix not too long ago. IIRC you briefly worked at visa also, at least your signature had visa email address instead of Informix email address. BTW one of the better Mysql implementation is Sabre which I think uses a massive cluster of 100 Linux box. http://www.mysql.com/customers/customer.php?id=21 Someone from Sabre said that "if we have used Oracle, then the license cost alone would have killed us". I do feel that open source database are going to eat the lunch of commercial databases like SQLServer,DB2,Oracle. The question is not if, but when. One day it will cross the critical point and after which there will be a flood of migration. As companies start looking to cut cost (recollect how skeptical were people about outsourcing to India even 6/7 yrs ago), Mysql will be a major beneficiary. There was a case of Canadian Online Pharmacy company (selling drugs to US residents :-)) which migrated from IIS/COM/SQLServer to LAMP model and they were very happy.

rkusenet wrote: > "Mark Townsend" <markbtownsend@comcast.net> wrote > > >>Oracle was using three pricing metrics in those days - unlimited user >>pricing based on the MHz of the CPUs, named users and concurrent user. >> >>In 2000 the unlimited user pricing for Oracle Standard Edititon was $15 >>per MHz. Most machines at the low end were around 600-700 Mhz in those >>days - so around $10K per CPU for Standard Edition. >> >>FYI - People hated the MHz pricing, and as the speed of the CPUs >>increased the pricing model simply wasn't sustainable. So for this >>reason the fixed $15K per processor pricing for unlimited users Standard >>Edition came out with Oracle9i in 2001. >> >>I'm not sure where the $22K figure comes from ? Or the $50K figure ? > > > $22k per processor is what I saw last year when trying to prepare > a feasibility study of migrating out of Informix (I no longer work > for that company). $22k is Canadian Dollars. > > $50K (in 2000) was I think referred to some high end Solaris/HP box running > 8 or more CPUS. I don't recollect anything more at this time. I > wasn't working with Oracle, not even remotely concerned. The guy > who told me was in a O project. > > Are you suggesting that $50K per CPU isn't just possible for a high end machine. > > >>There is an implication here that Yahoo Finance migrated from Oracle to >>MySQL. They didn't. > > > there is no such implication. I posted that bcos I know how easily > non IT folks misunderstand an ad like this "YAHOO USES ORACLE" or > "ORACLE POWERS YAHOO". I bet the first thing they think of is a > site like yahoo/finance. That's why I clarified from my friend > working at Visa (San Mateo) on which DB they use for their most > intensive OLTP application. It was DB2. > > Jonathan: Wasn't Visa using Informix not too long ago. IIRC you > briefly worked at visa also, at least your signature had visa > email address instead of Informix email address. > > BTW one of the better Mysql implementation is Sabre which I think > uses a massive cluster of 100 Linux box. > > http://www.mysql.com/customers/customer.php?id=21 > > Someone from Sabre said that "if we have used Oracle, then the > license cost alone would have killed us". > > I do feel that open source database are going to eat the lunch > of commercial databases like SQLServer,DB2,Oracle. The question > is not if, but when. One day it will cross the critical point > and after which there will be a flood of migration. As companies > start looking to cut cost (recollect how skeptical were people > about outsourcing to India even 6/7 yrs ago), Mysql will be a > major beneficiary. > > There was a case of Canadian Online Pharmacy company (selling > drugs to US residents :-)) which migrated from IIS/COM/SQLServer > to LAMP model and they were very happy. > > Interesting thread. For the record I agree largely with Mark. Companies like Oracle and IBM do not feed of the commodity market. The features implemented into the open source DBMS are following the trail blazed by the major vendor in the last 25 years. I dare say that a similar situation holds for OS. To the best of my knowledge Linux has yet to provide innovative value. I'm not aware of any open source project that has provided true innovation and I'm at a loss at how that would be achieved. As long as there are advanced features required there is room for commercial products. If there isn't, companies like Oracle and IBM move on to higher ground (see Lenovo, Hitachi, ...). Again this isn't about taking over the world with ones code base. It's about feeding shareholders expectations. If cheese slicers were still high value IBM would still sell them. Don't expect a publicly traded company to have any emotional attachement to a given product. Cheers Serge -- Serge Rielau DB2 SQL Compiler Development IBM Toronto Lab

Neil Truby wrote: > "Data Goob" <datagoob@netscape.net> wrote in message > news:58kme.3661$g6.3487@fe65.usenetserver.com... > <quite a lot about DB2 UDB> > > Yes, your analysis is pretty good. And it's interesting that, as a US-based > correspondent, you seem to recognise SQL Server as the true competition: > you'll recall that the original debate was sparked by the American speakers > at the recent UK Infobahn being convinced, in defiance of all UK partner > experience, that Oracle is the main competition to Informix. > I live in a very strong Microsoft market, that is also strong with Oracle. But many of the shops in my region are heterogenious, having a mix of several products for whatever reason(s). I came to my conclusions based on some research that led me to DB2 as being a very good product, and easy-to-support if you put it into a very strong SQL-Server shop--in other words, people who don't really know a good DB can get up to speed with DB2 relatively quickly simply because of the ease-of-use of DB2. Not to mention the pricing structure for the Workgroup DB2 compares with SQL-Server. It also reduces the risk to suggest a product you know will be very successful in an environment where the people will depend on it pretty much taking care of itself. Sorry, but Informix does not fit that model well in SQL-Server shops--nobody is old enough to know about Informix, nor do they care, and without strong brand recognition like Oracle, they ignore it, or fear it. Not to mention that the people taking care of it won't know the difference between a logical log and a phyical log, or buffer cache, LRUs, are you kidding! > But I'm very interested that you agree with the IBM line that DB2 is the > product with which to compete against SQL Server In SQL-Server ( read heavy Microsoft ) environments it is a long haul to get anything else into the shop that doesn't have Redmonds' blessing on it. There is as much a rhythm to an MS shop as much as a UNIX or Linux environment, and as everyone here already knows, if it doesn't have a good ( solid ) GUI, it ain't gonna fly well in a strong MS shop. There is just too little interest in a lot of MS people learning anything outside of MS. What got the last MS shop I was in to actually buy DB2 after almost TWO YEARS of selling it was to get them to see that the migration was indeed possible, and relatively painless ( not a lot of learning involved ) and it ran faster on Windows than SQL-Server. From an Informix mind, it is simple, voila, add real memory management to a database engine and amazing things can happen. It would have been great to get Informix into this particular environment, but who will support it? Me? No way, I'm not gonna be the only one in town that can support the product, or be the point of blame when things go wrong. With mainstream products you can ease the fear that the right choice was made, no way Informix would have been the right choice. DB2 was right simply because of the very thing you alluded to, it's a mainframe product, and if banks can use it, well, it's probably pretty darn good and good enough to replace SQL-Server. This is the kind of logic that MS shops understand. And for all the poo-poo against DB2 here, I do not see table-level memory management in Informix that DB2 has, which for big data is pretty darn nice. (you go on to suggest that > Informix should be made open source). Open source does not have to be "free". MySQL is thriving as open source. But it also helps to understand why MySQL is such a success. In many of the environments it exists in, many of the instances are unsupported except for the local system administrator. But this is also what keeps out the competition! MySQL as a company realises that they will have x number of unsupported sites for every supported site. Eventually many of those unsupported sites move into a supported model when they take it seriously and realise that support is important. Many also will not, but again, MySQL is sitting there and not Oracle or DB2 or Informix or even Sybase for that matter. But the pricing is also reasonable, not so over-the-top like DB2 or Oracle. IBM is probably not ready for everyone to get DB2 or they would adopt the open source model. Probably not ready for Informix to go open source either, that would probably screw up a lot of the marketing. LOL! :-) The real confusion for IBM though is that they still want to car-sale the db market, instead of just selling at a price that sells into the 1-CPU system space at open source pricing. When you see DB2 EEE ( or whatever it is now ) going for the same prices as MySQL, you will see IBM finally taking open source seriously. ( This is where Informix could really be interesting, as an open source product. ) My experience as a partner - and I > fully accept that this battle is lost - is that IN THE MARKETS IN WHICH > INFORMIX WAS STRONG - DB2 is nowhere; it is (inaccurately in my view, but my > view doesn't matter) overwhelmingly perceived as a mainframe product. > > IBM's inabiliy to grasp this point is the true tragedy of Informix IMHO. > I agree, DB2 is considered a mainframe or AS400 product, that can run on Windows, or UNIX, or Linux. Until they market it as a Linux leader, and explain clustering to the masses, it will take more time to get it to the top in Linux. It is superior to every open source database out there, but don't bother telling ISPs or web server farms, they already have MySQL! This by the way should have been Informix SE's big play, but we all know that SE is all but dead. SE should be where MySQL is today, as the market leader in Linux, with a great opportunity for businesses to move into a shared-memory engine when they outgrow SE and start to really get big data. MySQL also appeals to developers, something Informix thought was unimportant, and now IBM will dictate the future of Informix products with their own take on what's important for developers. Sad, but that's the way it goes. SE could have become open source and totally dominated the market.

Serge Rielau wrote: > Interesting thread. For the record I agree largely with Mark. > Companies like Oracle and IBM do not feed of the commodity market. Actually IBM is trying to push DB2 as the next commodity db as a part of the operating system. What platform is it? iSeries? that uses DB2 as part of the system to maintain the system. > The features implemented into the open source DBMS are following the > trail blazed by the major vendor in the last 25 years. And because of the perceived high-cost of top tier databases, you're watching the reinvention of the database market coming out of the open source market. MySQL, PostGres, etc. etc. But please, go ahead and ignore them. By the way Cloudscape is interesting, but not the big play, it will have to come over in MySQL's space. > I dare say that a similar situation holds for OS. To the best of my > knowledge Linux has yet to provide innovative value. Flame-bait right? > I'm not aware of any open source project that has provided true > innovation and I'm at a loss at how that would be achieved. Uhm, actually, there are a lot of internet-based marketing companies using MySQL exclusively, and no big-gun database anywhere on site. I know of several in my region who are doing quite well, one to the tune of $300 Million USD annually in revenue, all on MySQL. The innovation is not just the technology but how the business takes cheap software and hardware and competes with companies that use expensive hardware and expensive software. Now THAT is interesting, and innovative. To also say that Linux has no innovative value is to be a real idiot, and not understand the market. It's not about "advanced" features, it's about reasonable pricing and marketing of your products. > As long as there are advanced features required there is room for > commercial products. If there isn't, companies like Oracle and IBM move > on to higher ground (see Lenovo, Hitachi, ...). > Again this isn't about taking over the world with ones code base. > It's about feeding shareholders expectations. If cheese slicers were > still high value IBM would still sell them. > Don't expect a publicly traded company to have any emotional attachement > to a given product. > > Cheers > Serge

Serge Rielau wrote: > > EEE is a DB2 V7 release. In DB2 V8 that would match DB2 ESE (formaly DB2 > V7 EE) + DPF (Data partitioning feature). > You are kind of contradicting yourself if you state that any serious > enterprise needs DPF (EEE). That is definitely not true and you fall > into the same pit you just discussed above. Don't know what pit that is. You obviously do not work in the real world. > DB2 ESE + DPF comepetes only with Teradata and to a small extend with > Oracle RAC. > DB2 is selling just fine into the BI market with DB2 ICE (integrated > cluster environments) for Linux. OK so the branding and labeling for DB2 is a mess, thanks for clearing it up for me. ICE is shared-nothing--isn't that EEE or whatever-the-f*** IBM calls it. Very confusing marketing for DB2. As I understand it there are only two Linux products for DB2, one with clustering and one without. Call them whatever you want. One is a workgroup product the other is a "enterprise" product that **can** cluster. I don't know what else there is I've been too busy with MySQL lately. :-) But back on point, to trivialize clustering is to show that you have no grasp of the market or where it's heading. Teradata is a niche player and extremely expensive, and very limited in how you can use it last time I checked. It's also bolt-on, buy and add, very expensive. Commodity hardware and software, clustered, IS where things are going. If you buy anything over 4 CPUs you better have a big budget to absorb equipment that will be old in less than a year. So it's a better play to buy 1 or 2-CPU boxes, cluster them and rotate out old hardware faster, cheaper, etc. It's also a lot easier to repurpose a small system than a big system with big visibility to management. > Also let's not forget that in this profession the game changer for teh > game changer is always just around the next corner. > Take a look at > http://linux.sys-con.com/read/87960.htm > .. .. like I said... .. > So where does DB2 ESE + DPF compete with mySQL? This makes no sense to > me.... AFAIK, mySQL isn't playing in BI.... > Not yet but it's coming. MySQL 5 is coming out with partitioning, and a shared-nothing architecture is available TODAY. Why buy expensive DB2 when you can get MySQL for less? Drop the price and get in the game. >> But this is the big problem >> for that IBM DB2 sales guy, who is salivating to get multiple CPU >> sales, and >> now Power5-6-7 comes out, and blade servers. The hardware people are >> pushing >> systems with fewer CPUs, virtualization, consolidation, totally >> ruining big >> ticket DB2 sales. > > "Tel Aviv Mondamin" c'est la guerre.... > Same is true for anyone selling per CPU, not just DB2. > >> >> Over in Windows-land a lot of the braindead decisions are still >> settling into >> SQL-Server because it is more of a known quantity, whereas DB2 is >> not--despite >> a lot of market share propaganda. Maybe IBM is comfortable with their >> Windows >> market which is what I've been told is their real interest, and where >> they've >> made the most sales. > > You need to switch your radio station. > Yes Windows is important. But it's not IBM's and it's not DB2's "real > interest". I question that DB2 makes the most money on Windows. > I question that DB2 is anywhere but on an AS400 or mainframe, yet I hear how it's the market leader. I **do** know that I actually SEE MySQL on a lot of systems, as well as MS-SQL-Server, and Oracle, on Linux and Windows. I'm not sure where DB2 is outside of AIX or AS400, or a few Windows shops, but it does exist, and I'm probably not in step with all the hidden DB2 going on out there, but the last IBM rep I talked to last year told me they sold the most on Windows, not Linux. And all the press says IBM has a big market share, so it must be there but I just haven't been lucky enough to see it. :-) >> As far as Informix, they should just open-source the product completely. > > Which one? XPS, IDS, RedBrick, ...? > All of them. >> Now they have to compete with Postgres, and MySQL, both of which have >> made >> dramatic improvements. Postgres now has tablespaces, and MySQL 5 is >> coming >> out with partitioning, not to mention their in-memory shared-nothing NDB >> clustering. How much longer do you think your favorite "commercial" >> database has a chance in this market? Not long unless it goes open >> systems/open source. > > At the end of the day, this isn't about having a codebase winning. > This is about making money. If IBM thought that there is more money to > be made by open sourcing then they might do it. IBM can't admit to anything the wrong way or they lose credibility. Just like Microsoft. Eventually the market will decide. > But what for? Cloudscape is an entry point for DB2. Cloudscape is an entry point for Java, and that's about it but has nothing to do with DB2. It is more similar to an Informix SE, or even MySQL. > It makes economic > sense to have an open source teaser. (A bit like Access for SQL Server). Then they should have a DB2 lite or something, but Cloudscape is not DB2. You would have to increase your LSD intake to get there on that one. Cloudscape is interesting for somebody who wants a Java database, but that's about it. > If IBM wants to make money on an open source database for its own sake, > then they could equally well decide to play with mySQL. This is by the > same token as IBM isn't open sourcing AIX. Instead pusing Linux.... > It's really about IBM understanding the database market, which I am still not sure they do. They understand their own market, but I've yet to see much about IBM thinking about databases outside their own box. It just hasn't happened big yet. Like I said when they start to actually respond to the MySQL space then I'll believe that they actually acknowledge a market outside themselves. Currently they do not. MS-SQL-Server is a relatively easy play compared to Linux land, where in Linux they have to compete on price first before anything else, and then convince people that it is better than MySQL or some other product that is cheaper. Oracle actually has a better chance competing head-to-head with MySQL simply because most Linux people know the Oracle brand, but like we already talked about, DB2 is a mainframe product as far as a lot of Linux people know. Oracle will probably win a lot in this arena simply on its own momentum and similarity-befuddlement of tech terminology to MySQL they will use to win the hearts and minds of Linux shops. It's almost a play out of the Microsoft playbook. :-)

> Neil Truby wrote: >> "Data Goob" <datagoob@netscape.net> wrote in message > This is the kind of logic that MS shops understand. And for all the > poo-poo > against DB2 here, I do not see table-level memory management in Informix > that DB2 has, which for big data is pretty darn nice. Most people don't want to be negative about DB2 - as you say, it has a number of attractive features that IDS lacks. But it sticks in the craw that DB2 UDB is remorselessly promoted as a product (the "We only promote the Brand" mantra is bullshit) to the at-one-time complete, and now just virtual, exclusion of Informix. And, - this is comp.databases.informix btw :-) - we believe that Informix is a superior product, and four years ago had a stronger recognition in its market (that's largely been dissipated now of course).