February 9, 2017 at 11:45 am
Gary Varga - Thursday, February 9, 2017 8:09 AMEric M Russell - Thursday, February 9, 2017 6:49 AMAn operating system is to a database server what a buyer/seller agent, attorney, and county clerk are to a real-estate purchase. It's entirely reasonable to ask what the middle men bring to the transaction in terms of financial, administrative, and performance overhead. Like I mentioned before, there is nothing essential that the operating system provides for the database that the database engine couldn't theoretically provide for itself better if it were extended to include that low level functionality.Didn't Oracle try that in the 90s? Didn't they call it something like Raw Iron?
Yes, but it wasn't popular with customers. It makes more sense today in cloud hosted environments.
"Do not seek to follow in the footsteps of the wise. Instead, seek what they sought." - Matsuo Basho
May 9, 2017 at 3:55 am
Eric M Russell - Thursday, February 9, 2017 11:45 AMGary Varga - Thursday, February 9, 2017 8:09 AMEric M Russell - Thursday, February 9, 2017 6:49 AMAn operating system is to a database server what a buyer/seller agent, attorney, and county clerk are to a real-estate purchase. It's entirely reasonable to ask what the middle men bring to the transaction in terms of financial, administrative, and performance overhead. Like I mentioned before, there is nothing essential that the operating system provides for the database that the database engine couldn't theoretically provide for itself better if it were extended to include that low level functionality.Didn't Oracle try that in the 90s? Didn't they call it something like Raw Iron?
Yes, but it wasn't popular with customers. It makes more sense today in cloud hosted environments.
Although they announced in in 1998, they had gotten nowhere with doing it at the beginning of 2002 when they announced the agreement with Sun to develop it (using chunks of Solaris to do the low level bits); but that "agreement" turned out to be more an agreement to have an agreement, there was still lot of detail to be worked out over who would do what, how it would be sold, and so on. Ellison had claimed agreement on hardware was close with Compaq and IBM and had announced agreements with Dell and HP some time earlier, but none of Dell, HP, and Compaq had anything to say about the project in late 2001/early 2002 (I don't know whether IBM did). The general feeling seemed to be that nearly 4 years after the project was first announced it was never going to fly (no-one believed Oracle could do n OS substitute, not even with Sun's help if they did get that. Shortly afterwards, with no customer interest and hopelessly slipping schedules, it was effectively dead. Some time in the next six years Oracle decided to have a new "database appliance" project, again involving IBM, HP, Dell and Sun and with EMC now in the mix, and when talking about that Oracle's Mendelssohn commented "We tried it with Raw Iron 5 years ago" and "Customers didn't see the imperative" (I guess he thought that 5 years on they would).
So saying the tried it in the 90s is not really accurate - what they did in the 90s was announce they were going to try it, and start some mostly ineffectual negotiations with some possible partners in such a project. They canned it some time in 2002, I think. And started a new attempt at the same sort of thing in 2007.
As I hadn't touched Oracle since 1996 I stopped keeping track of what they were up to when I retired in 2009 (reckoning that my rule that I would only take on really interesting stuff absolutely ruled out anything to do with Oracle) I don't know what (if anything) happened to the "Oracle Database Appliance" idea after that; but surely it must be dead now, replaced by something cloud-oriented?
Tom
May 9, 2017 at 8:48 am
TomThomson - Tuesday, May 9, 2017 3:55 AMEric M Russell - Thursday, February 9, 2017 11:45 AMGary Varga - Thursday, February 9, 2017 8:09 AMEric M Russell - Thursday, February 9, 2017 6:49 AMAn operating system is to a database server what a buyer/seller agent, attorney, and county clerk are to a real-estate purchase. It's entirely reasonable to ask what the middle men bring to the transaction in terms of financial, administrative, and performance overhead. Like I mentioned before, there is nothing essential that the operating system provides for the database that the database engine couldn't theoretically provide for itself better if it were extended to include that low level functionality.Didn't Oracle try that in the 90s? Didn't they call it something like Raw Iron?
Yes, but it wasn't popular with customers. It makes more sense today in cloud hosted environments.
Although they announced in in 1998, they had gotten nowhere with doing it at the beginning of 2002 when they announced the agreement with Sun to develop it (using chunks of Solaris to do the low level bits); but that "agreement" turned out to be more an agreement to have an agreement, there was still lot of detail to be worked out over who would do what, how it would be sold, and so on. Ellison had claimed agreement on hardware was close with Compaq and IBM and had announced agreements with Dell and HP some time earlier, but none of Dell, HP, and Compaq had anything to say about the project in late 2001/early 2002 (I don't know whether IBM did). The general feeling seemed to be that nearly 4 years after the project was first announced it was never going to fly (no-one believed Oracle could do n OS substitute, not even with Sun's help if they did get that. Shortly afterwards, with no customer interest and hopelessly slipping schedules, it was effectively dead. Some time in the next six years Oracle decided to have a new "database appliance" project, again involving IBM, HP, Dell and Sun and with EMC now in the mix, and when talking about that Oracle's Mendelssohn commented "We tried it with Raw Iron 5 years ago" and "Customers didn't see the imperative" (I guess he thought that 5 years on they would).
So saying the tried it in the 90s is not really accurate - what they did in the 90s was announce they were going to try it, and start some mostly ineffectual negotiations with some possible partners in such a project. They canned it some time in 2002, I think. And started a new attempt at the same sort of thing in 2007.
As I hadn't touched Oracle since 1996 I stopped keeping track of what they were up to when I retired in 2009 (reckoning that my rule that I would only take on really interesting stuff absolutely ruled out anything to do with Oracle) I don't know what (if anything) happened to the "Oracle Database Appliance" idea after that; but surely it must be dead now, replaced by something cloud-oriented?
No, the Oracle database appliance is still alive and well. The latest iteration is called Oracle Exadata and it is a pretty awesome platform. Sun Hardware, Linux OS and Oracle RAC database. Not cheap though 🙁
May 9, 2017 at 11:51 am
loyd.gravitt - Tuesday, May 9, 2017 8:48 AMTomThomson - Tuesday, May 9, 2017 3:55 AMEric M Russell - Thursday, February 9, 2017 11:45 AMGary Varga - Thursday, February 9, 2017 8:09 AMEric M Russell - Thursday, February 9, 2017 6:49 AMAn operating system is to a database server what a buyer/seller agent, attorney, and county clerk are to a real-estate purchase. It's entirely reasonable to ask what the middle men bring to the transaction in terms of financial, administrative, and performance overhead. Like I mentioned before, there is nothing essential that the operating system provides for the database that the database engine couldn't theoretically provide for itself better if it were extended to include that low level functionality.Didn't Oracle try that in the 90s? Didn't they call it something like Raw Iron?
Yes, but it wasn't popular with customers. It makes more sense today in cloud hosted environments.
Although they announced in in 1998, they had gotten nowhere with doing it at the beginning of 2002 when they announced the agreement with Sun to develop it (using chunks of Solaris to do the low level bits); but that "agreement" turned out to be more an agreement to have an agreement, there was still lot of detail to be worked out over who would do what, how it would be sold, and so on. Ellison had claimed agreement on hardware was close with Compaq and IBM and had announced agreements with Dell and HP some time earlier, but none of Dell, HP, and Compaq had anything to say about the project in late 2001/early 2002 (I don't know whether IBM did). The general feeling seemed to be that nearly 4 years after the project was first announced it was never going to fly (no-one believed Oracle could do n OS substitute, not even with Sun's help if they did get that. Shortly afterwards, with no customer interest and hopelessly slipping schedules, it was effectively dead. Some time in the next six years Oracle decided to have a new "database appliance" project, again involving IBM, HP, Dell and Sun and with EMC now in the mix, and when talking about that Oracle's Mendelssohn commented "We tried it with Raw Iron 5 years ago" and "Customers didn't see the imperative" (I guess he thought that 5 years on they would).
So saying the tried it in the 90s is not really accurate - what they did in the 90s was announce they were going to try it, and start some mostly ineffectual negotiations with some possible partners in such a project. They canned it some time in 2002, I think. And started a new attempt at the same sort of thing in 2007.
As I hadn't touched Oracle since 1996 I stopped keeping track of what they were up to when I retired in 2009 (reckoning that my rule that I would only take on really interesting stuff absolutely ruled out anything to do with Oracle) I don't know what (if anything) happened to the "Oracle Database Appliance" idea after that; but surely it must be dead now, replaced by something cloud-oriented?No, the Oracle database appliance is still alive and well. The latest iteration is called Oracle Exadata and it is a pretty awesome platform. Sun Hardware, Linux OS and Oracle RAC database. Not cheap though 🙁
From what I read, Oracle Database Appliance is just a tightly integrated cluster sharing disk storage, and not a MPP (massively parallel shared) architecture to the same extent that Azure SQL Warehouse is.
"Do not seek to follow in the footsteps of the wise. Instead, seek what they sought." - Matsuo Basho
May 10, 2017 at 12:49 pm
Eric M Russell - Tuesday, May 9, 2017 11:51 AMloyd.gravitt - Tuesday, May 9, 2017 8:48 AMTomThomson - Tuesday, May 9, 2017 3:55 AMEric M Russell - Thursday, February 9, 2017 11:45 AMGary Varga - Thursday, February 9, 2017 8:09 AMEric M Russell - Thursday, February 9, 2017 6:49 AMAn operating system is to a database server what a buyer/seller agent, attorney, and county clerk are to a real-estate purchase. It's entirely reasonable to ask what the middle men bring to the transaction in terms of financial, administrative, and performance overhead. Like I mentioned before, there is nothing essential that the operating system provides for the database that the database engine couldn't theoretically provide for itself better if it were extended to include that low level functionality.Didn't Oracle try that in the 90s? Didn't they call it something like Raw Iron?
Yes, but it wasn't popular with customers. It makes more sense today in cloud hosted environments.
Although they announced in in 1998, they had gotten nowhere with doing it at the beginning of 2002 when they announced the agreement with Sun to develop it (using chunks of Solaris to do the low level bits); but that "agreement" turned out to be more an agreement to have an agreement, there was still lot of detail to be worked out over who would do what, how it would be sold, and so on. Ellison had claimed agreement on hardware was close with Compaq and IBM and had announced agreements with Dell and HP some time earlier, but none of Dell, HP, and Compaq had anything to say about the project in late 2001/early 2002 (I don't know whether IBM did). The general feeling seemed to be that nearly 4 years after the project was first announced it was never going to fly (no-one believed Oracle could do n OS substitute, not even with Sun's help if they did get that. Shortly afterwards, with no customer interest and hopelessly slipping schedules, it was effectively dead. Some time in the next six years Oracle decided to have a new "database appliance" project, again involving IBM, HP, Dell and Sun and with EMC now in the mix, and when talking about that Oracle's Mendelssohn commented "We tried it with Raw Iron 5 years ago" and "Customers didn't see the imperative" (I guess he thought that 5 years on they would).
So saying the tried it in the 90s is not really accurate - what they did in the 90s was announce they were going to try it, and start some mostly ineffectual negotiations with some possible partners in such a project. They canned it some time in 2002, I think. And started a new attempt at the same sort of thing in 2007.
As I hadn't touched Oracle since 1996 I stopped keeping track of what they were up to when I retired in 2009 (reckoning that my rule that I would only take on really interesting stuff absolutely ruled out anything to do with Oracle) I don't know what (if anything) happened to the "Oracle Database Appliance" idea after that; but surely it must be dead now, replaced by something cloud-oriented?No, the Oracle database appliance is still alive and well. The latest iteration is called Oracle Exadata and it is a pretty awesome platform. Sun Hardware, Linux OS and Oracle RAC database. Not cheap though 🙁
From what I read, Oracle Database Appliance is just a tightly integrated cluster sharing disk storage, and not a MPP (massively parallel shared) architecture to the same extent that Azure SQL Warehouse is.
I guess I need an Oracle update. Or maybe I should just forget all about Oracle. My last experience with them was pretty awful.
Tom
May 10, 2017 at 1:05 pm
TomThomson - Wednesday, May 10, 2017 12:49 PMEric M Russell - Tuesday, May 9, 2017 11:51 AMloyd.gravitt - Tuesday, May 9, 2017 8:48 AMTomThomson - Tuesday, May 9, 2017 3:55 AMEric M Russell - Thursday, February 9, 2017 11:45 AMGary Varga - Thursday, February 9, 2017 8:09 AMEric M Russell - Thursday, February 9, 2017 6:49 AMAn operating system is to a database server what a buyer/seller agent, attorney, and county clerk are to a real-estate purchase. It's entirely reasonable to ask what the middle men bring to the transaction in terms of financial, administrative, and performance overhead. Like I mentioned before, there is nothing essential that the operating system provides for the database that the database engine couldn't theoretically provide for itself better if it were extended to include that low level functionality.Didn't Oracle try that in the 90s? Didn't they call it something like Raw Iron?
Yes, but it wasn't popular with customers. It makes more sense today in cloud hosted environments.
Although they announced in in 1998, they had gotten nowhere with doing it at the beginning of 2002 when they announced the agreement with Sun to develop it (using chunks of Solaris to do the low level bits); but that "agreement" turned out to be more an agreement to have an agreement, there was still lot of detail to be worked out over who would do what, how it would be sold, and so on. Ellison had claimed agreement on hardware was close with Compaq and IBM and had announced agreements with Dell and HP some time earlier, but none of Dell, HP, and Compaq had anything to say about the project in late 2001/early 2002 (I don't know whether IBM did). The general feeling seemed to be that nearly 4 years after the project was first announced it was never going to fly (no-one believed Oracle could do n OS substitute, not even with Sun's help if they did get that. Shortly afterwards, with no customer interest and hopelessly slipping schedules, it was effectively dead. Some time in the next six years Oracle decided to have a new "database appliance" project, again involving IBM, HP, Dell and Sun and with EMC now in the mix, and when talking about that Oracle's Mendelssohn commented "We tried it with Raw Iron 5 years ago" and "Customers didn't see the imperative" (I guess he thought that 5 years on they would).
So saying the tried it in the 90s is not really accurate - what they did in the 90s was announce they were going to try it, and start some mostly ineffectual negotiations with some possible partners in such a project. They canned it some time in 2002, I think. And started a new attempt at the same sort of thing in 2007.
As I hadn't touched Oracle since 1996 I stopped keeping track of what they were up to when I retired in 2009 (reckoning that my rule that I would only take on really interesting stuff absolutely ruled out anything to do with Oracle) I don't know what (if anything) happened to the "Oracle Database Appliance" idea after that; but surely it must be dead now, replaced by something cloud-oriented?No, the Oracle database appliance is still alive and well. The latest iteration is called Oracle Exadata and it is a pretty awesome platform. Sun Hardware, Linux OS and Oracle RAC database. Not cheap though 🙁
From what I read, Oracle Database Appliance is just a tightly integrated cluster sharing disk storage, and not a MPP (massively parallel shared) architecture to the same extent that Azure SQL Warehouse is.
I guess I need an Oracle update. Or maybe I should just forget all about Oracle. My last experience with them was pretty awful.
Their licensing and business practices are not very much fun, but the base product line, the database itself is top notch. For throughput and concurrency on OLTP it is tough to compete with. SQL Server may perform better on some read only scenarios with small to mid size workloads, but I love Oracle if I have a high transactional load (assuming my employer wants to pay the higher cost).
It always takes me longer to get an Oracle instance set up correctly than it does MSSQL, but once I get it locked in, I have had very little reason to ever complain.
Viewing 6 posts - 31 through 35 (of 35 total)
You must be logged in to reply to this topic. Login to reply