IDS and grid computing?
A conceptual question rather than a technical fault: can Informix XPS be called "grid computing" (dynamically adding/removing heterogeneous machines) as opposed to clustering? Alexey Sonkin answered that XPS is essentially a shared-nothing cluster, not a grid or resource-provisioning system, since adding a node generally means costly re-fragmentation of tables. Jack Parker countered that adding even a diskless "computational" node to an XPS cluster does improve performance, as work is shipped to it and it scales reasonably; Alexey accepted this as newer behaviour. The thread then drifted into comparisons with Oracle RAC and complaints about Oracle's optimizer, with no formal conclusion beyond the consensus that XPS is a cluster, not a grid.
Auto-generated by DrWatson from the posts below — may be imperfect; read the full thread.
Hi all, would IDS XPS (i don't know much about it) be considered "grid computing" a= s defined by: =20 .=2E.from an article on Grid Computing: =20 "=2E. while grids and clustering both share resources across multiple machi= nes, grids, according to Clabby [expert person], are different because they= allow "distributed resource management of heterogeneous systems." In other= words, with grids you can quickly add and subtract systems-without regard = for location, operating system or normal purpose-as needs dictate. Clusters= are built from the ground up to function as a single pool of compute power= and consequently aren't as flexible. ..." =20 thanks, Norma Jean =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D The information contained in this message may be privileged and confidential and protected from disclosure. If the reader of this message is not the intended recipient, or an employee or agent responsible for delivering this message to the intended recipient, you are hereby notified that any reproduction, dissemination or distribution of this communication is strictly prohibited. If you have received this communication in error, please notify us immediately by replying to the message and deleting it from your computer. Thank you. Tellabs =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D
apologies... I am doing mail from a web access client at this time... don't= know how to un-html it. sorry. ________________________________ From: forum.subscriber@iiug.org on behalf of Sebastian, .... Sent: Tue 2/8/2005 9:38 AM To: ids@iiug.org Subject: IDS and grid computing? [4202]=20 Hi all, would IDS XPS (i don't know much about it) be considered "grid computing" a= =3D s defined by: =3D20 .=3D2E.from an article on Grid Computing: =3D20 "=3D2E. while grids and clustering both share resources across multiple mac= hi=3D nes, grids, according to Clabby [expert person], are different because they= =3D allow "distributed resource management of heterogeneous systems." In other= =3D words, with grids you can quickly add and subtract systems-without regard = =3D for location, operating system or normal purpose-as needs dictate. Clusters= =3D are built from the ground up to function as a single pool of compute power= =3D and consequently aren't as flexible. ..." =3D20 thanks, Norma Jean =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D The information contained in this message may be privileged and confidential and protected from disclosure. If the reader of this message is not the intended recipient, or an employee or agent responsible for delivering this message to the intended recipient, you are hereby notified that any reproduction, dissemination or distribution of this communication is strictly prohibited. If you have received this communication in error, please notify us immediately by replying to the message and deleting it from your computer. Thank you. Tellabs =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D
Norma, XPS is mostly a cluster solution. If You distribute a table across several cluster nodes (horizontal fragmentation), then adding another node either doesn't add any power to queries accessing that table, or one should re-distribute it (very expensive operation) At the same time, after adding nodes to the cluster, one can use them for storing data for another subsystem... Anyway, it is not 'grid'. It is not 'resource provisioning' ------------------------------------------ Alexey Sonkin > -----Original Message----- > From: forum.subscriber@iiug.org [mailto:forum.subscriber@iiug.org] On > Behalf Of Sebastian, .... > > Hi all, > would IDS XPS (i don't know much about it) be considered "grid computing" > as defined by: > with grids you can quickly add and subtract systems-without regard > for location, operating system or normal purpose-as needs dictate. > Clusters are built from the ground up to function as a single pool of [AS] > compute power > and consequently aren't as flexible. ..." > Norma Jean >
Interesting. Would you consider 'grid' computing, 'resource provisioning' = to predominantly take place at the o/s level? like Sun's N1 technology or = Solaris Resource Manager.... technology mainly driven by hardware vendors? = Or software level like Oracle 10g (don't know much about it beyond the hyp= e). ________________________________ From: Alexey Sonkin [mailto:alexeis@grandvirtual.com] Sent: Tue 2/8/2005 1:13 PM To: Sebastian, Norma J. Cc: ids@iiug.org Subject: RE: IDS and grid computing? [4202]=20 Norma, XPS is mostly a cluster solution. If You distribute a table across several cluster nodes (horizontal fragmentation), then adding another node either doesn't add any power to queries accessing that table, or one should re-distribute it (very expensive operation) At the same time, after adding nodes to the cluster, one can use them for storing data for another subsystem... Anyway, it is not 'grid'. It is not 'resource provisioning' ------------------------------------------ Alexey Sonkin > -----Original Message----- > From: forum.subscriber@iiug.org [mailto:forum.subscriber@iiug.org] On > Behalf Of Sebastian, .... > > Hi all, > would IDS XPS (i don't know much about it) be considered "grid computing" > as defined by: > with grids you can quickly add and subtract systems-without regard > for location, operating system or normal purpose-as needs dictate. > Clusters are built from the ground up to function as a single pool of [AS] > compute power > and consequently aren't as flexible. ..." > Norma Jean > =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D The information contained in this message may be privileged and confidential and protected from disclosure. If the reader of this message is not the intended recipient, or an employee or agent responsible for delivering this message to the intended recipient, you are hereby notified that any reproduction, dissemination or distribution of this communication is strictly prohibited. If you have received this communication in error, please notify us immediately by replying to the message and deleting it from your computer. Thank you. Tellabs =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D
You might say that, but it's been proven that adding another node to an XPS cluster (even without disk) does in fact improve performance. Work is shipped to and performed by the additional 'computational' node. It even scales decently. j. ----- Original Message ----- From: "Alexey Sonkin" <alexeis@grandvirtual.com> To: <ids@iiug.org> Sent: Tuesday, February 08, 2005 3:19 PM Subject: RE: IDS and grid computing? [4205] > Norma, > > XPS is mostly a cluster solution. > > If You distribute a table across several cluster nodes > (horizontal fragmentation), then adding another node either > doesn't add any power to queries accessing that table, or one > should re-distribute it (very expensive operation) > > At the same time, after adding nodes to the cluster, one can > use them for storing data for another subsystem... > > Anyway, it is not 'grid'. It is not 'resource provisioning' > > > ------------------------------------------ > Alexey Sonkin > > > > > -----Original Message----- > > From: forum.subscriber@iiug.org [mailto:forum.subscriber@iiug.org] On > > Behalf Of Sebastian, .... > > > > Hi all, > > would IDS XPS (i don't know much about it) be considered "grid > computing" > > as defined by: > > with grids you can quickly add and subtract systems-without regard > > for location, operating system or normal purpose-as needs dictate. > > Clusters are built from the ground up to function as a single pool of > [AS] > compute power > > and consequently aren't as flexible. ..." > > Norma Jean > > > >
Jack, Nice to know, that XPS is now able to ship data to a node, that doesn't own any fragment... It was 5 years ago, when I was dealing with XPS the last time The technology is evolving.. As for Oracle RAC/"grid", it is not clear to me whether O is able to distribute a single query across several computers in a cluster. AFAIK Oracle RAC is mostly for OLTP, and comparison with XPS is not an 'apples to apples' comparison. -Alexey > -----Original Message----- > From: Jack Parker [mailto:vze2qjg5@verizon.net] > Sent: Tuesday, February 08, 2005 7:59 PM > To: ids@iiug.org; Alexey Sonkin > Subject: Re: IDS and grid computing? [4205] > > > You might say that, but it's been proven that adding another node to an > XPS > cluster (even without disk) does in fact improve performance. Work is > shipped to and performed by the additional 'computational' node. It even > scales decently. > > j. > ----- Original Message ----- > From: "Alexey Sonkin" <alexeis@grandvirtual.com> > To: <ids@iiug.org> > Sent: Tuesday, February 08, 2005 3:19 PM > Subject: RE: IDS and grid computing? [4205] > > > > Norma, > > > > XPS is mostly a cluster solution. > > > > If You distribute a table across several cluster nodes > > (horizontal fragmentation), then adding another node either > > doesn't add any power to queries accessing that table, or one > > should re-distribute it (very expensive operation) > > > > At the same time, after adding nodes to the cluster, one can > > use them for storing data for another subsystem... > > > > Anyway, it is not 'grid'. It is not 'resource provisioning' > > > > > > ------------------------------------------ > > Alexey Sonkin > > > > > > > > > -----Original Message----- > > > From: forum.subscriber@iiug.org [mailto:forum.subscriber@iiug.org] On > > > Behalf Of Sebastian, .... > > > > > > Hi all, > > > would IDS XPS (i don't know much about it) be considered "grid > > computing" > > > as defined by: > > > with grids you can quickly add and subtract systems-without regard > > > for location, operating system or normal purpose-as needs dictate. > > > Clusters are built from the ground up to function as a single pool of > > [AS] > compute power > > > and consequently aren't as flexible. ..." > > > Norma Jean > > > > > > > >
Ah, it's been a while for me and XPS as well, although I have tried to keep abreast. I've been doing quite a bit with Orrible these days - the thing is a piece of sh*t. Cute tools, great language features, but I wouldn't want to run a database over 50GB on an oracle engine. And to think, that was the competition. It can't even handle a local query properly without hints. Just the other day I was astounded to see it apply the filters to the query tree AFTER all of the joins - yeesh. j. ----- Original Message ----- From: "Alexey Sonkin" <alexeis@grandvirtual.com> To: <ids@iiug.org> Sent: Wednesday, February 09, 2005 10:52 PM Subject: RE: IDS and grid computing? [4222] > Jack, > > Nice to know, that XPS is now able to ship data to a node, that > doesn't own any fragment... It was 5 years ago, when I was dealing with > XPS > the last time > > The technology is evolving.. > > As for Oracle RAC/"grid", it is not clear to me whether O is able > to distribute a single query across several computers in a cluster. > AFAIK Oracle RAC is mostly for OLTP, and comparison with XPS is not an > 'apples to apples' comparison. > > -Alexey > > > > -----Original Message----- > > From: Jack Parker [mailto:vze2qjg5@verizon.net] > > Sent: Tuesday, February 08, 2005 7:59 PM > > To: ids@iiug.org; Alexey Sonkin > > Subject: Re: IDS and grid computing? [4205] > > > > > > You might say that, but it's been proven that adding another node to > an > > XPS > > cluster (even without disk) does in fact improve performance. Work is > > shipped to and performed by the additional 'computational' node. It > even > > scales decently. > > > > j. > > ----- Original Message ----- > > From: "Alexey Sonkin" <alexeis@grandvirtual.com> > > To: <ids@iiug.org> > > Sent: Tuesday, February 08, 2005 3:19 PM > > Subject: RE: IDS and grid computing? [4205] > > > > > > > Norma, > > > > > > XPS is mostly a cluster solution. > > > > > > If You distribute a table across several cluster nodes > > > (horizontal fragmentation), then adding another node either > > > doesn't add any power to queries accessing that table, or one > > > should re-distribute it (very expensive operation) > > > > > > At the same time, after adding nodes to the cluster, one can > > > use them for storing data for another subsystem... > > > > > > Anyway, it is not 'grid'. It is not 'resource provisioning' > > > > > > > > > ------------------------------------------ > > > Alexey Sonkin > > > > > > > > > > > > > -----Original Message----- > > > > From: forum.subscriber@iiug.org [mailto:forum.subscriber@iiug.org] > On > > > > Behalf Of Sebastian, .... > > > > > > > > Hi all, > > > > would IDS XPS (i don't know much about it) be considered "grid > > > computing" > > > > as defined by: > > > > with grids you can quickly add and subtract systems-without regard > > > > for location, operating system or normal purpose-as needs dictate. > > > > Clusters are built from the ground up to function as a single pool > of > > > [AS] > compute power > > > > and consequently aren't as flexible. ..." > > > > Norma Jean > > > > > > > > > > > > > > > >
Ummm... <sarcastic> Maybe an _update statistics_ is in order ;) </sarcastic> -----Original Message----- From: "Jack Parker" <vze2qjg5@verizon.net> To: ids@iiug.org Date: Thu, 10 Feb 2005 01:06:57 -0500 (EST) Subject: Re: IDS and grid computing? [4223] Ah, it's been a while for me and XPS as well, although I have tried to keep abreast. I've been doing quite a bit with Orrible these days - the thing is a piece of sh*t. Cute tools, great language features, but I wouldn't want to run a database over 50GB on an oracle engine. And to think, that was the competition. It can't even handle a local query properly without hints. Just the other day I was astounded to see it apply the filters to the query tree AFTER all of the joins - yeesh. j. ----- Original Message ----- From: "Alexey Sonkin" <alexeis@grandvirtual.com> To: <ids@iiug.org> Sent: Wednesday, February 09, 2005 10:52 PM Subject: RE: IDS and grid computing? [4222] > Jack, > > Nice to know, that XPS is now able to ship data to a node, that > doesn't own any fragment... It was 5 years ago, when I was dealing with > XPS > the last time > > The technology is evolving.. > > As for Oracle RAC/"grid", it is not clear to me whether O is able > to distribute a single query across several computers in a cluster. > AFAIK Oracle RAC is mostly for OLTP, and comparison with XPS is not an > 'apples to apples' comparison. > > -Alexey > > > > -----Original Message----- > > From: Jack Parker [mailto:vze2qjg5@verizon.net] > > Sent: Tuesday, February 08, 2005 7:59 PM > > To: ids@iiug.org; Alexey Sonkin > > Subject: Re: IDS and grid computing? [4205] > > > > > > You might say that, but it's been proven that adding another node to > an > > XPS > > cluster (even without disk) does in fact improve performance. Work is > > shipped to and performed by the additional 'computational' node. It > even > > scales decently. > > > > j. > > ----- Original Message ----- > > From: "Alexey Sonkin" <alexeis@grandvirtual.com> > > To: <ids@iiug.org> > > Sent: Tuesday, February 08, 2005 3:19 PM > > Subject: RE: IDS and grid computing? [4205] > > > > > > > Norma, > > > > > > XPS is mostly a cluster solution. > > > > > > If You distribute a table across several cluster nodes > > > (horizontal fragmentation), then adding another node either > > > doesn't add any power to queries accessing that table, or one > > > should re-distribute it (very expensive operation) > > > > > > At the same time, after adding nodes to the cluster, one can > > > use them for storing data for another subsystem... > > > > > > Anyway, it is not 'grid'. It is not 'resource provisioning' > > > > > > > > > ------------------------------------------ > > > Alexey Sonkin > > > > > > > > > > > > > -----Original Message----- > > > > From: forum.subscriber@iiug.org [mailto:forum.subscriber@iiug.org] > On > > > > Behalf Of Sebastian, .... > > > > > > > > Hi all, > > > > would IDS XPS (i don't know much about it) be considered "grid > > > computing" > > > > as defined by: > > > > with grids you can quickly add and subtract systems-without regard > > > > for location, operating system or normal purpose-as needs dictate. > > > > Clusters are built from the ground up to function as a single pool > of > > > [AS] > compute power > > > > and consequently aren't as flexible. ..." > > > > Norma Jean > > > > > > > > > > > > > > > > Jean Sagi jeansagi@myrealbox.com jeansagi@yahoo.com