RE: thank you IBM
A discussion thread (not a support question) about IBM's "Project Arrowhead" convergence of Informix IDS and DB2. Participants debate whether the two engines will ever merge: one poster argues they can't, citing DB2's process-based vs. Informix's thread-based architecture and DB2's lack of checkpoints, while others counter that features are flowing both ways, that such internals are implementation details that can be changed, and that the real obstacle is application-level/SQL differences (e.g. non-ANSI Informix databases). An IBM engineer notes DB2 uses trickle flush rather than checkpoints by design. No resolution or conclusion is reached.
Auto-generated by DrWatson from the posts below — may be imperfect; read the full thread.
Interesting you that you say.... >"Just remember that Project Arrowhead will bring Informix and DB2 closer, so that > eventually the 2 will be 1, so for Informix DBA's the change will be gradual." I was in a database futures meeting here the other day and I brought up the whole IFMX to DB2 Arrowhead thingy.... Senior management looked at me and asked, "How much longer does IBM need?" I dunno. -----Original Message----- From: Dirk Moolman [mailto:DirkM@mxgroup.co.za] Sent: Friday, June 11, 2004 9:21 AM To: rkusenet; informix-list@iiug.org Subject: RE: thank you IBM -----Original Message----- From: owner-informix-list@iiug.org [mailto:owner-informix-list@iiug.org] On Behalf Of rkusenet Sent: Wednesday, June 09, 2004 9:55 PM To: informix-list@iiug.org Subject: Re: thank you IBM "nobody" <nobody@devnull.org> wrote > I'm not sure exactly the point you are trying to make. > You'll see leakage from the Informix customer base for several > reasons.... > 1) 3rd Party Aps not ported to Informix in future releases. > 2) Coporations changing their IT direction > 3) Customer want to consolidate their vendors and platforms. >the thanks is for excellent marketing they have done to 'promote' >Informix and dispelling any FUD to existing user base and ISVs. >I have been told by an IBM employee that more than 50% of ex-Informix >employees are now working on Db2. >Yeah they are going ahead with 9.50. Who cares? What is the point in >upgrading when eventually they have to migrate to some other DB. Just remember that Project Arrowhead will bring Informix and DB2 closer, so that eventually the 2 will be 1, so for Informix DBA's the change will be gradual. DB2 has already built - into their latest version - HA (replication), taken from Informix. A couple of Informix features have already been ported to DB2. I choose to be optimistic - technology is still improving. ________________________________ << ella for Spam Control >> has removed 1754 Spam messages and set aside 712 Newsletters for me You can use it too - and it's FREE! www.ellaforspam.com sending to informix-list sending to informix-list
"Hamilton, Jerry" <hamiltoj@fleishman.com> wrote \\ > I was in a database futures meeting here the other day and I brought up the > whole IFMX to DB2 Arrowhead thingy.... Senior management looked at me and > asked, "How much longer does IBM need?" I don't believe there is any Informix to DB2 convergence. What will happen is that lot of Informix features will go to DB2. Now that is different from merging. DB2 is process based architecture. Informix is thread based. Db2 has no concept of checkpoint. A page can not be modified more than once, before it is written to disk. How smart :-). There are too many differences between them to merge.
"rkusenet" <rkusenet@sympatico.ca> wrote in message news:2iu0r7Fr1n5fU1@uni-berlin.de... > > "Hamilton, Jerry" <hamiltoj@fleishman.com> wrote \\ > > > I was in a database futures meeting here the other day and I brought up the > > whole IFMX to DB2 Arrowhead thingy.... Senior management looked at me and > > asked, "How much longer does IBM need?" > > I don't believe there is any Informix to DB2 convergence. What will happen is > that lot of Informix features will go to DB2. Now that is different from merging. > > DB2 is process based architecture. Informix is thread based. Db2 has no concept > of checkpoint. > A page can not be modified more than once, before it is written > to disk. How smart :-). Really...I must ask at the event I am going to next week! > > There are too many differences between them to merge. > >
rkusenet wrote: > "Hamilton, Jerry" <hamiltoj@fleishman.com> wrote \\ > >> I was in a database futures meeting here the other day and I brought >> up the whole IFMX to DB2 Arrowhead thingy.... Senior management >> looked at me and asked, "How much longer does IBM need?" > > I don't believe there is any Informix to DB2 convergence. What will > happen is that lot of Informix features will go to DB2. Now that is > different from merging. And a whole heap of DB2 features will go into IDS. Further, DB2 is embarked on a major project to add a lot of simplicity that IDS contains. Check out a very large posting I made about it maybe one or two months ago. It contains most of the info that they are making public about the technical directions. > Db2 has > no concept of checkpoint. A page can not be modified more than once, > before it is written to disk. How smart :-). This is only an implementation detail. It can be changed given enough will and programmer hours. > There are too many differences between them to merge. Perhaps at the source level, but even one feature will be made from a heirarchy of modules, and at the top-most level, where we get to experience the engine, the same feature can be implemented using a variety of underlying algorithms. Programming 101.
"rkusenet" <rkusenet@sympatico.ca> wrote in message news:2iu0r7Fr1n5fU1@uni-berlin.de... > > "Hamilton, Jerry" <hamiltoj@fleishman.com> wrote \\ > > > DB2 is process based architecture. Informix is thread based. Db2 has no concept > of checkpoint. A page can not be modified more than once, before it is written > to disk. How smart :-). Are you sure about that? It was my understanding that DB2 didn't use checkpoints because it used a trickle flush technique which worked similar to the way that the IDS fuzzy checkpoints work. I know that the way that data pages and index pages are logged and structured on DB2 are such that it would be fairly easy to not need to do periodic checkpoints, use a physical log, or go through physical recovery. Also, I don't really see why we would want to introduce the checkpoint concept in DB2. Especially since one of the more common questions I'm asked about IDS is how to avoid long checkpoint times. ;-0 M.P. > > There are too many differences between them to merge. > >
rkusenet wrote: > I don't believe there is any Informix to DB2 convergence. What will happen is > that lot of Informix features will go to DB2. Now that is different from merging. convergence is not equal to merging. > DB2 is process based architecture. Informix is thread based. Db2 has no concept > of checkpoint. A page can not be modified more than once, before it is written > to disk. How smart :-). > > There are too many differences between them to merge. I would use neither of them as a handicap because they are internal. Only DBA's worry about that one. What is a lot more important than internal details are externals. That's where the 100kloc app breaks. And that's where Informix is really interesting due to the large population of non-ANSI DBs. Some of those non-ANSI properties are "merely" extentions and can be made to fit into DB2, others can simply not be married with ANSI. Quite a conundrum. Cheers Serge -- Serge Rielau DB2 SQL Compiler Development IBM Toronto Lab