yipeee!

[Niall Litchfield]

Re: yipeee!

"Serge Rielau" <srielau@ca.e ye-be-em.com> wrote in message
news:bvrqg4$2l$ 1@hanover.torol ab.ibm.com...[color=blue]
> Mike,
>
> I get the feeling this is not a high concurrency and high throughput
> environment? So, no matter what DBMS you end up with, I'd be hesitant to
> let myself being talked into any high-end, pay extra, features....
>
> Cheers
> Serge
> --
> Serge Rielau
> DB2 SQL Compiler Development
> IBM Toronto Lab[/color]

I agree - there is what might be an interesting development for Mike with
regards to 10g, namely that standard edition Oracle will come with RAC
essentially for free. This would mean that they could buy a (relatively)
cheap edition of Oracle with the possibility of doing the 2 node failover
later. I'd agree with Daniel that 2 2cpu Intel boxes might make more sense
for this than a risc box.

I'd also suggest that you take a look at what expertise you have in house
and try and match that.


--
Niall Litchfield
Oracle DBA
Audit Commission UK

[Serge Rielau]

Re: yipeee!

Well, AFAIK amazon.com is running Teradata in their warehouse.
--
Serge Rielau
DB2 SQL Compiler Development
IBM Toronto Lab

[Serge Rielau]

Re: yipeee!

You would a slightly different configuration, that is correct.
An extra node for standby. If you take a look at the disclosured you
will find that an extra node will not disturb the total cost by much.

Cheers
Serge
--
Serge Rielau
DB2 SQL Compiler Development
IBM Toronto Lab

[Mike]

Re: yipeee!

In article <bvrqg4$2l$1@ha nover.torolab.i bm.com>, Serge Rielau wrote:[color=blue]
> Mike,
>
> I get the feeling this is not a high concurrency and high throughput
> environment? So, no matter what DBMS you end up with, I'd be hesitant to
> let myself being talked into any high-end, pay extra, features....
>
> Cheers
> Serge[/color]

It's not that it is a high-throughput, but rather that the company
has been built around the application that has grown for 25 years
and now the company is operating on three continents. The app
cannot be down.

Mike

[Mike]

Re: yipeee!

In article <bvrv1q$mr$1@ha nover.torolab.i bm.com>, Serge Rielau wrote:[color=blue]
> Actually if we go back that far I don't think the topic was shared
> nothing vs shared disk to begin with.. but for the sake of peace:
>
> Daniel Morgan wrote:
>[color=green]
>> Serge Rielau wrote:
>>[color=darkred]
>>> The topic was that you alleged DB2 cannot recover from a failing node.[/color]
>>
>>
>> Not as I recall. The topic was someone considering DB2 or Oracle. And
>> considering shared nothing vs shared everything. The architectures are
>> different and lead to different outcomes.
>>
>> Can you fail-over with shared nothing if you lose one or more nodes?[/color]
> Yes! And I'm not going to repeat myself. Read the thread.[color=green]
>>
>> Can you add or remove nodes without some code rewrite with shared
>> nothing (small but some)?[/color]
> Given that this is a brand new app: Yes, written for shared nothing.
> You start with 2 nodes you scale to 999[color=green]
>> Can you add or remove nodes without repartitioning the data with shared
>> nothing?[/color]
> Yes, to offload the system burden. Repartitioning would be advised a
> some point to retune the system the extra node stays put.
>[color=green]
>> Can you add or remove nodes without a shutdown with shared nothing?[/color]
> I honestly don't know the answer to that one. Never tried it.
>
> Now let's close this of, and get back to OP's questions:
>
> OP states he needs HA. OP so far shows no indication that clustering for
> sclability is needed. OP is putting up the requirements not the group :-)
>
> I would propose two boxes. Clusterware, DB2 licensed for the CPU's on
> the one box and 1 CPU for the other (it's in standby).
> I would NOT, repeat: NOT recommend DPF because there is NO indication
> that it's needed.
>
> Now that will come out at price X
>
> Now one can compare that with whatever the Oracle crowd would propose
> which will be price Y.
>
> That plus whatever other input OP has (skillset, TCO (let's not even get
> into that one), company strategy (special discounts?), ...) will yield
> the database to choose from.
>
> Cheers
> Serge
>[/color]

Currently I have one s390 with a home-grown app and the data in vsam.
What is required (either db2 or oracle) such that the app and data
can be moved from the s390 to rs/6000 with greater availability, as
fast of retrieval, etc?

Mike

[Mark A]

Re: yipeee!

"Serge Rielau" <srielau@ca.e ye-be-em.com> wrote in message
news:bvtd8b$a0f $1@hanover.toro lab.ibm.com...[color=blue]
> Well, AFAIK amazon.com is running Teradata in their warehouse.
> --
> Serge Rielau
> DB2 SQL Compiler Development
> IBM Toronto Lab[/color]

Serge, Daniel does not understand the difference between a data warehouse
application (which benefits from share nothing, or quasi-share nothing) and
an OLTP application (which benefits from share everything with fallover
capability). It's really scary that he is getting paid to teach people about
database.

[Serge Rielau]

Re: yipeee!

OK, so you need HA. What else?
Any specific SQL Features?
What's your companies strategie for App development? Java, .Net, both?
Any OS/Hardware preference?
What's you companies DB skillset in house?

Cheers
Serge
--
Serge Rielau
DB2 SQL Compiler Development
IBM Toronto Lab

[Mark A]

Re: yipeee!

"Serge Rielau" <srielau@ca.e ye-be-em.com> wrote in message
news:bvtfmn$a5e $1@hanover.toro lab.ibm.com...[color=blue]
> OK, so you need HA. What else?
> Any specific SQL Features?
> What's your companies strategie for App development? Java, .Net, both?
> Any OS/Hardware preference?
> What's you companies DB skillset in house?
>
> Cheers
> Serge
> --
> Serge Rielau
> DB2 SQL Compiler Development
> IBM Toronto Lab[/color]

I think they want to move to CICS on RS/6000 to save money on hardware
costs. I would bet money they don't have the time or bucks to re-write the
application in a totally different technology (Java, Net, etc).

[Niall Litchfield]

Re: yipeee!

"Mark A" <ma@switchboard .net> wrote in message
news:hlrUb.4$Ki 4.11047@news.us west.net...[color=blue]
> "Serge Rielau" <srielau@ca.e ye-be-em.com> wrote in message
> news:bvtd8b$a0f $1@hanover.toro lab.ibm.com...[color=green]
> > Well, AFAIK amazon.com is running Teradata in their warehouse.
> > --
> > Serge Rielau
> > DB2 SQL Compiler Development
> > IBM Toronto Lab[/color]
>
> Serge, Daniel does not understand the difference between a data warehouse
> application (which benefits from share nothing, or quasi-share nothing)[/color]
and[color=blue]
> an OLTP application (which benefits from share everything with fallover
> capability). It's really scary that he is getting paid to teach people[/color]
about[color=blue]
> database.[/color]

So now that we have established that the app is transactional and requires
HA, and setting aside the personal remarks which way would you jump?


--
Niall Litchfield
Oracle DBA
Audit Commission UK

[Mark A]

Re: yipeee!

> > Serge, Daniel does not understand the difference between a data
warehouse[color=blue][color=green]
> > application (which benefits from share nothing, or quasi-share nothing)[/color]
> and[color=green]
> > an OLTP application (which benefits from share everything with fallover
> > capability). It's really scary that he is getting paid to teach people[/color]
> about[color=green]
> > database.[/color]
>
> So now that we have established that the app is transactional and requires
> HA, and setting aside the personal remarks which way would you jump?
>
>
> --
> Niall Litchfield
> Oracle DBA
> Audit Commission UK
>[/color]
As I mentioned on another post, I doubt they want to pay for a complete
rewrite of the mainframe application. I think they want to port to CICS on
RS6000 to save money, and that means they probably can't afford a completely
new custom application. They might be able to use a VSAM emulator on
CICS/RS6000 or convert the application to DB2 running under CICS on RS6000.
I don't know if Oracle works with CICS on RS/6000.

I would probably stay on the mainframe until they find a packaged
application that runs on a distributed platform. This may require that the
company change parts of their business process to fit the package, but that
is what I would do. If they don't want a new application, I would stay on
the mainframe.

The odds of a medium scale application development project ever being
completed with code in production is less than 50%. Of the remaining 50%
that make it into production, the vast majority of those projects will be
over way budget and/or seriously late to completion.

In a situation like this, picking the DBMS product is not the first decision
to be made.

[Noons]

Re: yipeee!

"Mark A" <ma@switchboard .net> wrote in message
news:OhsUb.15$K i4.17007@news.u swest.net...
[color=blue]
> The odds of a medium scale application development project ever being
> completed with code in production is less than 50%.[/color]

they decrease to 0% if the developer is IBM/GSA...
[color=blue]
> Of the remaining 50%
> that make it into production, the vast majority of those projects will be
> over way budget and/or seriously late to completion.[/color]

You forgot: dumbed-down so that SOMETHING can be delivered to the client...

--
Cheers
Nuno Souto
wizofoz2k@yahoo .com.au.nospam

[Buck Nuggets]

Re: yipeee!

Daniel Morgan <damorgan@x.was hington.edu> wrote in message news:<107594886 8.103447@yasure >...
[color=blue]
> As has been frequently stated in the DB2 usenet group and elsewhere ...
> the TPC benchmarks favor shared-nothing. Pull the plug on one of those
> shared nothing nodes (hardware failures do happen in the real world) and
> see what happens to performance.
>
> My point being that if I made one change to the benchmark ... say added
> the following ... "one hour into the test pull the plug on one node and
> complete the job" ... shared nothing and federated databases wouldn't
> even be able to compete.[/color]

Kind of neat how as soon as product a has a new feature its apologists
declare it the essential and distinguishing feature in the
marketplace..

Now, if it works as smoothly as you describe - that's great, and I'll
look forward to using it. On the other hand, in my experience with
MPP databases (primarily Informix on AIX using SP2), I could go a year
without having to reboot any nodes. And when I did, it was often due
to a configuration change to the frame - so I had to take all of them
down anyway.

So, nice feature, but hardly essential for most applications.

[Serge Rielau]

Re: yipeee!

My DB2 "offer" would be DB2 without DPF on two AIX boxes (OP wants AIX
it seems). The seoncd box licenced as idle standby only (1 CPU).
With clusterware to handle the failover.
This is under the assumption that a rewrite of the app to a relational
DBMS is intended.
There is not enough information to home in on which edition or box-size
to home in to.

It seems Mark A. believes CICS would be less invasive. I'm not familiar
with either VSE or CICS so I keep my mouth shut.

Let's presume 100% scalability for RAC (if you want to use it) for the
sake of math (and to not start another flame war) and similar resource
requirements (AIX, RAM/box, comparable disk overall).

Your turn
Serge
--
Serge Rielau
DB2 SQL Compiler Development
IBM Toronto Lab

[Mark A]

Re: yipeee!

"Serge Rielau" <srielau@ca.e ye-be-em.com> wrote in message
news:bvtjk7$bvp $1@hanover.toro lab.ibm.com...[color=blue]
> My DB2 "offer" would be DB2 without DPF on two AIX boxes (OP wants AIX
> it seems). The seoncd box licenced as idle standby only (1 CPU).
> With clusterware to handle the failover.
> This is under the assumption that a rewrite of the app to a relational
> DBMS is intended.
> There is not enough information to home in on which edition or box-size
> to home in to.
>
> It seems Mark A. believes CICS would be less invasive. I'm not familiar
> with either VSE or CICS so I keep my mouth shut.
>
> Let's presume 100% scalability for RAC (if you want to use it) for the
> sake of math (and to not start another flame war) and similar resource
> requirements (AIX, RAM/box, comparable disk overall).
>
> Your turn
> Serge
> --
> Serge Rielau
> DB2 SQL Compiler Development
> IBM Toronto Lab[/color]

I think that what they want to do is change the COBOL application data
access from VSAM to DB2, running under CICS transaction monitor on RS6000.
This could be done without a complete redesign of the application, although
database access in each program would have to be changed by hand from VSAM
to DB2 and thoroughly tested. VSAM KSDS is an inverted list database (not
hierarchical like IMS) so there could be a one-to-one mapping of existing
VSAM datasets to DB2 tables.

But again, I think their main motivation is to save money, and the
conversion certainly would not be cheap, although much less expensive than a
complete re-write.

The hardware redundancy issues could be worked out in a number of ways, as
already mentioned.

[Niall Litchfield]

Re: yipeee!

"Serge Rielau" <srielau@ca.e ye-be-em.com> wrote in message
news:bvtjk7$bvp $1@hanover.toro lab.ibm.com...[color=blue]
> My DB2 "offer" would be DB2 without DPF on two AIX boxes (OP wants AIX
> it seems). The seoncd box licenced as idle standby only (1 CPU).
> With clusterware to handle the failover.
> This is under the assumption that a rewrite of the app to a relational
> DBMS is intended.
> There is not enough information to home in on which edition or box-size
> to home in to.
>
> It seems Mark A. believes CICS would be less invasive. I'm not familiar
> with either VSE or CICS so I keep my mouth shut.[/color]

ditto.

[color=blue]
>
> Let's presume 100% scalability for RAC (if you want to use it) for the
> sake of math (and to not start another flame war) and similar resource
> requirements (AIX, RAM/box, comparable disk overall).
>
> Your turn
> Serge[/color]

:(

I think I would go with one of

two AIX boxes one running DB with Oracle DataGuard to handle standby to the
second pretty much as per your solution.
n box implementation of 9i RAC with transparent application failover.


On the other hand the more we learn about the app the more business critical
it seems and so the more conservative (read why move it at all?) I become.


--
Niall Litchfield
Oracle DBA
Audit Commission UK