Re: So now Informix has RAC

DA Morgan wrote:
> Serge Rielau wrote:
>> DA Morgan wrote:
>>> Serge Rielau wrote:
>>>> DA Morgan wrote:
>>>>> Obnoxio The Clown wrote:
>>>>>> ibmsoftwarequestions@[EMAIL PROTECTED]
 said:
>>>>>>> On 29 Apr, 22:44, dcrunch...@[EMAIL PROTECTED]
 wrote:
>>>>>>>>
http://www.eweek.com/c/a/Database/IBM-UnCages-Cheetah-2-Data-Server/

>>>>>>>>
>>>>>>>>
>>>>>>>> "IDS (Informix Dynamic Sever) 11.5, code-named "Cheetah 2," 
>>>>>>>> builds on
>>>>>>>> the
>>>>>>>> failover-recovery capabilities of its predecessor by now 
>>>>>>>> sup****ting full
>>>>>>>> read/write transactions across all nodes. The last version of 
>>>>>>>> IDS only
>>>>>>>> sup****ted
>>>>>>>> reads on secondary nodes, the company said."
>>>>>>> Great news from IBM and shows continued commitment to Informix and
>>>>>>> developing Informix technology.
>>>>>>
>>>>>> I don't think it's fair to compare Cheetah 2 to RAC: Cheetah 2 
>>>>>> will scale.
>>>>>> :o)
>>>>> Unlike those 16 node RAC clusters at Amazon.com eh.
>>>> 16? Have the numbers come down or has someone finally realized that 
>>>> it's not about the total number of nodes in ALL clusters that
matters.
>>>> I seem to recall claims of over a hundred...
>>> Au contraire mon ami.
>>>
>>> It depends on what hardware you are running on. I just got back from
>>> Redwood Shores where I was attending a conference jointly sponsored by
>>> Oracle and IBM (http://zseriesoraclesig.org/index.php).
>> I thought the promise was for infinite scaling on low cost hardware.
>> Now you are saying less nodes on big boxes is better.
>> hard to keep up with the messaging here.
>>
>> Aside we were talking about Amazon, not system z.
>> Cheers
>> Serge
> 
> I am not an Oracle employee and my academic standing gives me full
> freedom to say it as it is.
> 
> RAC scalability is well do***ented but, as you should know, is
> dependent upon cache fusion and the private network's latency.
> Obviously there is a difference between GigE and CPU speeds in
> an LPAR just as obviously there is a substantial difference
> between VMware and zVM.
> 
> Lost cost hardware may well be a z10 with a nice green stripe.
> TCO is not the cost of the individual box ... it is the cost of
> the entire implementation both hard and soft. I have a customer
> right now with in excess of 10,000 Oracle servers. The cost of
> power is a significant part of their calculation.
> 
> You and the pSeries and zSeries guys don't talk much do you. <g>
Oh yes, we do. I just wanted to hear the invalidatation of the RAC 
advertisement from your mouth/pen.
Will be interesting to see how you (and Oracle) will manage maintain the 
mainframe and DB2 for system Z being a "legacy platform".

Thank you
Serge
-- 
Serge Rielau
DB2 Solutions Development
IBM Toronto Lab