Oracle FAQ Your Portal to the Oracle Knowledge Grid

Home -> Community -> Usenet -> c.d.o.server -> Re: Database or store to handle 30 Mb/sec and 40,000 inserts/sec

Re: Database or store to handle 30 Mb/sec and 40,000 inserts/sec

From: Double Echo <>
Date: Thu, 23 Feb 2006 22:06:51 -0500
Message-ID: <dLuLf.4824$>

Noons wrote:
> Double Echo wrote:

>> I couldn't agree more Daniel.  My own experience has shown scalability
>> to be a number one problem with SQL-Server.  But the new features
>> coming out in SS-2205 should not be ignored.  There are improvements,
>> and like other people I am waiting to see how it shakes out.  You and
>> I both know SQL-Server is not going away, if anything it is increasing
>> its footprint by leaps and bounds.  As a technologist, I am practically
>> required to understand the product in order to make sure I understand
>> it's usage in an environment, simply because of its pervasiveness,
>> something even Oracle can't claim as much of--there will probably
>> alway be orders of magnitude of MS-SQL compared to Oracle.  That is
>> largely in part because of marketing and branding of these products;
>> Oracle pushing down from big shops into small shops, and MS-SQL pushing
>> up from small shops into big shops.  The collision, as we have seen
>> in this thread is that thermocline that produces flame wars.

> Yup, very well said. What I'm seeing here as well.
>> Unfortunately businesses are doing just that, putting their mission
>> critical data in SQL-Server, but you already knew that.

> Indeed they are. I still think it's a major strategic error as well
> as tactical suicide. Diferentiation between businesses starts at
> how well each takes advantage of the information it needs to be
> successfull. Make processing that information into a single commodity
> style baseline and you got a recipe for very good bottom lines for a
> while and no competitiveness whatsoever in the medium term. Some
> businesses can afford that. Others won't. IME, size-related.
> But we'll see how things will pan out.

Well you've hit the nail on the head, but you don't need me to validate that. I worked with a firm that was using SQL-Server and they hit the wall on performance. The executive question was, "do we need to call Oracle". It is comical in a certain sense, but here is where the branding takes over and minds set to quiescent mode to consider the options. Talent pool comes to mind, time to bring forth the consultants. But the bottom line is, what really _are_ your options? In a world with little time to think, the obvious choices come to mind ( pick the easy choices ). But these choices are not religious, they are business. If your 'solution' cannot bring in a 'complete solution' you lose the executive set on the first pass, and quite frankly it it brutal, it is life or death. I have seen presentations die within 5 minutes of starting, only to have the presenter escorted out the side door. Laughable as it is, it reminds me of Gordon Gecko in Wall Street. Performance is in the minds of the beholder, and as management moves from one level to the next, the landscape changes. As technologists, we are forced to take sides, but as managers, it's a different ball game, it's all about performance of the business, and if that involves buying total shyte so be it, it fulfills a broader goal of simply increasing the bottom line. It means really in a broader sense that technologists that talk business-bottom-line will curry favor every time with management, whatever the brand. I think Microsoft is depending on application momentum over anything 'technically superior'.

> Hardware is not as critical as it used to be, neither is the
> OS. Things are different nowadays and maybe there is a business
> case for it being so.

Seems like a 6-month cycle these days is what a year cycle used to be.

  The other thing that folks should always be
> aware
> of is that Windows ES is NOT the same as vanilla desktop Windows:
> the two are only common in name. ES is a well rounded OS,
> with some limitations for sure but quite usable for general purpose
> computing. Cheaper than *n*x? I don't think so, but there will
> always be an accountant that will disagree.

I have worked with Windows up to 16 CPUs on Unisys ES7000, Advanced Data Center Server and for some strange reason, it runs as fast on my laptop, what's with that! :-) The ES7000 was supposed to be the UNIX killer, but it is $360,0000.00 of crap. The same money could be better spent on a legion of 4-CPU systems, using Linux, with some kind of clustering--or in today's world a legion of 2-cpu systems or a blade server/system.

The bottom line is to simply stay in touch with management, and meet expectations. I don't think MS-SQL-Server is beyond AAA ball just yet but who cares what I think, we all know MS will either be accepted into the major leagues or they will simply create their own ball club--as was suggested, everyone else does their homework for them. Smart people will simply work to meet management goals plain and simple, product or brand, it doesn't matter. Religious wars aside, it's all about the bottom line.


>> a dependency on a db.  Larry will be the first to tell us applications
>> drive databases, not the other way around.  And it is instructive for
>> all of us to think first in terms of supporting businesses with good
>> sound logic skills, and accept the fact that management does not always
>> apply the same logic we do, if at all.  Your skills should transcend
>> product, plain and simple.

> Well said.
> (applause)
Received on Thu Feb 23 2006 - 21:06:51 CST

Original text of this message