Oracle FAQ Your Portal to the Oracle Knowledge Grid
HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US
 

Home -> Community -> Usenet -> c.d.o.tools -> Re: Microsoft destroys TPC-C records!

Re: Microsoft destroys TPC-C records!

From: Michael D. Long <lead_dog_at_bellsouth.net>
Date: 2000/04/08
Message-ID: <KIMH4.694$Yo4.16009@news1.mia>#1/1

I took the time to read the article, and I found a real point of humor in a quote by Charles Fitzgerald (director of business development for Microsoft's Developer Group). On the topic of scalability, Mr. Fitzgerald says "There was a great study done by [DocuLabs], for example, in July...".

Apparently neither the development managers at Microsoft nor the staff at Doculabs have the technical ability to properly plan a benchmark and assess the results.

The major issue I take with the Doculabs benchmark is that the competitive products utilized transactions, while the Micrsoft VC++ based solution relied on implicit commits. I attended a conference where Lon Fisher, the chief architect of the team that implemented the code used in the benchmark, presented the techniques used to optimize performance. Some of them border on smoke and mirrors, as one technique was specifically used to hide the fact that transactions were not used.

I have strong experience implementing systems based on the Windows DNA framework - both with Oracle (on Unix and NT) as well as SQL Server. What annoys me is not whether Oracle or SQL Server is superior, but that the benchmarking techniques are misrepresented. The benchmarks imply a use of COM-based transactions, which is not true.

What amazes me is that my independent testing demonstrates that COM+ can deliver on its promise, though not by using the mickey-mouse sample applications and demo code that Microsoft publishes for developers to learn from.

Note: SQL Server does perform better that Oracle in the Windows DNA framework, but Oracle is working to improve the integration of their client layer. With Oracle 8i comes the addition of native support for OLE Transactions, which should eventually narrow the gap that resulted from relying on the XA protocol.

Mike

"Norris" <jcheong_at_cooper.com.hk> wrote in message news:8cmdsk$12b9$1_at_adenine.netfront.net...
> You can read this article:
>
> Microsoft's top developer executives take the offense against
> competitor's claims
>
> http://www.infoworld.com/articles/hn/xml/99/12/20/991220hnhotmsdev.xml
>
Received on Sat Apr 08 2000 - 00:00:00 CDT

Original text of this message

HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US