By: William Campbell (wcampbell.delete@this.realworldtech.com), May 23, 2006 8:17 pm
Room: Moderated Discussions
Kava (song_jj2003@yahoo.com) on 5/23/06 wrote:
---------------------------
>William Campbell (wcampbell@realworldtech.com) on 5/23/06 wrote:
>---------------------------
>>David Kanter (dkanter@realworldtech.com) on 5/23/06 wrote:
>>---------------------------
>>David, thank you for the review. Excellent work--and time consumming no doubt. ;)
>>
>>A general question for some feedback if i may. You make reference in your article
>>on several occasions about future performance gains using tuning and optimised software.
>>
>>In the real world, how likely is that to happen though. My experience is that
>>once a production platform becomes stable, the people who look after infrastructure
>>are _loathe_ to change to (for example) a new JVM etc. because the validation and
>>regression testing costs are prohibitive. Perhaps I've only worked in overly conservative
>>organisations? So while the theoretical numbers are impressive, whether Woodcrest
>>will be as dominant may not be so clear cut?
>>
>
>Tuning is very important, sometimes it pulls off huge improvement. Because CPU
>means nothing without software running on it, but softwares are written by people
>who very possibly do not know how to utilize CPU best. Intel definitely helps its
>big customers(BEA, SAP) tune their major software products. It's a win-win if they
>can sell the "platform". Do you want a JRokit with this much performance? Good,
>go buy a Woody because we only can reach there with it. What you go a Woody already
>but needing a JVM? Well go for BEA because that's the fastest master piece so far...
I see what you are saying, but in my current situation this holds no water. We already have too many differing servers and configurations. Our organisation wants to standardise on one. This leads to a nice tug-of-war when a particular application does not run/is not supported on the platform of choice. Performance on brand Y software running on brand Y hardware does not even enter into the planners considerations. In a small org, with only a few (probably of the shelf) applications, what you say is true. In a large org with much in-house software, probably less so.
---------------------------
>William Campbell (wcampbell@realworldtech.com) on 5/23/06 wrote:
>---------------------------
>>David Kanter (dkanter@realworldtech.com) on 5/23/06 wrote:
>>---------------------------
>>David, thank you for the review. Excellent work--and time consumming no doubt. ;)
>>
>>A general question for some feedback if i may. You make reference in your article
>>on several occasions about future performance gains using tuning and optimised software.
>>
>>In the real world, how likely is that to happen though. My experience is that
>>once a production platform becomes stable, the people who look after infrastructure
>>are _loathe_ to change to (for example) a new JVM etc. because the validation and
>>regression testing costs are prohibitive. Perhaps I've only worked in overly conservative
>>organisations? So while the theoretical numbers are impressive, whether Woodcrest
>>will be as dominant may not be so clear cut?
>>
>
>Tuning is very important, sometimes it pulls off huge improvement. Because CPU
>means nothing without software running on it, but softwares are written by people
>who very possibly do not know how to utilize CPU best. Intel definitely helps its
>big customers(BEA, SAP) tune their major software products. It's a win-win if they
>can sell the "platform". Do you want a JRokit with this much performance? Good,
>go buy a Woody because we only can reach there with it. What you go a Woody already
>but needing a JVM? Well go for BEA because that's the fastest master piece so far...
I see what you are saying, but in my current situation this holds no water. We already have too many differing servers and configurations. Our organisation wants to standardise on one. This leads to a nice tug-of-war when a particular application does not run/is not supported on the platform of choice. Performance on brand Y software running on brand Y hardware does not even enter into the planners considerations. In a small org, with only a few (probably of the shelf) applications, what you say is true. In a large org with much in-house software, probably less so.
| Topic | Posted By | Date |
|---|---|---|
| Woodcrest Preview online | David Kanter | 2006/05/23 09:30 AM |
| Woodcrest Preview online | savantu | 2006/05/23 09:40 AM |
| Woodcrest Preview online | David Kanter | 2006/05/23 09:43 AM |
| Forget about it | savantu | 2006/05/23 09:54 AM |
| Woodcrest Preview online | Kava | 2006/05/23 10:17 AM |
| Error in some of your numbers? | Emil Briggs | 2006/05/23 05:01 PM |
| I think I see your mistake | Emil Briggs | 2006/05/23 05:25 PM |
| I think I see your mistake | SLee | 2006/05/23 05:52 PM |
| Different comparisons | David Kanter | 2006/05/23 07:18 PM |
| The importance of tuning | William Campbell | 2006/05/23 06:37 PM |
| The importance of tuning | Kava | 2006/05/23 07:06 PM |
| The importance of tuning | William Campbell | 2006/05/23 08:17 PM |
| The importance of tuning | Dean Kent | 2006/05/23 07:18 PM |
| The importance of tuning | William Campbell | 2006/05/23 08:13 PM |
| The importance of tuning | Dean Kent | 2006/05/23 08:18 PM |
| Frame of reference | William Campbell | 2006/05/23 08:57 PM |
| The importance of tuning | David Kanter | 2006/05/23 07:24 PM |
| smile, you are now officially part of the media :) | mas | 2006/05/24 11:48 AM |
| and enjoy it while your engine lasts 8-) | anonymous | 2006/05/24 01:01 PM |
| Run...the media's here! (NT) | David Kanter | 2006/05/24 01:09 PM |
| Run...the media's here! (NT) | Kava | 2006/05/24 01:20 PM |
| Woodcrest Preview online | yuyuhakusho | 2006/05/28 12:49 PM |