Talk:Sun Java System Web Server

From Wikipedia, the free encyclopedia

[edit] Record for SPECweb2005 Performance

Toph3r recently removed text noting that Sun Java System Web Server holds the current record for SPECweb2005 performance. I believe this was useful NPOV information that is relevant in an article on Sun Java System Web Server. Indeed, when Zeus Web Server held the record, Toph3r thought it was worth mentioning on the Zeus Web Server page. I will revert Toph3r's change and leave a comment at User_talk:Toph3r encouraging him to explain his rationale here. — Elving 02:37, 3 May 2006 (UTC)

Hi Elving, first things first - I didn't add the SPECweb results in the ZWS article :) They were there long before I discovered that article. The reason I do not feel that the SPECweb score is valid, with regards to Sun JSWS, is if you look at the *hardware* used in this test - it by _far_ outranks any similar hardware used against a ZWS benchmark. Indeed, the most cores used in a specweb test for ZWS was 2 (iirc). Yet, Sun are using *8* cores - that's hardly a fair test, and i very much feel its inaccurate to state that Sun JSWS holds the highest score as a result. It's like me running the 100m, but with x8 the legs of Carl Lewis ;) -- simply not a fair and reasonable comparison. Cheers. --Toph3r 09:06, 3 May 2006 (UTC)

So the problem is that the hardware is not explicitly mentioned? Would something like "Sun Java System Web Server on the Sun T2000 server holds the current record for SPECweb2005 performance" resolve your objection? (Also, note that while the T2000 does have 2x the cores of its closest Xeon-based competition, the Xeons are clocked 3.1x faster. If one looks at simply cores × GHz, one would expect ZWS on Xeon to have the advantage.) — Elving 09:17, 3 May 2006 (UTC)

Hi Elving. Yep, I think that would be a fair and reasonable addition to the article. You make some good points re the cores. I'll investigate these later today. Thanks for your fair approach. --Toph3r 09:20, 3 May 2006 (UTC)