[Benchmarking] Some new raster benchmarks

[Andrea Aime]

Chris Tweedie ha scritto:
> Hi Jeff,
> 
> Yes I was aware of what happened pre-2009. Shawn was involved in some
> web chats but it seemed our involvement came too late with the platform,
> tests and other items already confirmed. Shawn was very keen to include
> some advanced raster workflows but was flat out rejected in very
> "succinct" language I believe.

We have a requirement of allowing people to redo the benchmarks.
His proposal to use 5TB of data was not feasible due to that, and to
the fact that we did not have such storage on the benchmark server.

He also spoke of 12 bands TIFFs and I think everybody just needed to
check if their own server was up to support such kind of TIFF, but I 
don't think we said no to that.

Anyways, looking at the data you used for your benchmarks I don't see
any problem, it's all within acceptable size ranges.
The GeoServer configuration is not ok with respect to the machine you
benchmarked on, but I'll write you about that separately.

> The event itself was a month before our major 2010 release so timing was
> also an issue. Shawn as the PM rightly made the call to pull out before
> we really were ever involved. 
> 
> As for the blog, that's his personal space much like mine which needs to
> be treated as such.

Sorry, this is not acceptable. His personal blog is not a pub
conversation in front a beer, it's a public place from a public facing
person in ERDAS, and in particular, from the person that approached us
for the benchmarks, the only one person we knew from ERDAS.

> What does boggle my mind is why ERDAS is hit from all sides on this and
> yet ESRI got off scot-free? 

Where do you see ERDAS being hit from all sides? I'm not saying it's not
happening, I would be kind of surprised of the opposite, but I did not
see significant blogs or twits about that.

Anyways, ESRI showed some basic respect, whilst the only position
we got from ERDAS, even if you say it was on a personal blog, had some
quite strong positions against the benchmark itself:
http://owston.blogspot.com/
The same person states "I am EXTREMELY glad I didn't waste any resources 
on attempting to participate".
That is basically _asking_ for adverse reactions.

In open source everybody brings on the table what he can. Sometimes
things get messy, and they surely did this year. If we had more
people joining the table and sharing the preparation work we could
have done better I guess. It was the first attempt
to run this benchmark with a diverse and distributed community accessing 
it. I've participated to the benchmarks in the past years and doing them
in a private network on machines you can admin directly, where you have
GUI access, and no connection speed hurdles, is _a lot_ easier.

I'm pretty sure we'll do better next year.

> Will we participate in Barcelona? I have no idea ... its not my call.
> Would I like us to? Yeah I would, but its pretty clear from my POV that
> the slate needs to be wiped clean and things start fresh for next year
> so all this can be left behind us.

I surely hope ERDAS is going to participate, the more servers the better.

I also hope you'll be representing ERDAS in the effort, working
with you in the past has been a pleasure and I'm sure it will be in the
future as well.

Cheers
Andrea



-- 
Andrea Aime
OpenGeo - http://opengeo.org
Expert service straight from the developers.