Not to take sides in this or anything, but I believe the study in question
looked specifically at database performance, where you have lots of opening
and closing of connections, etc. When you're talking about a streaming
appliance, you typically have some opening and closing of connections, but
it's mostly just sending constant packets of data, and in this area OS X
performed adequately.
Again, I'm not trying to take sides, and it's possible my recollection is
incorrect.
Colin Witt
Baylor University
> From: Heitzso <email@hidden>
> Date: Thu, 22 Dec 2005 13:00:00 -0500
> To: Charles <email@hidden>, <email@hidden>
> Subject: Re: Paid support for DSS?
>
> There was a recent comparison test of current Linux kernel
> versus current OS X on the same Apple hardware. OS X came out
> very poorly as the load increased (several factors worth).
> Problem seems to be the OS X kernel is not well designed re
> granularity of kernel locks so a heavily threaded (i.e. web/db
> app with lots of users beating on it) situation beats OS X up,
> at least in comparison w/ Linux. Current OS X improves the
> multitasking ability of OS X over Panther, but still cannot begin
> to compete with Linux kernel.
>
> I may have seen a variation of this in that a 450 mhz Pentium III
> laptop running Windows 2K Pro can display an embedded QuickTime
> movie in a Firefox frame while my 1ghz eMac struggles to do the
> same with Panther failing and Tiger just barely managing it.
> That surprised me because the cpu design in the eMac is supposed
> to provide more crunch per cycle than the old Pentium, so I
> would have expected the 1ghz eMac to be 3 times faster (or more)
> than my old Pentium III laptop, instead of struggling to match it.
>
> Odd side note is that Safari and IE both fail to handle the mix
> of embedded QuickTime in a frame and I have to give them a
> super simple web page for the embedded QT player otherwise
> they both break. While Firefox didn't seem to have a problem.
>
>
>
>
>> Not yet, and I'm very curious about this too. (It's not that I /don't/ want
>> to use Apple hardware, but I have to do due diligence.)
>>
>> I've found firms that offer load testing services, but I haven't found any
>> tools that I can use myself.
>>
>> -- Charles
>>
>>
>> -----Original Message-----
>> From: streaming-server-users-bounces+lists06=email@hidden
>> [mailto:streaming-server-users-bounces+lists06=email@hidden]
>> On Behalf Of email@hidden
>> Sent: Thursday, December 22, 2005 4:37 AM
>> To: email@hidden
>> Subject: Re: Paid support for DSS?
>>
>> Charles,
>>
>> Have you done any actual comparison testing on the hardware end? If so,
>> I'd be curious to know the relative performance of a dual 2.3ghz XServe
>> w/QTSS for streaming when compared to a recent vintage dual processor
>> Xeon/Opteron blade server running Linux + DSS. I am in much the same boat
>> as you are. The Xserves are tempting because of the "it just works out
>> of the box" aspect, the relative ease of maintenance with applecare.
>> However, recent articles I've seen haven't been very encouraging with
>> OSX's scalability on heavily loaded servers. Also, 1RU dual Xeon/Opteron
>> hardware solutions are so much cheaper in comparison. Times umpteen
>> servers, it really starts to add up.
>>
>>
>>
>
> _______________________________________________
> Do not post admin requests to the list. They will be ignored.
> Streaming-server-users mailing list
> (email@hidden)
> Help/Unsubscribe/Update your Subscription:
> http://lists.apple.com/mailman/options/streaming-server-users/colin_witt%40bay
> lor.edu
>
> This email sent to email@hidden
_______________________________________________
Do not post admin requests to the list. They will be ignored.
Streaming-server-users mailing list (email@hidden)
Help/Unsubscribe/Update your Subscription:
http://lists.apple.com/mailman/options/streaming-server-users/email@hidden
This email sent to email@hidden
