[rescue] Solaris vs sparclinux

[Patrick Giagnocavo]

The best performance I found for Solaris, was lots of memory being used for
processes, some disk access and some swapping going on.  Linux would
basically grind to a halt on the same hardware, Solaris would not, even on
a dual 300Mhz or 400Mhz E250.  And that was a real-world example of stuff
going on.  Nowadays with RAM being so cheap, it is less of a use-case than
before.

On Fri, Jun 12, 2015 at 3:22 PM, Jerry Kemp <sun.mail.list47 at oryx.us> wrote:

> As far as getting UFS root file system on Solaris, the easy way is just to
> run some version of Solaris 10.  Solaris 10 update 11 was the last released
> version.
>
> There is a way to get newer versions on UFS, but it is not for the meek or
> timid, and requires jumping thru various flaming rings of fire.
>
> Also, as an alternative to Sun/Oracle Solaris, BSD or linux, there are
> also misc open-source Solaris based distro's.  On Sparc, I would suggest
> OpenSXCE, which is championed by Martin Bochnig.
>
> An additional note when benchmarking Sun/Oracle T-series boxes, especially
> early boxes like the T1000/T2000.
>
> These boxes were never fast boxes for single threaded applications, even
> when they were brand new, fresh from Sun.
>
> Sure, its awesome to run mpstat, and the operating system is seeing 64 or
> more CPU's available in a 1U or 2U box, but if your benchmarking
> application is only using one or two of those, its really not a relevant
> test.
>
> Back when those boxes were still relevant, my $EMPLOYER had some very
> highly threaded Java apps we were running, and once our T2000's, or T5xxx
> boxes got spooled up, the T-series boxes were great, and no x86/x64 box
> could touch them.  Not that there were some groups that didn't try.
>
> OTOH, we had a web application with some non-threaded modules that did
> horrible.  I fought with my boss for 18 months that T-series boxes were not
> the right system to be running that application on.  We even brought Sun in
> and had them look everything over, do all kinds of fine tuning, etc.  Sun
> kept telling him the T-series was where he wanted to be, and nothing I
> could say would change his mind.
>
> At one point, I finally got a vendor to bring in an M3000 to benchmark the
> app on.  No tuning or anything else.  Just threw the app on and tested.
> The M3000 moved three times the data that the T-series boxes did.  It
> wasn't that one was better than the other, is was just a matter of choosing
> the right tool for the job.
>
> What's the point of me writing all of this?  Just hope you are comparing
> oranges to oranges, especially when you are working with T-series systems.
> Don't get me wrong, they are certainly cool boxes, but if you are looking
> for maximum raw performance, you need to make sure you have the right
> workload to get it out of the t-series box.
>
> Food for thought,
>
> Jerry
>
>
>
>
> On 06/12/15 03:27 PM, Meelis Roos wrote:
>
>> Hmm, it would be interesting to install Solaris on a spare disk in T2000
>>> and run some bencharks.
>>>
>>
>> Well, I did that - grabbed Solaris 11.2 and installed it on identical
>> disk.
>>
>> Debian unstable with 4.0.0 kernel and my custom configurationf for that
>> machine (left out most drivers I will never need, left out some features
>> I would not use).
>>
>> Server is T2000, 1 GHz, 32 threads, 16G RAM, 10kRPM SAS disks. Bare
>> hardware, no virtualization if any kind (no ldoms, zones, containers).
>>
>>  And, should I run single-disk Solaris on ZFS or journaled UFS for this?
>>>
>>
>> ZFS was the only option.
>>
> _______________________________________________
> rescue list - http://www.sunhelp.org/mailman/listinfo/rescue