[Info-vax] BASIC compiler in the hobbyist distribution

[Stephen Hoffman]

On 2015-05-31 16:38:10 +0000, johnwallace4 at yahoo.co.uk said:

> On Sunday, 31 May 2015 16:17:23 UTC+1, seasoned_geek  wrote:
>> On Saturday, May 30, 2015 at 4:01:00 PM UTC-5, johnwa... at yahoo.co.uk wrote:
>>> 
>>> The Proliant will also be qualified to run a couple of business class 
>>> Linuxes (the Packard Bell? Why?)> >> > Think about how the *systems* 
>>> compare:
>>> . Resilient (Proliant) vs cheap (consumer)
>>> .. e.g. redundant PSU vs single PSU, ECC memory vs non-ECC, etc
>>> . Serviceable (Proliant) vs disposable (consumer)
>>> . Expandable (Proliant) vs bounded (consumer)
>>> . Documented (Proliant) vs 'good luck with that, mate' (consumer)
>>> . Expected lifetime of years (Proliant) vs months (consumer)
>>> . Multiple OS multiple version support (Proliant) vs you get what
>>> you're given and that's it (consumer)

Intel has been rolling the Itanium RAS features — now known as Intel 
Run Sure Technology to Intel marketeers everywhere — into Xeon for a 
while now:
<http://labs.hoffmanlabs.com/node/95>
<http://www.intel.com/content/www/us/en/architecture-and-technology/intel-run-sure-technology.html> 


>> You know, it never ceases to amaze me just how short of a memory people have.

>> It was not all that long ago Amazon made the nightly news for roughly a 
>> week due to a massive outage where they had to replace so many 
>> thousands of servers it exceeded their suppliers capacity to provide 
>> them.

Yet Amazon still has customers.  Lots of them.  
<http://highscalability.com/blog/2015/1/12/the-stunning-scale-of-aws-and-what-it-means-for-the-future-o.html> 


One size does not fit all.

>> Most of the other "cloud" providers have had similar outages where 
>> hundreds or thousands of these supposedly enterprise quality servers 
>> committed harry-kari in a Jonestown like fashion.
>> 
>> Quit drinking the KOOL-AID

FWIW, presuming that customers and computer users are stupid?  Probably 
not the most auspicious start for a sales pitch.   Those are some very 
dark references, too.

Might want to start out with a pitch that includes the features and 
benefits, maybe some plots around uptime or reliability or total cost 
of ownership or such.   Maybe find out what is and is not working for 
folks, too.    Some folks can and should use cloud providers — and for 
various and good reason.  Other folks can and should use commodity 
servers or private clouds or bespoke servers.  Or boxes running a 
somewhat more unusual operating system, such as OpenVMS or NSK or z/OS.

One size does not fit all.

> I don't know whose servers you think AWS/EC2/etc use but I do hope you 
> don't think it's Proliant.

Ayup.  The servers will likely be bespoke.  Either buyer-specific:
<http://labs.hoffmanlabs.com/node/954>
<http://blogs.microsoft.com/firehose/2014/01/27/microsoft-contributes-cloud-server-designs-to-the-open-compute-project/> 

or buyer-tailored products from an OEM or ODM vendor, or designs based 
on Open Compute or otherwise.
<http://www.opencompute.org>

I'd expect that HPE and the other OEMs are likely selling most of their 
server gear to folks in the low- to middle-range of sales volumes, 
these days.  Moderate- to large-scale buyers — Amazon, Google, 
Microsoft — and buyers with big budgets or unusual requirements — can 
and probably will purchase tailored designs.  Or high-end boxes.  Or 
both.  Or they go "four walls" or they outsource their data centers.  
Or some combination.

One size does not fit all.

> You'd be a lot closer if you were thinking Amazon talk direct to white 
> box builders like Foxconn who will, given sufficient volume, build 
> whatever the customer wants.

Which can be reasonable, as each OEM vendor tends to have their own 
management tools and add-ons and options, and Amazon, Google or 
Microsoft other large-scale buyers likely have their own expectations 
and tools here, and may not or do not want or don't need the OEM 
features.

One size does not fit all.

> You may even have noticed the recent HP-Foxconn deal, which HP seem to 
> think  will allow them to make money in the ultra low margin ultra high 
> volume world of cloud server sales. (I think they're mad to even think 
> about competing with Foxconn directly, but I don't get out much).

It looked like HP was a reseller of Foxconn-branded gear.   Like that 
HP iPod of a few years back.

Post-split, it'll be interesting to see if and how quickly Foxconn or 
one of the other ODMs or OEMs might acquires HP Inc.  Or if HP Inc 
starts acquiring other entities.

> Unbadged whitebox and Proliant 'enterprise' are at different ends of 
> the server hardware spectrum, because they address different sets of 
> requirements, frequently conflicting requirements.

Ayup.  Some of the folks looking for white-box servers can be looking 
for cheap, or others for higher-end and far more reliable white-box 
configurations.   Some folks running applications on x86 want low-end 
gear, and some want multi-socket and ECC and RAID configurations.   x86 
— for all its warts — does solve the problems that a whole lot of 
people have.  OEM, ODM and open-sourced and white-box server designs 
run the gamut, too.

One size does not fit all.

> Some are fundamental requirements such as price and performance, others 
> may be slightly woolier, such as reliability, spares availability, etc.

Or running a redundant array of servers, something that OpenVMS just 
isn't very good at.

> One size does not fit all.

Ayup.


-- 
Pure Personal Opinion | HoffmanLabs LLC