[Info-vax] OpenVMS servers and clusters as a cloud service

[Kerry Main]

> -----Original Message-----
> From: Info-vax [mailto:info-vax-bounces at rbnsn.com] On Behalf Of
> Johnny Billquist via Info-vax
> Sent: January 2, 2018 5:59 AM
> To: info-vax at rbnsn.com
> Cc: Johnny Billquist <bqt at softjar.se>
> Subject: Re: [Info-vax] OpenVMS servers and clusters as a cloud
service
> 
> On 2018-01-01 16:45, Kerry Main wrote:
> 
> >>> There are basic things which have to happen for every server
> instance
> >>> and yes, smart tools and custom automation absolutely do help, but
> >>> there is a limit on what tools can do without some level of
> experience
> >>> at the controls.  That is why large companies like Google have L1,
L2
> >>> and L3 escalation support models for each level of their IT dept.
> >>>
> >>> Do these escalation teams count as part of their "server" SysAdmin
> >> ratio's?
> >>>
> >>> Make no mistake - there is lots to learn about scalability from
the
> >>> Googles and Amazons of the world, but we also need to keep open
> >> minds
> >>> and not blindly accept whatever their marketing depts. pump out
> >> about
> >>> how good they are.
> >>
> >> OpenVMS has installation, deployment, patch management, remote
> >> management, app deployments, recovery and system re-installation,
> >> and
> >> troubleshooting and related comparative-competitive difficulties at
> >> deployments of scale one.
> >>
> >> As for the scale of deployments, what passes for a very large
> OpenVMS
> >> installation is smaller than what many of us have seen of Microsoft
> >> Windows clients in use on one floor of one building in one office
park,
> >> too.
> >>
> >
> > Different culture as well - lets not forget that Windows and Linux
> typically deploy one bus app or one App service per OS instance. That
> drives much higher server numbers, but usually at the pain of VM
> servers that run at 10-15% busy most of the day.
> >
> > With OpenVMS, it is very common to run many Apps on one OpenVMS
> instance.
> 
> What kind of nonsese is this? Google runs Linux. Do you think Google
> only runs one app per machine??? They do not. Like I said before,
> machine management is all automated, including scheduling jobs on
> machines. And the system knows the load of different applications, and
> can combine many applications on one machine, as long as there are
> resources available.
> 

To clarify - the common model for Linux and Windows is one "business"
App per server. 

It is very uncommon to run bus apps from different groups on the same OS
instance. Each group wants there own server instances (P/V).

To quote David's earlier reply - 
" If I can run a decent sized company on one VMS system, with all apps,
with the 
exception of office automation and web servers, does that qualify as
"many 
apps"?  Warehouse control, order processing, purchasing, forecasting,
AR, AP, 
GL, database, and more?"

How many Linux and Windows environments would run all these on one OS
instance?

I know many, many (most?) OpenVMS environments where David's example
here is the norm, not the exception.

I know of one very large lottery that is still cranking out billions $
each year on a 2 Alpha server cluster. That homogeneous cluster has
15-20 different Apps / DB's running on the two server cluster. I know,
because they were looking at a move to Integrity (support reasons, not
performance) and each App had to be looked at for the porting impact
analysis.

> One application per machine is such a nonsensical and uninformed
belief
> that it almost blows my mind. I can't comment on Windows at all, but
in
> the Unix/Linux world, nothing could be further from the truth.
> 

See above. More simply - how many Linux/Windows servers would put the
Web server, App server and DB server on the same single OS instance?

> There is a reason containers have been developed/pushed in Linux. To
> increase isolation between multiple applications running on the same
> machine. However, I do think all the knobs in VMS can be very useful
in
> this context. Unix sometimes certainly suffer because of the lack of
> those (hello page quota).
> 
>    Johnny
> 

In some respects, containers are an attempt to address the growing
challenge of VM sprawl i.e. reduce the numbers of separate OS instances.

Not very well known, but OpenVMS also has an native class scheduler
which can also be used to ensure certain processes or even groups of
processes do not take more resources than they are allocated. This is
another required feature when one looks at App stacking.

$ mcr sysman
Sysman> help class 

Also, reference the following:
<http://h41379.www4.hpe.com/openvms/journal/v15/class_schedule.pdf>
" The class scheduler provides the ability to limit the amount of CPU
time that a system's users may receive by placing the users into
scheduling classes. Each class is assigned a percentage of the overall
system's CPU time. As the system runs, the combined sets of users in a
class are limited to the percentage of CPU execution time allocated to
their class. To invoke the class scheduler, use the SYSMAN interface.

Class scheduling is implemented in the SYSMAN utility, which allows
users to define classes based on username, UIC, or account. SYSMAN
allows users to create, delete, modify, suspend, resume, and display
scheduling classes."

Regards,

Kerry Main
Kerry dot main at starkgaming dot com