[Info-vax] Using VMS for a web server

[David Froble]

Stephen Hoffman wrote:
> On 2015-06-03 23:28:39 +0000, Dirk Munk said:
> 
>> I'm aware of all these things, but if WASD can be the web server of 
>> choice for VMS, then it would be possible to drop Apache. After alle, 
>> WASD is a pure VMS product, it may be easier to implement on VMS than 
>> Apache.
> 
> As good as WASD is, Apache is the web server that folks expect and are 
> familiar with.  Increasing the delta between familiar platforms and 
> tools isn't usually the best approach.  Do not ever underestimate the 
> advantages of being compatible with expectations, training and tools.
> 
> 
>>
>>>
>>> Then there's the question of pricing.  VSI is also competing with a
>>> zero-cost entry-price for Linux and BSD distros; for those folks
>>> experimenting with and just getting started.  If hosting services are in
>>> play, then you're competing with Amazon and folks like
>>> https://www.digitalocean.com/ or http://www.rackspace.com/ or
>>> http://azure.microsoft.com/ here.  Spooling up a guest is seriously fast
>>> on most boxes, too.  A few clicks and you're off and running, whether
>>> hosted services or with an end-user-targeted server OS configuration.
>>>
>>
>> I'm referring to the type of web server that is a front end to your 
>> business, not some box at a hosting organization.
> 
> Seems you're following the one-box one-architecture model, and that was 
> something familiar, comfortable and common back in the 1990s and 2000s, 
> but most places are now dealing with heterogeneous server installations, 
> or with their own existing and installed web front-ends. OpenVMS isn't 
> at the center of nearly as many business configurations.  While there 
> are some web-facing OpenVMS boxes at businesses — running Apache, BTW — 
> there are more than a few folks that are running tools and content 
> management and the rest that are predicated on other platforms.  Until 
> and unless those front-ends and tools are available on OpenVMS, moving 
> OpenVMS into the roll of a web host is going to be a tough sell.  Then 
> there's the question of whether you want to have your production OpenVMS 
> boxes in your DMZ.
> 
> It's possible to be your own hosting organization, BTW.  Makes for a 
> handy way to quickly increase the scale of your hosting by either 
> rolling in a rack or two of servers — something OpenVMS is not very good 
> at — and/or by temporarily adding outside hosting for some tasks.

If I was still installing new systems I have the opinion that I could 
roll in that rack or two of new systems and have them up and running 
without a lot of fuss.

What you're talking about is whether knowledge and capability are 
required, or whether any old dummy that can figure out a mouse can do 
so.  Maybe there are systems where any old dummy can set them up, but I 
have to wonder, how well are they actually set up, and is this type of 
thing where some of today's security and other problems might be coming 
from?  You get what you pay for.  Pay for the dummy, if that's all you want.

> Individually managing specific servers is also becoming rare — servers 
> are resources, and having to individually configure and manage and name 
> and address them and load applications and modify startup files is 
> inefficient at best.
> 
> 
>> I suppose it all depends on what you want to do with your web server. 
>> I'm not suggesting that a VMS web server could or should do anything 
>> any other web server can do. It's like with buying a car, you can buy 
>> a very amall one, a big one, a sports car, a saloon, a station wagon, 
>> a pickup, it all depends on your needs and wishes.
>>
>> And if a VMS webserver can do the things we need it for, then that is 
>> good enough.
> 
> Trying other server platforms and tools means you'll know much more 
> about both the sorts of things that OpenVMS does better, and the things 
> that OpenVMS does worse.   OpenVMS still solves the same sorts of 
> problems that were common back in the 1980s and 1990s, but those sorts 
> of problems aren't necessarily the sorts of problems and limitations and 
> environments that many folks are working with now, and interested in 
> acquiring servers for now.   OpenVMS also solves many of those problems 
> in ways different from other approaches; not necessarily for the 
> better.  Those 1980 and 1990 data centers and configurations basically 
> don't exist anymore, outside of a few parts of the installed base.

You actually open up a topic that isn't often considered.  Who were 
those people in the 1980s and 1990s, and 1970s for that matter, that 
were solving problems with what was then very expensive computers?  To 
answer my own question, they were people who vitally needed the 
capabilities, and had the resources to pay for the capabilities.  Don't 
think those people no longer exist.  What has happened is that the costs 
have dropped so low that just about anyone can make use of some of the 
capabilities.  So yes, the majority are the newer users, and yes, they 
expect things to be easy, and cheap.  But the "more serious" (my term) 
users are still there, and they sometimes still need the "more serious" 
capabilities.

You claim it's "a few parts of the installed base".  Mostly because the 
installed base has grown so much larger.  Not because there is no longer 
any need.

> If you're not looking around and learning about other approaches and 
> other tools and other platforms, then your ability to even try to 
> discuss and to promote OpenVMS — if that's your goal here — will be 
> poor.   Beyond the basic marketing and comparisons, you'll also have the 
> sorts of technical issues around just managing and supporting OpenVMS 
> servers, as there'll be requests to integrate with various newer tools — 
> everybody's managing more and more servers, or deploying more and more, 
> or reconfiguring, etc.
> 
> You might learn that if you're going to be different, then you usually 
> need to bring specific advantages over the other approaches.  Being 
> different because you're better integrated with OpenVMS might not be 
> something folks evaluating OpenVMS care all that much about, for 
> instance.  The folks might prefer a better-integrated Apache, for 
> instance, because that means they can use their Apache skills and 
> tools.  Existing OpenVMS Apache users will probably want to see that 
> continue, which means that Apache will either continue and that adds to 
> the work, or it gets deprecated and that might be a problem for some in 
> the installed base.
> 
> To repeat some of my comments from a boot camp a whole back, OpenVMS is 
> far too piecemeal in its installation and configuration and management — 
> there are other platforms around which are *massively* better at this 
> — and I'd really like to have what are now core features — web services 
> among those — in the base distro.  Apache would be the local preference, 
> just because that's familiar to many, already present, already in use, 
> and very common in the industry.   But whether that integrated web 
> server becomes WASD or Apache or nginx or something else?

Or most of the above ....