[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 Arne
> Vajhøj via Info-vax
> Sent: January 1, 2018 12:54 PM
> To: info-vax at rbnsn.com
> Cc: Arne Vajhøj <arne at vajhoej.dk>
> Subject: Re: [Info-vax] OpenVMS servers and clusters as a cloud service
> 
> On 12/31/2017 9:46 AM, Kerry Main wrote:
> >> -----Original Message-----
> >> From: Info-vax [mailto:info-vax-bounces at rbnsn.com] On Behalf Of
> Arne
> >> Vajhøj via Info-vax
> >> On 12/30/2017 9:31 PM, DaveFroble wrote:
> >>> Arne Vajhøj wrote:
> >>>> There are a lot of commodity stuff in IT operations.
> >>>>
> >>>> Windows client PC's
> >>>> MS Office
> >>>> Windows file & print servers
> >>>> Apache on Linux web servers
> >>>> some database servers (like FaceBook running 1800 MYSQL
> servers)
> >>>> Low end ERP systems (not SAP and Oracle)
> >>>> Etc.
> >>>>
> >>>> Are all standardized and measured in hundreds or thousands.
> >>>>
> >>>> According to internet gossip then Google has one sysadm
> >>>> per 28000 Linux servers.
> 
> >>> The IT of yesterday, today, and tomorrow is in many cases much
> more
> >>> specialized.  Just because google has 28000 servers, I'm betting none
> of
> >>> those servers can monitor and run many of today's jobs.  Industry.
> >>> Power.  And such.  Does google's 28000 servers make the rest less
> >>> important?  I think not.
> >>
> >> I wrote that Google (according to gossip) have one sysadm per 28000
> >> servers.
> >>
> >
> > That’s crap and/or are following the age old misdirection whereby they
> do not count a whole lot of resources doing what people would consider
> sysadmin work.
> > - Is that one sysadmin managing IP addresses for 28,000 servers?
> > - is that one sysadmin monitoring backups for 28,000 servers?
> > - is that one sysadmin monitoring security logs for 28,000 servers?
> 
> Tools, automation and self service.
> 

Technicolour dreaming. 

IT automation is great in long term, but it is hugely complex, very expensive (eng / consulting design, deployment, tools) to get there. Its why many SMB companies get frustrated that after paying big $'s for enterprise suites, a year later they are frustrated because they do not feel they are getting what they were sold. The reason is that to get the full benefit, they need to customize these tools and they did not invest in the post product buy custom engineering efforts required.

Good example - a router goes off line. If it is paired, then the indicator on the single pane of glass that OPS uses shows a yellow alert. If not paired, then the indicator shows red. What services are impacted so you can give those groups a heads up? If it’s a single router, every IT group is likely being blasted with phone calls that their server (storage) or "network is down". Remote App groups will be calling in to help desk that their App is down. 

Given all this calamity, how do you ensure the ticket is sent to the group responsible for the router, but at the same time, send an email to other groups that the issue is a router down, so no need to look at their server / storage / applications and avoid the associated ticket bounce? If after hours or on weekend, should the same email or auto ticket be generated or is there a special escalation flow after hours?

The above is all customization that takes time, resources, $'s and loads of understanding of the companies help desk, escalation flows and overall integration in the company.

> > - is that one sysadmin working with firewall groups to address issues
> with firewall issues associated with 28,000 servers?
> 
> That is not a sysadm task.
> 

The firewall admin makes the changes in the firewall, but the SysAdmin is consulted, because the App group will not know the source IP, dest IP, port and protocols required for the FW rule to be put in place in order to make some new App functionality work. Neither will the firewall resource.

> But likely to be automated as well.
> 

Good luck with automating internal firewall rules for DB, App, DMZ and the many other internal FW zones. 

😊

> >> I did not write that Google has 28000 servers.
> >>
> >> They have a lot more servers.
> >>
> >> Internet gossip says that they have 900000 servers.
> >
> > Huge numbers I have no doubt (likely 90% are VM's).
> 
> I assume physical servers with 10-20 times more VM's.
> 
> > A more impressive question to ask is "what is the average utilization of
> all these VM servers?"
> 
> Most likely much much higher than what the small shops can achieve.
> 
> > In case anyone doubts the costs of VM sprawl on an organization, here
> is a simple test:
> >
> > Ask your IT provider to give you a quote of what their monthly
> > support
> > charges are for 10 physical servers and 10 OS instances. Then ask them
> > what their monthly support charges are for 10 VM's running on one
> > physical server. The latter estimate for 10 VM's will be approx. 70-80%
> > of the former 10 physical servers.
> >
> > Now you know what outsourcers / cloud providers really understand all
> > to well - the real costs are associated with the management and
> > support of the OS instance - not the server hardware.
> And that is exactly why you are so wrong when you think IaaS cloud
> means out-sourcing a lot of IT operations.
> 
> It is OS and up - especially the applications that generate
> work - not the hardware.
> 
> Arne
> 

Think of IT as a series of layers. 

With each layer you outsource (move to public cloud) you shed more internal IT jobs which, to the C-level's anyway,  is a great thing - especially if their bonuses are based on revenue per employee. Do the math when you reduce employees. Implementing this and even with revenue flat, they still get bonuses.

1. Collocation - shed DC facility and related support jobs and costs
2. IaaS - reduce network + 1. related support jobs and costs
3. PaaS - reduce server support / OS + 2. + 1. related support jobs and costs
4. SaaS - reduce App support + 3. + 2. + 1. related support jobs and costs

Btw, the same applies to the decades old outsourcing models. Outsourcers will provide as much or as little as you want to pay.

Regards,

Kerry Main
Kerry dot main at starkgaming dot com