[Info-vax] Distributed Applications, Hashgraph, Automation
Kerry Main
kemain.nospam at gmail.com
Sun Feb 25 10:19:12 EST 2018
> -----Original Message-----
> From: Info-vax [mailto:info-vax-bounces at rbnsn.com] On Behalf Of
> DaveFroble via Info-vax
> Sent: February 25, 2018 2:18 AM
> To: info-vax at rbnsn.com
> Cc: DaveFroble <davef at tsoft-inc.com>
> Subject: Re: [Info-vax] Distributed Applications, Hashgraph,
Automation
>
> Kerry Main wrote:
>
> > Only if not well planned. The argument for "one bus app per OS
> instance" is
> > common in the commodity OS world because of not only technical
> challenges,
> > but also culture i.e. "no way I am running my bus App on the same OS
> as
> > another Bus App".
>
> This can be traced to the usage of PCs to get around IT departments
who
> may or
> may not have been responsive enough for users. It was anarchy then,
> and now it
> is just a big mess.
>
Yep, the centralized glass house was not responding to the needs of its
users, so when cheap IT technology became available locally, most IT
departments jumped into it with no little planning other than "lets go
buy servers and do stuff".
Today, this is often referred to as the wild west of distributed
computing. HP internally had a major push to root this out and called it
"Shadow IT".
Now, when the overall $'s started to dry up ad the real costs of
managing so much IT infrastructure in an uncoordinated manner and 5-10%
busy servers really became known to C-Level execs, that is when products
like VMware were born.
Unfortunately, while products like VMware addressed HW sprawl, they do
not address the cultural issue of every BU still wants to do their own
IT strategy. The only difference is that they now do their own plans
with individual VM's and not separate server HW. Hence, VM sprawl is now
a scourge in many companies and this is a much, much tougher issue to
address.
So, to summarize, the answer is NOT totally distributed or totally
centralized compute models, but something in between that adopts the
best of both strategies.
Both have pro's and con's.
- The totally distributed model - side is that this model is expensive
to maintain and is difficult to deploy common standards. However, on the
+ side, this model is more responsive to local end users service
requirements.
- The totally centralized model + side provides very low management
costs and is easy to deploy common standards. However, on the - side,
this model is much less responsive to local users service requirements.
Each company needs to find the mix that works best for their unique
needs.
Regards,
Kerry Main
Kerry dot main at starkgaming dot com
More information about the Info-vax
mailing list