[linux-elitists] Get off my lawn.

[Don Marti]

An entire new web company where nobody has to run
anything as root.

The Serverless Start-up
http://highscalability.com/blog/2015/12/7/the-serverless-start-up-down-with-servers.html

We would like to introduce Teletext.io, also
known as the serverless start-up - again, entirely
built around AWS, but leveraging only the Amazon
API Gateway, Lambda functions, DynamoDb, S3 and
Cloudfront.

Database as a service, function calls as a service, and
of course storage and CDN as a service. All tied
together by a single-page JavaScript application.

"If you upload your Lambda code, Amazon will take
care of everything required to run and scale your
code with high availability. Lambda executes in
parallel. So, if a million requests are made, a
million Lambda functions will execute without loss
of speed or capacity. According to Amazon, "there
are no fundamental limits to scaling a function"

(Kind of like a bigger version of a 1990s web hosting
account with your own `cgi-bin` directory.)

This highscalability.com web site has a lot of
thought-provoking stuff on it. Another good example:

Uber Goes Unconventional: Using Driver Phones as a Backup Datacenter
http://highscalability.com/blog/2015/9/21/uber-goes-unconventional-using-driver-phones-as-a-backup-dat.html

[Phil Mayers]

Post by Don Marti
Database as a service, function calls as a service, and
of course storage and CDN as a service. All tied
together by a single-page JavaScript application.

But, but... my years of now-obsolete experience?!?!

I'm sure the industry will keep clinging to the best the 1970s has to
offer for a few more years yet. After all, there's no reason IT should
be any different to any other human endeavour :o(

Sarcasm aside, I wish them luck - our current approaches work well for
some problems, but badly for others. Even if their approach fails, the
manner of failure should be instructive, and we desperately need to
start thinking as a profession about the long-term endpoint of our efforts.

I'm not sure if it'll happen or if I'll live long enough to see it, but
if hardware performance eventually tops out and we reach a stable
saturation point of computing for any length of time - say a couple of
human lifespans - I suspect we'll see one of three outcomes:

1. A software engineering discipline with well-accepted principles of
design, widespread commonality of tools and standards, and a few well
accepted ways to attack a given problem - sort of like Civil
Engineering, but for code.

2. What Vernor Vinge called a "mature programming environment" - this is
not what you might think if you've not come across his work. "Programmer
archaeologist" is a job title. Bugs lurk dozens of layers deep in code
centuries old. And given the volume of code, ground-up rewrites are,
basically, impossible.

3. Software written by software written by software written by... Either
AI or something so close to it that the resultant code is as
incomprehensible to an unaided human as the full DNA code of a living
creature.

FWIW my money is on #2 - half my job already feels like archaeology, and
if I have to spend another half-day doing something as apparently simple
as deploying a WSGI application, I might quit and become a plumber. At
least it's up-front that you have to deal with s**t.

So good luck to 'em - maybe it'll steer us towards option #1.

Quite how they'll license it to avoid widening the already substantial
power-gap between infrastructure owners and users however...

[Aaron Burt]

Post by Don Marti
An entire new web company where nobody has to run
anything as root.
The Serverless Start-up
http://highscalability.com/blog/2015/12/7/the-serverless-start-up-down-with-servers.html

Yep, service LEGO. I remember a fellow discussing his EBS/SQS/RDS app a
few years ago at an AWS meetup. I've long been under the impression
that's what AWS has in mind.

If your service offering is on a fairly generic stack with not that much
specialized user-interface or algorithms behind it, you're in danger of
being disrupted by J. Random Salesman and his AWS LEGOneer. Engineering
effort? 1-2 orders of magnitude less. Scalability? Yeah, whatevs.

That's frightening.

My own thinking has been:
(a) Be ready to price and build stuff with AWS LEGO.
(b) Pay attention to the AWS work-alike stacks.
(c) Think about the missing pieces and how to make them, e.g. on EC2.
(d) Look for things that don't fit the AWS model.

(c) and (d) are where the fun (and remunerative) things are.

[Don Marti]

Post by Don Marti
An entire new web company where nobody has to run
anything as root.
The Serverless Start-up
http://highscalability.com/blog/2015/12/7/the-serverless-start-up-down-with-servers.html

Yep, service LEGO. I remember a fellow discussing his EBS/SQS/RDS app a few
years ago at an AWS meetup. I've long been under the impression that's what
AWS has in mind.
If your service offering is on a fairly generic stack with not that much
specialized user-interface or algorithms behind it, you're in danger of
being disrupted by J. Random Salesman and his AWS LEGOneer. Engineering
effort? 1-2 orders of magnitude less. Scalability? Yeah, whatevs.
That's frightening.

No more frightening that the MSCEization of
run-of-the-mill client-server applications in the
pre-web days. Or the vendor-certified companies that
design and deploy POS systems. Or the fact that an
electrician can wire a house to code -- you don't
need an Electrical Engineer to design a unique system
any more.

(a) Be ready to price and build stuff with AWS LEGO.
(b) Pay attention to the AWS work-alike stacks.
(c) Think about the missing pieces and how to make them, e.g. on EC2.
(d) Look for things that don't fit the AWS model.
(c) and (d) are where the fun (and remunerative) things are.

One fun part is offering a pluggable service with
"LEGOneers" as your users.

Also makes "scaling" more of a concern for low-level
projects (OS, database, language runtimes...).
Instead of those projects selling to, and designing
for, an IT professional who serves thousands of users,
they're serving a SaaS component company that serves
hundreds of "serverless" projects, each with thousands
of users.