After thinking about it a bit more I think my thoughts are more along the
lines of recognizing a need for a split between a traditional sysadmin role
and a devops role. What do you wish developers knew when they asked for
resources? What would you require them to know before they have access to
production servers? (I agree that's contrary to best practices but some
definitions of 'devops' go that far.) What about ad hoc infrastructure,
e.g., what would expect developers to know before they start creating
running their own docker images on a k8s cluster?

I'm in an unusual, but not unheard of, position since we have a stable ops
environment but our product requires us to support an incredible variety of
servers/configurations. Many developers know how to create an instance on
EC2 and how to use yum to install the latest version of some server but I'm
currently on a joint team trying to figure out how we can offload a lot of
non-production work from the ops team without blowing up the budget or
destroying security. Ops simply doesn't have the bandwidth to understand
all of the different servers/versions we need to support.

I also want to reiterate one point - the LPIC exam shouldn't cover the same
material as the AWS exams even though so many devops candidates will be
doing most if not all of their work in a cloud environment. However the
test should be aware of it, e.g., if you're in a traditional environment
you'll set up postfix so your server can forward local email into your
corporate email. If you're in a cloud you'll probably use SES for the same
purpose. Why would you want to have to maintain an additional server if
you're only using it as a relay? Every server you run on your EC2 instances
increase the risk surface for a compromise. Likewise OpenLDAP should
mention that it can be backed by a relational database even if no details
are required. In a traditional environment you'll set up a database
somewhere, in the cloud you might decide it's better to use RDS. I think
AWS also offers AD.

A sample question could be:

Which server/service can be used to send email (Choose two)
  - postfix
  - postgres
  - SMS
  - SES
  - RDS

but that can be a little tricky since someone familiar with AWS will know
that you can easily set up pipelines between different services directly or
with simple lambdas. That means the client may post to a JMS topic and it
magically transform into outbound email. That means that 'SES' (simple
email service) is the correct answer but 'SMS' (simple messaging service,
iirc), 'SNS' (simple notification service), and possibly even SQS (simple
queuing service) would have to be an acceptable answer if the person is
knowledgeable.

BTW AWS services aren't limited to the cloud. It's easy to set up desktops
to use AWS to send email, to use as the database, etc. You could use the
AWS gateway so the services are transparently visible on your VPN or just
use secure protocols. (E.g., access SES via SMTPS.) For some small- and
medium-sized businesses it may be more cost-effective to do this than to
maintain their own servers.

Bear
_______________________________________________
lpi-examdev mailing list
[email protected]
http://list.lpi.org/cgi-bin/mailman/listinfo/lpi-examdev

Reply via email to