Learning Continuously

A field that keeps moving

What you learned for your first job will partly expire. Exams are revised every few years: CompTIA's A+ moved to a new version (220-1201/220-1202) in 2025, and the previous version was retired in English that September. Products are renamed, tools are replaced, and practices that were normal (passwords without MFA, servers configured by hand) become warnings. Even this encyclopedia dates fast-changing facts "as of 2026" for that reason.

That sounds tiring, but it has a good side. The fundamentals change slowly: how a network moves packets, how an operating system runs processes, how to troubleshoot methodically. Someone who understands those learns each new tool faster. Continuous learning is less about chasing everything new and more about building on a solid base, in small regular steps.

Learning techniques that work

Learning research is unusually clear on a few points. A large 2013 review by Dunlosky and colleagues compared ten common study techniques. Two were rated as having high utility: practice testing and distributed practice (spreading study over time). Highlighting and rereading, the two most popular, were rated low.

Retrieval practice

Retrieval practice means learning by recalling information from memory, for example by answering questions, explaining a topic without notes or rebuilding a configuration from scratch, instead of reading it again. The act of retrieving is what strengthens the memory.

Roediger and Karpicke (2006) showed the effect, and its trap. Students who reread a text did better on a test five minutes later. But on tests after two days and after one week, students who had practiced recalling it remembered substantially more, even though the rereading group felt more confident. Rereading feels like learning because the text becomes familiar; familiarity is not the same as being able to produce the answer at 03:00 without the book.

Practical forms of retrieval practice in IT:

• Close the documentation and write down from memory the steps to do something (create a VLAN, restore a file from backup), then check.
• Before you read a page of this encyclopedia, write down what you already know about its subject; afterwards, check what you got wrong or left out.
• Explain a topic to a colleague, or write a short knowledge note about it. Gaps show up immediately.
• Rebuild a lab exercise without looking at the guide you followed the first time.

Spaced repetition

Spaced repetition is reviewing material at increasing intervals, for example after one day, three days, a week and two weeks, instead of cramming it in one session. Each successful review lets you wait longer before the next. Forgetting is fastest right after learning (the "forgetting curve", first measured by Hermann Ebbinghaus in the 1880s), and each well-timed review slows it down.

Flashcard apps schedule this for you, but the logic is simple. The sketch below uses fixed intervals: after a successful recall the card moves to the next, longer interval; after a failure it starts again at one day.

from datetime import date, timedelta

# Review gaps in days: each successful recall roughly doubles the gap.
INTERVALS = [1, 3, 7, 14, 30, 60]

def next_review(last_review, step, recalled):
    """Return (next date, next step) after a review."""
    if not recalled:
        step = 0                      # forgot: start again with a short gap
    else:
        step = min(step + 1, len(INTERVALS) - 1)
    return last_review + timedelta(days=INTERVALS[step]), step

# Amara's card "Which port does LDAPS use?", first learned on 1 September.
review = date(2026, 9, 1)
step = 0
results = [True, True, False, True, True]   # recalled at each review?

print("learned ", review.isoformat())
review += timedelta(days=INTERVALS[0])
for recalled in results:
    when = review
    review, step = next_review(when, step, recalled)
    outcome = "recalled" if recalled else "forgot  "
    days = INTERVALS[step]
    print(f"review {when.isoformat()}  {outcome}  next in {days:>2} day{'s' if days > 1 else ''}")

learned  2026-09-01
review 2026-09-02  recalled  next in  3 days
review 2026-09-05  recalled  next in  7 days
review 2026-09-12  forgot    next in  1 day
review 2026-09-13  recalled  next in  3 days
review 2026-09-16  recalled  next in  7 days

Spaced repetition suits facts you must recall quickly: port numbers, command options, exam definitions. It is less suited to skills like troubleshooting, which you learn by doing. Combine it with retrieval practice: the review itself should be a recall, not a reread.

Interleaving

Interleaving is mixing different types of problems in one practice session instead of doing a block of one type before the next. Twenty subnetting questions in a row train the method; a mix of subnetting, routing and VLAN questions also trains you to recognize which method a problem needs, which is what real tickets ask. The Dunlosky review did not rate it as highly as practice testing and spaced study, so treat it as a useful addition, not a replacement.

Deliberate practice, not hours

Deliberate practice is focused practice on a specific weakness, slightly beyond your current ability, with quick feedback on what went wrong. The term comes from the psychologist K. Anders Ericsson's research on expert performance. Doing the same routine task for years does not make you much better at it; practicing the parts you are bad at, and checking the result each time, does.

Ericsson's 1993 study of violinists is also the source of the "10,000 hours" rule, popularized by Malcolm Gladwell in Outliers (2008). Ericsson later called that reading "a popularized but simplistic view" of his work: 10,000 was an average for a group of elite players, not a threshold, and what mattered was the kind of practice, not the count.

What deliberate practice looks like in IT:

• You are slow at reading packet captures: take ten captures from a lab, write down what happens in each before checking with a colleague or a reference.
• Your scripts break on unexpected input: write the tests first for three edge cases, then the code (see Error Handling for the Python side).
• Your ticket notes come back with questions: ask a senior colleague to review five of them and point out what was missing, then apply that on the next five.

Building a home lab

A home lab is a small set of computers, virtual machines or network devices you run at home for learning and experimenting. Its value is that you can break things: misconfigure a DNS server, lock yourself out with a firewall rule, fill a disk, and learn how to recover, without hurting anyone.

It does not need to be expensive. Common starting points:

• One computer with virtualization. A laptop or old desktop with enough memory runs several virtual machines: a Linux server, a Windows client, a small firewall appliance. Free hypervisors such as Proxmox VE are popular for a dedicated lab machine.
• A single-board computer such as a Raspberry Pi for services that run all the time: DNS filtering, a monitoring server, a Git server.
• Containers for quick experiments with applications and their dependencies.
• Cloud free tiers for cloud skills. Set a budget alert first: a forgotten resource can cost real money.
• Network simulators (for example Cisco Packet Tracer) for routing and switching practice without hardware.

Give each lab project a goal and write it up, even briefly: what you built, what broke, how you fixed it. The write-up turns tinkering into learning (retrieval again), and it becomes material for a portfolio; see Building Your Career in IT.

Learning on the job

Most of what experienced engineers know, they learned at work. You can speed that up by treating daily work as material:

• Read closed tickets and changes from senior colleagues: how did they diagnose it, what did they check first?
• Read postmortems of incidents, including old ones. They are compressed experience; see Blameless Postmortems and Learning from Mistakes.
• Shadow a colleague during a change or an on-call shift, with their agreement, and ask why at the end rather than during.
• Ask questions well. Show what you tried and what you think; people are more willing to teach someone who has done the first steps (see Asking Good Questions and Gathering Information).
• Keep a learning log. A few lines a day about something new. It is retrieval practice, a source for knowledge articles and a record of growth.

Mon  Port-channel stayed down after CHG-3098: LACP mode was "on" on
     one side, "active" on the other. "on" does not negotiate, so the
     two sides never agreed. Check both ends before blaming the cable.
Wed  Tomasz showed how to read interface counters: input errors with
     CRC errors growing = physical layer (cable, SFP, duplex).
Thu  Read the PRB-0412 problem record: monitoring had no alert on certificate
     expiry. Question for Friday: do our switches' management certs
     have one now?

Choosing and keeping certifications

A certification proves to an employer that you passed an independent exam against published objectives. It is useful for getting interviews, especially early in a career or when switching fields, and preparing for one gives your learning a structure. It does not prove you can do the job; employers look for experience and practical examples as well.

Some guidelines for choosing:

• Start from the role. Look at ten job postings for the job you want next and count which certifications they mention.
• Vendor-neutral first, vendor-specific later is a common route: CompTIA A+, then Network+ and Security+, is often recommended as a starting path, followed by vendor certifications (Cisco, Microsoft, AWS, Linux) for the platforms you actually use.
• Pair each exam with practice. Build the lab exercises for every objective; an exam passed only from question banks is quickly forgotten.
• Ask your employer. Many employers pay for exams and study time when the certification fits your role. Ask before you book.

The topics of this encyclopedia are tagged with the exams they support (shown at the top of each page). For AI credentials, see AI certifications.

Renewal and continuing education

Many certifications expire. CompTIA certifications are valid for three years. You can renew by earning continuing education units (CEUs): credits for activities such as training, higher certifications or relevant work, recorded with the certification body. You can also renew by passing the current exam version or a higher certification.

Figures as of 2026, from CompTIA's renewal pages; check them before you plan. The fee covers the whole three-year cycle, not each year. Renewing a higher certification also renews the lower ones in the same series (for example, renewing Security+ also renews Network+ and A+). Other vendors have their own rules: some require a free online renewal assessment every year, others a new exam. Put the expiry dates in your calendar on the day you pass.

Communities, colleagues and AI tools

You do not have to learn alone, and you learn faster when you do not:

• Colleagues and mentors. A more experienced person who reviews your work and answers questions is worth more than any course. If nobody offers, ask someone for 30 minutes every two weeks; most people say yes. More on this in Building Your Career in IT.
• Communities. User groups and meetups (many cities have them for Linux, networking, cloud and security), forums, and open-source projects, where reading and answering questions is practice in itself.
• Study groups. Two or three people preparing for the same exam can quiz each other: retrieval practice with feedback built in.
• AI tools. An AI assistant can explain a concept another way, generate practice questions or review a lab write-up. It can also be confidently wrong, so check its answers against documentation, and never paste work data into a tool your employer has not approved. The rules are on Using AI at Work.

Making it sustainable

Continuous learning should not become a second job in the evenings. A few habits keep it manageable:

• Small and regular beats large and rare. Two or three sessions of 30–45 minutes a week, spaced out, fit the research above better than a weekend of cramming.
• Use work time where you can. Learning that benefits your employer is legitimately work; agree on time for it with your manager and put it in your calendar as a "schedule" item (see the Eisenhower matrix).
• One goal at a time. Studying for two exams while learning a programming language usually means finishing none.
• Accept not knowing. Nobody knows all of IT. Being able to say "I don't know yet, I'll find out" is a professional skill, not a weakness.

Goal:      pass Network+ by mid-January; employer pays the exam,
           2 hours of study time a week agreed with my lead
Per week:  Tue 30 min + Thu 30 min (work time), Sat 60 min (lab)
Method:    read one objective, then close the book and write it down;
           flashcards for ports and standards (spaced, 10 min/day max);
           lab exercise per objective, written up in the lab log
Check:     practice exam in week 6 and week 11; focus on weakest domain
Not now:   Python course (after the exam)