A solution architect on one of our programmes asked me last spring what he should read. I sent him my enterprise architecture list. He came back three weeks later, polite about it, and said something I’ve been chewing on since: “Most of this is about convincing people. My problem is that I don’t know whether the design works.”

He was right and I was lazy. The EA reading list is a list about leverage, consensus and operating models, because that is what the job is. A solution architect’s week is not that. It is a design review where someone asks what happens when the event stream replays, a non-functional requirement nobody wrote down until it broke in UAT, and two teams who both believe they own the customer record. That work needs a different shelf.

So here is the list I should have sent. Five books, sequenced, with what each one actually fixes. The order matters — it climbs from the inside of the systems outward to the people who fund them, which is the same direction a solution architect’s career tends to travel.

Why the SA list isn’t the EA list

The honest difference between the two roles is altitude, not seniority — I’ve made that argument at length in enterprise architect vs solution architect. The reading follows the altitude.

An enterprise architect can get a long way on judgement, vocabulary and political patience. A solution architect cannot, because a solution architect gets found out. You stand in front of engineers who have read the source and say “this will hold at ten times the volume,” and within two quarters everyone knows whether you were right. That exposure is the best thing about the role and it sets the reading priority: technical depth first, organisational subtlety second. Reverse that order and you become the architect whose designs are diplomatically impeccable and operationally naive.

1. Designing Data-Intensive Applications — the spine

Designing Data-Intensive Applications

· Full review →

Start here and do not negotiate. Almost every solution design that falls over in production falls over at the data layer: two systems holding the same fact and disagreeing, a retry that isn’t idempotent, a “real-time” integration that is actually a nightly batch with better marketing.

What Kleppmann gives you is the vocabulary to have those arguments before the design is built. Read-your-writes. At-least-once versus exactly-once. Change data capture as a first-class integration style rather than a database trick. The chapters on derived data are the ones I’ve used most — once you see that most enterprise integration is a badly specified derived-data pipeline, a whole category of vendor claim stops working on you.

The concrete change in your week: you stop accepting “we’ll keep the two systems in sync” as a design statement. You ask which is the source of truth, what the replication lag budget is, and what the business does during the window when they disagree. Nobody has ever thanked me for that question in the meeting. Several people have thanked me afterwards.

Buy the print edition. You will annotate it, and it is the one book on this list you will still be using in ten years.

2. Software Architecture: The Hard Parts — for the decomposition decisions

Not on my affiliate shelf, so take this as a plain recommendation: Ford, Richards, Sadalage and Dehghani’s Software Architecture: The Hard Parts is the most directly useful book I know for the specific work of pulling something apart. Service granularity, distributed transactions you cannot avoid, what to do when the data wants to be in one place and the teams want it in four.

Its real contribution is method. It refuses to give you answers and instead teaches trade-off analysis as a discipline: enumerate the options, name the forces, write down what you’re giving up. That is the single most transferable skill in solution architecture and the one most often replaced by preference dressed up as principle.

Read it when you are mid-decomposition and the arguments have started going in circles. Read DDIA first, though — The Hard Parts assumes the data vocabulary and will feel thin without it.

3. Building Evolutionary Architectures — for designs that have to change

Building Evolutionary Architectures, 2nd ed

· Full review →

The second edition is the book the first one was trying to be, and the reason it belongs on an SA list rather than an EA list is the fitness function.

Most architecture governance is a document that says what ought to be true. A fitness function is an executable check that fails the build when it stops being true. The difference is the whole ballgame. If you are a solution architect, you have almost certainly written a target architecture that decayed within a year because nothing was enforcing it and you weren’t in the room for the next forty decisions. Fitness functions are how you stay in the room without being in the room — cycle checks on module dependencies, automated latency budgets, a test that fails if a service starts reaching into another’s database.

This is also the book that has transferred best to agent-generated code. When a large share of the changes arriving in your repository weren’t written by a human who absorbed the design intent, the only architecture you actually have is the architecture you can assert automatically. I’ve written about that pressure in architecture governance at agent speed.

The concrete change in your week: your design documents start shipping with tests.

4. Team Topologies — because most “architecture problems” are boundary problems

Team Topologies

· Full review →

Here is the pattern I see most often. A solution architect is asked to fix an integration that keeps breaking. The integration keeps breaking because two teams own halves of a flow with no agreed contract and no shared on-call. The architect proposes a technical fix — a message broker, a canonical model, an anti-corruption layer — and the problem reappears next quarter in a new costume.

Team Topologies is short, slightly repetitive, and gives you the language to name what’s actually wrong. Cognitive load as a design constraint. The four team types. Most importantly the three interaction modes, because “are these two teams collaborating or consuming a service?” is the question that determines whether your interface should be a conversation or an API with a version policy.

I’m careful about the failure mode here: it is very easy to read this book and start redrawing the org chart, which is not your job and will not go well. Use it diagnostically. When you can say “this design requires continuous collaboration between a stream-aligned team and a platform team that is staffed for X-as-a-service,” you have turned a vague delivery worry into a statement someone with budget can act on.

5. The Software Architect Elevator — for the altitude shift

The Software Architect Elevator

· Full review →

Last, and only last. Read this when the technical work is no longer the thing holding you back.

Hohpe’s elevator is the metaphor that explains why your excellent design didn’t get funded: you pitched it in engine-room language to people who live on the top floor. “Architects sell options” is the framing I’ve stolen most shamelessly — it reframes every “why are we spending two sprints on abstraction we don’t need yet?” conversation into a price-of-optionality conversation, which a finance-minded executive can actually evaluate.

It is also the only book on this list that is a genuine pleasure to read, which matters when you’re doing it after work.

For a solution architect this book does one specific job: it converts you from someone who is right into someone who is believed. If you’re finding that your designs are technically sound and keep losing to worse proposals with better narratives, this is the gap.

The EA-track reading — only if you’re eyeing the floor above

Everything above makes you better at the job you have. Two additions only matter if you want the job above it.

Enterprise Architecture As Strategy (Ross, Weill and Robertson) is the one to read if you’re seriously considering the EA track — the operating-model diagnostic turns “what should our architecture be?” into “what kind of company are we?”, and that substitution is most of what enterprise architecture actually does. I’ve reviewed it here, and the rest of the EA-specific shelf is in best EA books for 2026.

The framework material — TOGAF and friends — I’d leave alone until someone pays you to care. If your organisation runs on it you’ll absorb enough by osmosis. If you’re chasing the certification for a job spec, that’s a different decision and I’ve argued it separately in is TOGAF certification worth it.

Do not read the EA books early as a signal of ambition. The fastest way to become an enterprise architect is to be an undeniably good solution architect, and the fastest way to be that is the first three books on this list.

Bottom line

Buy Designing Data-Intensive Applications and read it properly, over months, with a pen. Buy

Building Evolutionary Architectures, 2nd ed

next and make one fitness function real in your current codebase before you finish it. Read

Team Topologies

in a weekend the next time a technical fix fails for organisational reasons. Save The Software Architect Elevator for the point where your constraint is persuasion rather than correctness.

That’s a year of reading, not a week, and it’s roughly the order in which a solution architect’s problems change shape.

Solution architect reading list — quick answers

What should a solution architect read first?
Designing Data-Intensive Applications by Martin Kleppmann. Most solution designs fail at the data layer — duplicated sources of truth, non-idempotent retries, replication lag nobody budgeted for — and this is the book that gives you the vocabulary to catch those problems in design review rather than in UAT.
Is the solution architect reading list different from the enterprise architect one?
Yes, and the difference is altitude. The EA list leans towards operating models, strategy and influence. The SA list leans towards systems depth, trade-off analysis and making designs enforceable, because a solution architect is held accountable for whether a specific thing works. Read the EA material when you're genuinely considering that move, not before.
Do I need to read TOGAF material to be a solution architect?
No. Framework literature is worth your time if your organisation runs on it or a job spec demands the certification. It will not make you better at sizing a service boundary or specifying an integration, which is the work you are actually judged on.
Why is Team Topologies on a technical reading list?
Because a large share of problems presented to solution architects as integration or platform problems are team-boundary problems wearing a technical costume. Team Topologies gives you the language — cognitive load, the four team types, the three interaction modes — to name that and propose an interface rather than another broker.
How long should this reading list take?
About a year if you do it properly. Designing Data-Intensive Applications deserves months and a pen. Team Topologies is a weekend. Building Evolutionary Architectures is best read slowly enough that you implement a real fitness function in your own codebase before you finish it.