Search the practice set
275 questions, 991 terms and 600 topics in 30 areas.
56 results for “Offline-First”
Local First Write
Committing to local storage first and treating the server as an eventual replica, which makes the interface responsive and connectivity an optimisation.
First-Principles Reasoning
Reducing a problem to the physical, mathematical or economic facts it rests on, then reasoning up, rather than reasoning from analogy or convention.
Spec-First Development
Writing and reviewing the API specification before implementing, so the contract is designed deliberately rather than emerging from code.
Weighted Shortest Job First
Sequencing work by dividing the cost of delaying it by the effort to deliver it, which maximises value delivered per unit of time.
Airbnb's Service-Oriented Migration
Airbnb decomposed a large Rails monolith by first extracting a unified data-access layer, so that services were built on owned data rather than on shared database tables.
Assumption Excavation
Deliberately surfacing the unstated beliefs behind a design or a requirement, to test which are constraints and which are merely habits.
Bottom Line Up Front
Placing the conclusion and the requested action in the first lines, on the assumption that most readers will not reach the end.
Graph Database
A store whose first-class citizens are nodes and the relationships between them, making multi-hop traversal cheap.
Hedged Request
Sending a duplicate of a request to a second replica after a short delay and using whichever response returns first, to cut tail latency.
Instagram's Early Scaling
Instagram reached tens of millions of users on Django and PostgreSQL with a handful of engineers, by deliberately choosing boring technology and doing the simple thing first.
Instance Family
A group of instance types sharing a resource profile — general purpose, compute optimised, memory optimised, storage optimised, accelerated — chosen by which resource the workload…
Netflix's Recommendation Architecture
Netflix splits personalisation into offline, nearline and online layers so that expensive computation happens ahead of time and the request path stays fast.
Priority Queueing
Classifying requests by business importance so that overload sheds the least valuable work first rather than an arbitrary slice.
Query Plan
The database's chosen strategy for executing a query, and the first thing to look at when one is slow.
Schema Compatibility Mode
The rule stating which schema changes a registry will accept, which encodes whether producers or consumers are expected to upgrade first.
Service Catalogue Entry
The record that makes a service a first-class object in the platform — owner, dependencies, docs, runbooks and health in one place with a single identity.
Try-Confirm-Cancel
A three-phase distributed transaction where each participant first reserves resources, and a coordinator then confirms or cancels all reservations.
Zero Touch Provisioning
A device obtaining its identity and configuration on first boot without a human configuring it, which is the only approach that scales past a few hundred units.
eBay's Architectural Generations
eBay rewrote its core platform several times across its first decade, each time because the previous generation had hit a limit that could not be tuned away.
gRPC
A contract-first RPC framework using Protocol Buffers over HTTP/2, with generated clients and servers and first-class streaming.
You are designing a field-service application for engineers who work in basements and rural sites with no reliable connectivity. What is the architecture, and what are the decisions that will bite later?
The shape: local first, with the server as an eventual replica Connectivity is an optimisation, not a precondition. The application writes to a local store, ren
A critical CVE is announced in a widely-used library. Walk me through the first four hours.
Hour 1 — determine exposure Query the SBOMs across the estate , including transitive dependencies. This is the moment that justifies having them: without, this
A team is launching a new service and asks what its SLO should be. How do you help them decide, and why is "99.99%" usually the wrong first answer?
What the interviewer is testing Whether you treat reliability as a cost benefit decision with a budget, or as a virtue to maximise. Why not four nines It costs
A user in Mumbai types your URL and presses enter. Walk me through every hop until the HTML reaches their browser. Where would you look first if the page were slow?
What the interviewer is testing This is the classic architecture screening question, and it is asked because the answer's depth correlates strongly with real ex
An internal AI assistant gives confidently wrong answers. The team wants to upgrade to a better model. What do you check first?
Establish whether the model ever saw the right content Log the retrieved chunks alongside each answer, then take the wrong answers and check: was the correct so
An organisation is starting cloud adoption. Three teams want to deploy next month. What must exist first, and what can wait?
What must exist before anything reaches production Account structure. Separation by environment and by workload, because an account is the strongest isolation b
Anomalous access to a customer database is detected. Walk me through the first day, and say what determines whether you can answer the regulator.
The first hours Declare an incident and assign command. Named commander who does not debug, operations lead, communications lead, scribe. Security incidents add
Delivery feels slow and leadership wants a productivity initiative. You propose mapping the value stream first. What will you map and what do you expect to find?
Map every step from request to delivered change, with two numbers each Work time — how long the step actually takes when someone is doing it. Wait time — how lo
Knight Capital lost roughly $460M in 45 minutes in 2012 after a deployment reached seven of eight servers. Which architectural failures made that possible, and which one would you fix first?
The case, as publicly reported Per the SEC's 2013 order, Knight Capital deployed new order routing code to eight production servers ahead of a NYSE programme la
Maersk rebuilt roughly 4,000 servers and 45,000 PCs in about ten days after NotPetya in 2017, and recovered its directory only because one data centre had been offline during the attack. What does this say about DR design?
The case, as publicly reported In June 2017 the NotPetya malware — destructive rather than financially motivated — propagated through Maersk's network, encrypti
You are asked to lead a migration of 300 applications to cloud. What happens in the first ninety days?
Days 1–30: discovery, and the foundation in parallel Discovery from evidence, not interviews. Network flow logs for dependencies (the only trustworthy source),
You are asked to produce a technology radar for a 200-engineer organisation with no current standards. How do you build the first one?
Start from what is actually in use, not from what should be Inventory the estate: languages, frameworks, datastores, queues, CI tooling, observability. Sources
You are asked to restart an enterprise architecture function that was disbanded two years ago after being seen as a blocker. How do you approach the first six months?
Understand what it did that got it disbanded The pattern is consistent: it reviewed everything, decided slowly, produced documents nobody read, and became a que
You are forming a platform team of four to serve twelve product teams. What do you build first?
Not a platform. One golden path. Four people cannot build an internal developer platform for twelve teams. They can make one journey excellent, and expand from
You are modernising a fifteen-year-old monolith with the strangler pattern. Which slice do you extract first, and why?
The criteria, in order Low coupling to shared state. The extraction cost is dominated by data, not by code. A slice that reads and writes tables twelve other mo
You are the first architect at a 120-engineer company with no architecture function. What do you do in the first ninety days?
Do not start with governance A review board in month one, from someone with no credibility and no context, gets routed around and defines the role permanently a
You are the first architect at a 200-engineer company with no architecture function. What do you do in the first three months?
Understand before changing anything The fastest way to become irrelevant is to arrive with recommendations. Nobody has any reason yet to believe them. Listen wi
You inherit an estate of roughly 400 applications, no reliable inventory, and a mandate to reduce cost and risk. What do you do in the first ninety days?
What the interviewer is testing Whether you can sequence work at portfolio scale, and whether you go for evidence before strategy. This is the enterprise archit
Your organisation is preparing for its first SOC 2 audit. The security team is asking engineers for screenshots of configurations. What would you change, and what is the architectural argument?
What is wrong with screenshots Three things, and each is worth naming separately. They evidence a moment , not a period. The auditor is assessing whether contro
Your organisation receives its first subject access request. Nobody knows where the person's data is. How do you respond within thirty days?
Start the clock formally and verify identity first Record the date received. Verify the requester's identity robustly — responding to an impostor is itself a di
Your service will exceed capacity by 30% during a known peak. Do you shed load or brown out, and how do you decide what goes first?
The decision Brown out first; shed only if that is not enough. They are complementary rather than alternatives, and brownout is strictly less harmful when it is
Your system handles 1,000 requests per second today. Marketing says a campaign will bring 10,000 next month. What breaks first, and how do you find out?
What the interviewer is testing Whether you reason about bottlenecks systematically or start adding servers. The shape of the answer Scaling is not uniform. Som
A field application has been live for three months and support reports that engineers occasionally lose completed job reports. What do you investigate?
Suspect last writer wins on a record level merge The pattern fits exactly. Two writers touch the same record — the engineer's device and the office, or two devi
Your platform has 60 internal APIs. Teams repeatedly rebuild capabilities that already exist because they cannot find or understand them. What do you build?
The problem is discovery, then comprehension Duplication is the symptom. Teams are not choosing to rebuild; they cannot find what exists, or cannot tell in reas
Offline-First
Treating connectivity as an optimisation, with a local store as the source of truth.
First-Principles Reasoning
Reducing to physics, mathematics and economics, then reasoning up.
Caching for Performance
Layer choice, hit ratio as a first-class metric, and cold-cache recovery.
Degradation Modes
Deciding in advance what is shed first and what is protected.
Device Provisioning
Getting identity and configuration onto hardware at manufacture or first boot.
Horizontal vs Vertical Scaling
Scale out for stateless, scale up first for stateful.
Presenting to Executives
Decision first, cost, risk, and what happens if we do nothing.
Rendering Strategies
Client, server, static and incremental rendering, and what each costs on first paint.
Stress Testing
Pushing past target to learn what breaks first and how it fails.
gRPC APIs
Contract-first RPC, generated clients and protobuf compatibility rules.