Fixed capacity meets growth
A fixed-capacity structure meeting growing cardinality breaks the design.
What it is
The general principle: a fixed-capacity structure meeting growing cardinality breaks the design. The fix is never a bigger fixed capacity. It is a fallback path, just like widening a transaction ID or giving an undo log an overflow rule.
From Memory & Resource Management › Interning and dictionaries
At or below α = 1: a fixed cache size meeting a growing corpus gives hit-rate decay with no code change and no config change. The cache that gave 85% last year gives 78% this year, because the denominator grew. This is the general principle "a fixed width meeting exponential growth fails silently", and here it is nasty because nothing in the system reports it. The only symptom is a slow rise in backing-store load, which gets blamed on traffic.
From Caching › The popularity exponent
Where it appears
In 7 mechanisms. Each says what the mechanism is about, then how this principle shows up there, in its own words.
- Indexing
When does a separate "where is it?" structure pay for itself, and what does it cost to keep it true?
Quantization's primary function is to move an index back across that threshold. 100 million 768-dimensional vectors are 100,000,000 × 3,072 B ≈ 300 GB raw and about 10 GB quantized.
Read it inHigh dimension: exactness becomes unaffordableWhy indexes exist and when they win
- Memory & Resource Management
Who is allowed to hold how much memory, for how long, and what happens to the work when the answer is less than it asked for?
The fix is never a bigger fixed capacity.
Read it inWhat spilling actually costsString interning and dictionary reusePlan for running out of memory
- Caching
What happens when we deliberately keep a second copy of data that no protocol forces to be correct?
A TTL charges for freshness in proportion to the size of the cache, not to the churn of the data.
Read it inFreshness boundingThe popularity exponent and the phase transition at α = 1
- Cost Mechanics
What is metered, and what is the smallest increment you are charged for?
This is a design whose sign flips at a scale boundary that it crosses silently.
Read it inChatty requests, needless recompute and cross-region calls
- Streaming
What does it cost to remember, when the input never ends?
max_parallelism is fixed at job creation and caps scaling forever.
Read it inKeyBy and key groupsStateful operators and the state-size arithmeticComplex event processing
- Observability & Operations
What is the smallest description of a running system from which we can find a fault, fast enough to matter, at a cost we can still pay while the system is failing?
The operational test is severity-to-action.
Read it inAnomaly detection and alert fatigue
- Security & Privacy
How do you protect data from someone who watches your design and then chooses how to attack it?
A privacy budget is consumed by reading and cannot be refilled.
Read it inAuthorization modelsDifferencing and differential privacy
Related
Made on the same pages most often.