How do search engines search their entire cache for terms in .1 seconds?

How do search engines search their entire cache for terms in .1 seconds?

It's obviously exobytes of data, how could it be possible?

Attached: 1547387375625.jpg (480x480, 32K)

because it's not actually that many results.
Go to google and search for cheese. It'll come up with some bullshit number like 600 million results found, but the results cap out at page ~35 with less than 400 results displayed.

google exponential growth

>cheese
pizza?

no just cheese

how do they get away with this

yeah but is google searching it's whole cache for "cheese" for you or not?

searching though fifty million terabytes of data should take a few mins no matter what hardware you have

ha ha ha that IS a Jow Forums reference!

There's place for a better search engine than Google for specialized web forensics

such as

it is the power of billions of dollars

now google hashtables instead of cheese

habeeb it

Google literally has well published and documented papers on this. I fucking don't understand why you would ever ask on Jow Forums when this places is just full of software engineer larpers

Algorithms.

hashtables

God I wish that were me

The girl or the bag of spaghetti?

you can't hash everything, only certain things

>spaghetti without sauce
but for real, I think this is cute

Attached: 84cc3e7c620f75370057719171aac408.jpg (500x500, 43K)

No it doesn't search it's whole hash, why would it? It's already got weighed results to display on the first page, only something like 20? The secret is to have results ready to respond with before someone searches for it. Everything is already sorted and ready to go when you press enter. They don't actually go through 6*10^20 results because they don't have to. They just have to fill up the first page. If you go to the second page then you display the next 20 results in the sorted cache for cheese. If you have multiple keywords in your search its basic algebra to produce a new 20 results.

Unironically image related. I know it's a buzzword, but it's true.
Instead of designing your data to be vertically scaling (one large powerful machine with a few replicas), you structure the data over hundreds of thousands of machines.

ScyllaDB (Cassandra alternative written in C++) is the closest we have to Google's BigTable (that's also FOSS).

Attached: 1_iTUNwaZ5qHfJ4K_fzhQ6aw.jpg (960x340, 42K)

Attached: Google Lies.jpg (734x646, 124K)

(((they))) are merely pretending, in reality everything you see is hand picked to support narrative

It matches a pattern and ranked only for the page you requested.