Caching - Web Development Udacity English Japanese

Caching - Web DevelopmentYouTube↗ IA↗(long shot)
Education 12.1K views 102 likes 0 dislikes 23 Feb 2015

Udacity 651K subscribers United States YouTube↗

Click to Play Video

Caching refers to storing the result of an operation so

キャッシュというのは

that future request return faster. Basically, if you do something once

将来同じリクエストを受けた時に速く応えるため処理の結果を保存することです

you know whether it's a database query or rendering some

例えばデータベースクエリや

HTML or you know, anything that might be slow. You store

HTMLのレンダリングなどは動作を遅くさせます

the results so you don't have to do the computation

その結果を蓄積しておけば

a second time instead you can reference the previous result. So,

次に同じ処理をする時に参照できるのです

when do we cache? We cache things when the underlying

ではいつキャッシュをするのかというと

computation is slow, when the underlying computation will run multiple times.

計算が遅い時あるいは計算が何度も実行される時や

When the output of said computation is the same for

ある計算の入力に対して出力が同じ時などです

a particular input, so that we know you know, we don't

そのような時は毎回計算し直さなくて済みます

have to recompute it every time, because it's going to be the

なぜかと言うと

same every time; the output of this computation. And another good

その計算結果がいつも同じだからです

reason you know, for, for when we cache, when you are

また他にも理由があります

hosting provider charges for DB access, which applies to you right

仮にあなたがホスティングサービスを利用しているとします

now. Google APP engine, gives you a fixed number of reads

例えばGoogle App Engineでは 1日に決められた回数データベースに対して

and writes to the data store in a particular day and

読み込みと書き込みをすることができ

if you go over that, you have to pay for it. Even

超過すると代金を払う仕組みです

if your app doesn't get a whole lot of traffic Caching request so

アプリケーションが何度もキャッシュのリクエストを受けず

they don't have to hit the database over and over is a

データベースへのアクセス回数を

fine way to save some money. And that will be an example we

減らすことができればお金の節約につながります

start with shortly. So, whenever you have a situation where you have

これについてはあとで話すことにします

the slow computation that you're running

計算が遅くて

multiple times, with the same output over

常に出力が同じであれば

and over, you should cache it. You should store that result somewhere

キャッシュして結果をどこかに保存しておけば

else. So that you don't have to run a computation over and over.

同じ計算を何度もしなくて済むわけです

So let's, let's talk about how that algorithm looks. Let's say

ではそのアルゴリズムを見てみましょう

we have a function called db_read(), and this reads from the database.

データベースから読み込むdb_read関数を使います

And it's slow, you know, let's say it takes 100 ms to

この処理に仮に100ミリ秒かかるとします

run this. This query which is slow for a database query but

データベースクエリとしては遅い方で

not unheard of. And, and you're serving thousands of requests, you know.

サイトでは大量のリクエストを処理しなければなりません

If every request that comes in to, to your website has to

その際リクエストごとに

hit db-read and that takes 100 milliseconds, that means you can only

100ミリ秒かけてdb_read関数を実行していたら

do ten requests a second or you start, you know, you start

1秒で10個しか処理できません

doing multiple requests at the same time and your database starts to get

また複数のリクエストを同時に処理した場合

pummeled because it's trying to do this complicated query. All at the same time

データベースは複雑な処理をすることになり

and, and this 100 milliseconds maybe turns in to 200 or 300 milliseconds,

1つの処理に2～3倍かかるもしれませんし

or 500 milliseconds, who knows, you know. When your database is under a load

5倍かかる可能性もあります

it starts to really get angry at you. So, if we wanted to

データベースに負荷がかかると機嫌が悪くなるのです

cached this db read, let's talk about

そこでdb_read関数を使って

what this algorithm would look like. Prefer

キャッシュ機能を追加しましょう

it a kind of write in, in sitor code here, we will have

疑似コードを書いてみます

something that looks like this. If the request from making is in the cache,

リクエストがキャッシュにある時は if request in cache:とすると

return the cache version of that request, and in this case,

キャッシュ版のリクエストが返されます

I'm pretending cache is, is like a dictionary and the request

ここではcacheを辞書として扱っていて

that we're making is the key into this dictionary. And that's

requestは辞書のキーとします

generally the structure of the cache, the cache is basically a large,

キャッシュは基本的にこのような構造で

like a large hash table a large mapping of, of keys

大きなハッシュテーブルや値に対するキーのマップと言えます

to values. Now you know all about hash tables a hash table

以前説明したハッシュテーブルですが

works perfectly well for this. So, if a request is in

今回のようなケースに最適です

the cache. Returned Cache value of that request, Else start the value

リクエストがキャッシュにあれば値が返されるのです

of this DB Read in a variable, put that variable

else節ではdb_read()の値を変数に入れ

in the cache for future look ups. And then return that

キャッシュに保存すると将来同じリクエストをした時その変数が返ってきます

variable. So basically, instead of calling DB Read on every

つまり毎回db_read()を呼び出す代わりに

request, the first thing we do is check to see if

キャッシュを確認して

that request is in our Cache And if it is,

同じリクエストがあればその値を返します

we return the cached value. This is called a cache hit.

これをキャッシュヒットといいます

And only if this request isn't in our cache do we

つまりクエリを返すのはキャッシュにリクエストがない場合だけで

actually return our query. And this is called a cache miss.

この状態をキャッシュミスといいます

And what we do on a cache miss is we actually do

その際はリクエストを処理して

operation, and store the result of the operation in our cache, and

結果をキャッシュに保存してから結果を返すことで

then return that result. So future requests will just bounce off the

将来同じリクエストを受けた時に備えます

cache. Hash. Now if you're using a hash table, depending on the

ハッシュテーブルを使うと処理の時間は

size of that hash table. We're going to probably do a lot better

そのサイズにもよりますが

than 100 milliseconds, we're probably going to be talking about less than 1

100ミリ秒どころではなく1ミリ秒以下になるでしょう

millisecond. Which would be you know, quite a big speed improvement. Now

これはスピードの面で非常に大きな改善です

of course this hash table is huge and you're caching lots of things.

しかしハッシュテーブルは巨大ですし

You know, hash table have their own Performance characteristics that you'll

何度もキャッシュするので処理能力が

have to take into account. But you know, you can get

問題になってきます

substantial improvements just by taking the slow pices of your code

そこで処理を遅くするコードを修正すると

and wrapping this simple algorithm around it, which is the focus of

パフォーマンスに大きな改善が見られます

our next quiz. What I'd like you to do is implement

では次の小テストでPythonのキャッシュアルゴリズムを

that caching algorithm here in Python. So, what I've given you

実行してみてください

is a function called Complex Computation. And this takes two numbers

2つの数を足すcomplex_computation関数を使うと

and it adds them up. But it's going to take half a second

処理に0.5秒かかります

to do so. The time dot sleep function causes Python

time.sleep関数を用いると

to just sleep or wait for however much time you want.

指定した時間だけ処理を待ってくれます

In this case, I have it sleep for half a

今回は0.5秒と指定します

second. So, what I want you to do is use a

皆さんにやってほしいのは

dictionary as a cache. And it's defined right here as,

ここに定義したcacheとして辞書を使うことです

called cache. I want you to make this function cache computation.

その際cached_computation関数を使って

Use this cache to store the results of calling complex

入力ごとにcacheにcomplex_computation関数の結果を

computation on different inputs. So if we pass in the same

格納できるようにしてください

inputs a and b twice, I want the second running of this function to use the

そしてaとbという入力を使って 2回この関数を処理しますが2度目は

cache. The first one, we'll do the actual

キャッシュを使ってください

computation. And we'll be grading this by running picking

このアルゴリズムを評価する際は aとbの異なる値を使います

different values of a and b, seeing how long the first one takes to run and then

そして1度目と2度目の実行時間の差に

seeing how long the second one takes to

注目します

run. And the second one should run substantially faster.

2度目の方がずっと速いはずです

Similar videos:

Locking and Memcache - Web Development
Udacity
5.6K views 16 likes 0 dislikes 27 May 2012100% match

Making Links
Udacity
53.9K views 21 likes 0 dislikes 23 Feb 2015100% match

Italics Solution - Web Development
Udacity
5.7K views 7 likes 1 dislikes 23 Feb 2015100% match

Port - Web Development
Udacity
15.8K views 18 likes 0 dislikes 23 Feb 2015100% match

Italics
Udacity
49.1K views 16 likes 2 dislikes 23 Feb 2015100% match

Best Browser - Web Development
Udacity
14.9K views 7 likes 0 dislikes 23 Feb 2015100% match

Web 简介
 Udacity
25.7K views 60 likes 0 dislikes 23 Feb 2015100% match

Course Outline - Web Development
Udacity
46.2K views 169 likes 0 dislikes 23 Feb 2015100% match

Submitting Input - Web Development
Udacity
6.8K views 5 likes 0 dislikes 22 Apr 2012100% match

Best Browser Solution - Web Development
Udacity
9.6K views 22 likes 1 dislikes 23 Feb 2015100% match

Paragraph Tag - Web Development
Udacity
8.6K views 4 likes 0 dislikes 23 Feb 2015100% match

Precomputed Caching - Web Development
Udacity
12.2K views 80 likes 0 dislikes 27 May 2012100% match

Shopping List Take 1 - Web Development
Udacity
841 views 1 likes 0 dislikes 23 Feb 2015100% match

HTTP Responses - Web Development
Udacity
5.5K views 19 likes 0 dislikes 23 Feb 2015100% match

Correct URL - Web Development
Udacity
10.9K views 6 likes 1 dislikes 23 Feb 2015100% match

Adding Images
Udacity
12.9K views 36 likes 0 dislikes 23 Feb 2015100% match

Types of Databases - Web Development
Udacity
8.5K views 43 likes 4 dislikes 25 Apr 2012100% match

App Server Architecture - Web Development
Udacity
14K views 52 likes 5 dislikes 27 May 2012100% match

Introduction - Web Development
Udacity
23.3K views 107 likes 0 dislikes 23 Feb 2015100% match

User Input - Web Development
Udacity
7.7K views 3 likes 0 dislikes 22 Apr 2012100% match

Query Parameters - Web Development
Udacity
22.6K views 78 likes 3 dislikes 23 Feb 2015100% match

Part 2 - Web Development
Udacity
643 views 2 likes 0 dislikes 07 May 2012100% match

Caching Techniques - Web Development
Udacity
34.4K views 216 likes 7 dislikes 23 Feb 2015100% match

Google App Engine - Web Development
Udacity
10.1K views 13 likes 3 dislikes 17 Apr 2012100% match

Rot13 - Web Development
Udacity
9.3K views 46 likes 8 dislikes 23 Apr 2012100% match

Document Structure
Udacity
7.2K views 24 likes 1 dislikes 23 Feb 2015100% match

Security - Web Development
Udacity
5.7K views 64 likes 0 dislikes 23 May 2012100% match

App Server Scaling - Web Development
Udacity
28.2K views 277 likes 8 dislikes 23 Feb 2015100% match

Cookie Headers - Web Development
Udacity
4.2K views 23 likes 1 dislikes 23 Feb 2015100% match

Datastore Types Solution - Web Development
Udacity
656 views 1 likes 0 dislikes 25 Apr 2012100% match

Settings Privacy About Contact Donate