What is Object Storage?


技术77 阅0 评

原视频地址:What is Object Storage?

字幕 & 翻译(待):

If we think about object storage,

which is kind of a new-ish tier,

"object storage",

the idea is for relatively

low performance storage at a

relatively low cost

that is designed to serve the needs

of the Internet workload.

And so by Internet workloads, I'm

talking about Web applications.

I'm talking about website and

website hosting and delivering

content across the Internet,

but also a way to find

a new home for all of that data

that we used to store on tape.

Right. We have to have a way to

take big volumes of data,

write that down, put it somewhere

that is safe, that is secure,

and where we can hold on to it for a

long time for regulatory reasons,

legal reasons and other

requirements that make it

necessary for us to keep that data

for a long amount of time.

So let's start off with,

"what is an object?" and

why do we call it "object storage"?

So, the concept

in object storage is that you have

an object, right.

Now, that object

can be any kind

of file.

There aren't really any file

restrictions or anything like that.

There are some characteristics that

are very important.

But as a general rule, any

kind of file can be an object.

So, we'll throw

an object in there now, there

are 4 essential components

to that object that make

it usable in the computing sense.

The first one is for every

object, we have to have an ID,

we have to have some sort of a

unique identifier that lets

us know what this

object is when it comes time to

retrieve it. That's the first piece.

The second piece is, obviously,

you've got to have some data.

What's the point of having a file

and all of this information about it

if you don't have the data itself?

So this could be an Excel file

and could be a video file.

It could be an audio file, any kind

of file. But you've got to have the

actual data itself.

The third piece that you

have to have is

metadata.

So what is metadata?

Metadata is everything that

you need to know about

this file and about the

data itself.

Who created it?

When was it created?

What is it used for?

What is the file type?

How large is it?

The metadata is what makes it

possible to go out and find

it, search it, index it

so that you can bring it up and use

it whenever you need to.

Because, again, all that's being

displayed out right now is the ID.

The ID is not terribly useful.

And then the last thing that you

have to have are your attributes.

Now, attributes are related to

metadata. They aren't exactly the

same thing, but they're close.

Attributes can be, "are certain

users are allowed to override

it?", "are certain users are allowed

to download it?", "are certain users

are allowed to delete it?".

So the attributes are about the

object itself rather than about

the data.

But again, those are those 4

essential components.

This is what you use

when you actually create something.

So you've got your object here,

right.

And then what do you do with it?

Well, in the world of object

storage, which is unstructured,

you take all of your objects and you

drop them into a bucket.

So let's do this on this

side. So, imagine

I have a bucket, and

my bucket can have a handle, why

not?

And so into my bucket, I'm

going to put all of these objects

and now this is the neat part: these

buckets can be as big as you

need them to be.

They can scale to hold billions

of objects.

Now, it's not always a good idea to

put billions of objects in it, but

they can, in fact, scale to that

size.

And then what's really neat about

object storage - because remember,

we want to be able to use this for

cold storage, too, right, we need to

put things in it that can never be

deleted - is whenever you,

the user, you're going to interact

with this thing via an API.

Right. So you're not ever going to

interact with the bucket

individually here.

We'll throw you in.

There you are.

You're using object storage.

It's great.

So you're going to take your

objects. You're going to drop them

into this bucket. Now, on the back

end, the magic of object

storage is your object

here. Call it "O".

It is going to replicate out of this

bucket, the bucket is a virtual

construct, but we're going to take

it and we are going to say, all

right, well, I have

3 physically

separated devices.

And I'm going to take a copy of your

object and I'm going to replicate

it out into all 3 of them,

and my purpose in replicating

them out into all 3 of those

locations is data integrity,

data security.

That way, if the physical device

here, right, each of these buckets

is going to reside on a physical

device. So we'll call this P1,

P2, and P3.

So we're going to replicate it out,

your object is going to be on 3

physical, different devices, so if

there's a hardware problem or a

network outage or whatever in P1,

that's OK. You've still got copies

in P2 and P3, you can still get

the data that you need.

Now, when we get into

the details of object storage, all

of a sudden you start saying, man,

there's a lot of complexity in this.

Right? So not only do I have

the object buckets, but the way

that it's priced and metered and

build is kind of unique.

And people are asking me about how

many "gets" and how many "puts" and

how much storage am I taking up.

And the metering and billing around

object storage, everyone that sells

it has their own different metrics

on it. But here's the important

parts to consider.

The first is what

is your responsiveness or your

response time?

Right. So from a price perspective,

the higher performance that you want

and this is true for all cloud

storage, but the more performance

you want, the more it's

going to cost.

If you have data that you

have packaged up and it's data that

for 20 years your company has been

taking and writing it to tape,

sending that tape to vaults and

storing it in vaults where it will

probably never be seen again.

That can go to the very coldest of

cold tiers, right.

And that is going to be the least

expensive.

And so we're going to call that

"cold times 10".

That is the coldest of cold tiers.

Up from that, where you need

to get to it every so often, but you

don't really need to get to it much,

that's going to be a two dollar sign

one, right? We're going to charge a

little bit more for that.

We're just going to call that

"cold".

Maybe this is data that you need to

get to once every six months,

something like that.

But maybe you're earning a website

and you are using these object

storage buckets to host all of the

assets for your website.

Obviously that is going to cost

a little bit more, right, because

you are sending data out on a

regular basis.

So that's going to get our three

dollar signs.

And I'm going to call this

one called "cold-ish".

It's not really warm because

the concept of a warm here doesn't

really exist an object storage, but

it's cold-ish.

And so this is

videos, it is pictures,

it is physical files,

and what makes object storage really

neat is on the back end through

the magic - because remember, you,

you're interacting via an API,

you're not interacting with these

things directly - you

can do all kinds of nifty

front-end things for it.

And that's the really neat part of

object storage and that's the

benefits of it.

So let's talk about some of them,

right?

Hosting videos for streaming over

the Internet.

Perfect.

Perfect use case for object storage,

because like I said before, we can

replicate these things out.

Right. So for your video streaming,

rather than just replicating it into

three buckets that are all in the

same place, let's replicate it

out globally. Let's replicate it out

so that we are hosting those videos

not just in their primary location,

but in three different locations

in seven different countries around

the world, decreasing the latency

for your users so that their

streaming experience is better.

Let's talk about another good use

case: file

sharing. So let's imagine

a company where you're collaborating

on spreadsheets and PowerPoint, and

things like that, and you've got

collaborators that work all

over the world.

Well, for versioning

purposes, you can have

different versions that overwrite

the previous version as long as

those attributes allow for it.

And so if I've got my

friend Eric, who is working

in London, can create a file,

I can get a copy of that file,

make my changes to it, upload it

back in. And I've got version two.

That's the metadata.

And then I can have a colleague

in Singapore, Sam, who downloads

that file, make some more changes to

it, uploads it back in version

three, again, incrementing the

metadata.

But that allows for this Cross Globe

collaboration on a single file type,

and it makes everybody's lives

easier. But we're doing it at a very

low cost, much less expensive than

trying to send files

around via email because emails have

to get stored in a much more

expensive storage tier, much

less expensive than FedExing thumb

drives around.

And again, being able to be done

in a secure manner because we have

control over who can see and access

these objects.

Other Web hosting, obviously, is

going to be important. We talked

about regulatory data before, the

regulatory data and cold storage.

Let's say that you involved in a

digital archiving project

and you're working with a university

or a library and you're taking

digital photographs of manuscript

documents. And you want to host

those out for everyone in the world

to be able to see.

Object storage, perfect place to

store that kind of data - because

again, it's not getting accessed

very often.

It is write

once and then read many,

many, many times and you can host

it in a secure manner.

So hopefully you've enjoyed this.

This has been a good overview of

object storage.

As always, if you have any

questions, please feel free to leave

them in the comments. I'll do my

best to answer, or some of my

colleagues at IBM will.

And if you have any other questions,

just let us know.

Thank you so much for your time

today. If you have any questions,

please drop us a line below.

If you want to see more videos like

this in the future, please do like

and subscribe and let us know.

And don't forget, you can grow your

skills and earn a badge with IBM

Cloud Labs, which are free

browser-based interactive Kubernetes

labs that you can find more

information on by looking below.

Thanks again.

最后更新 2022-02-17
评论 ( 0 )
OωO
隐私评论