Thanks a lot, Yugen.

I was in Erlangen and I spent a couple of years here and I can tell you this was the

most challenging place but also the best place for me to be.

You can see the rate of graduation in the group of Professor Taish.

Everybody sitting here, if you are a student of Professor Taish, you may know that you

probably have 60% or 70% to become a professor after you graduate.

Normally, I just wanted to come in Erlangen again to reconnect and then to say hi and

maybe just sit back, relax and get something, wine and so on.

But Professor Taish is always very challenging, so he found a way to give me a work.

So I had to do this talk here today.

The title of my talk is about the work that we have been doing recently.

It's about hardware isolation.

This hardware isolation, we use it in embedded system, we use it in desktop, we use it in

the cloud and we believe that this is going to play a big role in the future.

Why did we came to this?

I will explain that later on.

But first, I wanted to draw a connection between the invasive computing and the cloud computing.

Because when the request came for me to give a talk, it was like, do we do a multi-processor

or multi-core?

I said, no, we do the cloud.

But what's the connection between the cloud and then these multi-core?

I will explain that and you will see that the connection is very, very evident.

And then FPGA accelerated cloud computing, the need for that.

Why do we need cyber security mitigation in those cloud and then some trend in isolation

and then I will explain the work that we've been doing.

So first, invasive computing and then cloud computing.

So what is the connection between those two fields?

Elasticity.

So cloud is usually characterized, but the infrastructure as a service in the cloud is

characterized by elasticity.

You can grow, you can shrink.

Depending on whatever you request in terms of resource.

So this is a picture that describes the invasive computing where you have tasks spreading around

and then occupying resource in various granularity and various configuration.

And then at some point when those tasks are done, what they do is that they shrink and

then they release those results so somebody else can use them.

So this is elasticity.

And in cloud computing, we do the same.

Basically the infrastructure that we have in cloud computing is set at a system level

where we have, so the most important part, this is the picture that describes the open

stack.

You have the neutron that takes care of the network.

Then the nova is in charge of managing all the computational resource.

So those are the two most important.

And then we have some switch for the storage and then the silo meter to compute how much

you're computing.

So all the constraints are built into the silo meter here that charge you if you cross

a certain boundary.

We may not be seeing that when we do invasive computing, but we have to define some form

of costs here.