Tip:
Highlight text to annotate it
X
I'm first going to set this up,
we're going to set up this BayLISA meetup,
we're going to have it here in the office on SmartOS.
I tried to let her down easy,
that actually not too many people are going to be interested in this.
This is an operating system, this is pretty esoteric stuff,
maybe we'll get a handful of people.
She kept at it, "No, we need to get an agenda."
I'm like, "Deirdré, we're kind of making a big fuss
about something that not a lot of people
are going to be interested in."
We threw it out there with a "limit" of 60 people,
like "oh yeah, a limit of 60 people, like, don't all rush in,"
and the next day all the slots are gone.
I'm like "wait a minute, what?"
And then we added like another 30 slots,
and they were all gone.
So we weren't actually sure if we were being put on
or if there was actually this many bodies.
We did actually take a room to 20 degrees below zeros
at like 2pm, thinking that you might actually all show up,
so it's great to see so many of you here.
You've definitely proved me wrong certainly
about the amount of interest that we have in SmartOS
just here in San Francisco.
So I just want to briefly describe SmartOS
and me or Peter will introduce the rest of the speakers.
In terms of Joyent, which is where we are,
the fundamental belief of Joyent,
the way you can most concisely summarize Joyent
is Joyent believes that the operating system still matters.
We are tackling the fundamental challenges of cloud computing:
multi-tenancy, density, observability, storage, network virtualization.
These are not all unique to cloud computing,
indeed we don't think any of these is unique to cloud computing.
We think that this is just computing and we believe, like you believe,
like a lot of folks believe in the broader illumos community
that damn it, the OS actually still matters,
in fact it matters more than ever before.
I think that- exactly, that's right. I think the other illumos variants,
just Leo from OmniOS in the back there.
We got to really get a lighter up there.
I actually think, to that point,
this is something that we in the entire illumos community,
not just SmartOS but OmniOS Illumian, OpenIndiana
and the other distros,
this is what we all emphatically believe.
This is the belief that actually binds us together,
that the OS still matters.
SmartOS embodies that belief.
A year ago, it's only been a year, it's kind of amazing,
we made SmartOS available with the announcement of KVM,
the illumos-based KVM. Since that time
we've deployed it into production
and many of you have deployed it into production as well.
We've deployed tens of thousands of VMS.
We've got great confidence in KVM,
which is kind of DTrace, once you know how it works
it's hard to get complete confidence in
because you know the number of kittens
that need to be slaughtered
every time you exit and enter the VM.
But it does actually work, thank God.
But it's not just KVM we've added in SmartOS.
We have taken the fundamental differentiators of illumos
and we have pushed them forward: Zones, DTrace, ZFS,
and the stuff that is actually-
the little stuff that is big stuff for those of us in the community,
little things like mdb and libumem.
In fact we're going to dive into libumem in great depth
tonight with Bob Mustacchi's presentation.
We love all application runtimes.
I would say we're application runtime-agnostic
but that's not quite true.
We love everybody, so prepare to be fed into that view or not.
Node.js we actually really enjoy here.
We see a lot of promise in node.js and enjoy its own platform
but we're pushing forward.
We are in particular focus on making SmartOS
the premier platform for node.js apps.
Out of curiosity, how many of you in the room
have deployed node on SmartOS?
Because I know there are at least a couple of you.
Right. So really only you guys.
I know you guys, okay, anyway.
A lot of folks have and it's great to see,
and it's great to see enthusiasm from the node community
for an operating system.
We'd like to think that we are teaching folks in the OS,
we are learning about these high level runtimes,
and folks in these higher level runtimes
are learning the power that you can get
from a leading edge operating system,
in particular introducing technology like DTrace, mdb, libumem.
I've been apologizing a lot for the core kernel in mdb.
Explaining the history of mdb doesn't tend to buy us much time
but we try to do it anyway.
So in terms of the SmartOS community,
we've been- surprised is not even the word, shocked.
When we made this available as a download,
when we announced KVM,
we kind of did that as just like good clean living,
we didn't actually expect anyone to download it,
but a lot of you did download it.
In fact a surprising amount downloaded it.
We don't have the numbers as of this second,
but in the first couple of months
there were tens of thousands of downloads
and that pace has continued. Even more surprising,
and I don't want to sound too surprised,
like it's not production grade because of course it is,
but it was a little stunning how many people
would just download stuff off the internet
and deploy it into production apparently.
We actually had one guy early on who had these on IRC,
had a KVM issue.
I could not reproduce this issue, didn't make sense.
I asked him in private chat:
"hey, can I have a couple of minutes on this machine?"
He says "Yes, yes, sure, hop on this machine".
Alright, so he gets me like some station
and I hop on to the machine.
And I'm like: "Can I just understand a little about the machine?
And he's like: "Yes, if you just be careful with it,
it's in production." I'm like-,
"it's in production?" Yes, yes, yes.
Then he gives me a website that you would all recognize
from a brand that you would all recognize.
Like "Go here". I'm like: "Alright."
"Oh yes, I'm there, I'm on the website."
"Yes, you're executing in that VM with this pit over here."
"Oh God!" "Do you know what you're doing?
Are you sure? He's like "Yes,
should I not be running it in production?"
"No, no, no, no, it's cool, it's cool, it's cool,
just like, wow, alright."
This was like eight weeks after we made it available,
so it's like "Whoa, this is exciting!"
So in particular the bug that he had on his system,
he had a bug where he couldn't spin up a new VM.
The reason he could not spin up a new VM
is because he had half a terabyte of physical DRAM on that box.
That's an expensive box. Half a terabyte of physical,
we hadn't tested it on a machine,
we don't have a machine with that much physical, sorry,
I'm embarrassed to say.
We hit the sweet spot a lot lower for us in the cloud,
we had 100-200 gigabytes. It was great, we fixed the bug
and he's happy running in production,
although it's scary as hell.
But no, it's totally production grade.
I want to make that absolutely clear, I'm emphatic about this.
It's just that that's crazy.
This has been a huge surprise for us.
I think one thing that I just want to apologize for a little bit,
there's been so much interest, it's been a bit overwhelming
and sometimes we haven't been able to get to people
as much as we want to.
So if you feel like you've been using it,
you've been reaching out, and we've been ignoring you,
it's not because we don't love you.
It's just because we're so surprised that you love us
and we're trying to find a way to return the love.
One of the things we've been doing,
Robert Mustacchi, who will be presenting later,
has kicked off office hours in #smartos on Freenode.
That's been terrific I think,
that's been a good way for people to know
we're going to be there, we're going to be answering questions.
If you've got ideas about how we can help the community,
we're all ears, because we're very humbled
by the enthusiasm of this community
and we want to serve it the best way we possibly can.
So please let us know if there are things that we can do
to help you, with our apologies
for not being able to help you as much as we'd like.
I think a tweet is worth a million words in some regards.
It just doesn't seem like more than a day or two passes
before you get these things.
As an OS developer, you're not really accustomed
in kernel development to generating enthusiasm.
Generally it is the absence of failure that constitutes success
for an operating system.
And people don't walk in and say: "Oh wow,
thank you operating system, you stayed up all night.
You didn't corrupt any of my data."
Hold on, listen, staying up all night
and not corrupting any of your data,
that's a cloud of work actually.
But that's fine, that's fine, don't thank me for it.
But no one gets excited about that stuff,
so for us it's very weird, "oh sweet mother",
this is not something that people utter
when they talk about their operating system
unless they just panicked with all of their production data
because they just downloaded this thing off the internet
two weeks ago and they're now running it in production.
But no, this is just great,
This is an amazing amount of interest
and we're really excited about it.
So we put together a night for you tonight
that we think is going to be pretty exciting.
For the recruiters in the room,
we're going to take you to Croschev tonight so prepare.
You're earning it tonight, I've got to tell you
because we are going to take you-
you're going to go through some deeply technical presentations.
You are not allowed to let them leave.
Yes, exactly, that's right.
This is going to be like a Clockwork Orange,
like "oh my God, it's assembly, oh God!"
If you have something for the slide, we could.
Robert may add that in time for his presentation.
I just one want to thank our sponsors, Telefonica.
It's a great joint partner,
has actually spun up a Joyent-based cloud,
a Smart beta center based-cloud in Spain.
I think I've got their slides here so I'll just-.
Lest you think I'm just shilling here, they paid for the beers,
so pay attention folks. Exactly. Thank you Telefonica.
They're instantservers.es offering.
And Alex, I don't know if you want to say anything
about Telefonica? Alex is here.
Yes, I'm a global product manager for Telefonica
on the instant servers product.
We're working closely with Joyent
and we're talking to lots of people who are running in Joyent clouds,
SmartOS in the US, are interested in deploying in London.
We're going to be live in London probably in a month or two.
We do have a live cloud in Spain at instantservers.es.
It's actually run by a subsidiary of Telefonica.
Can you get them as a US citizen?
You can get them as a US citizen.
Can you upgrade Brazil? You could do some good there.
We're working on it though we may do it via Miami.
Drink the beer please and listen to Bryan and his team, this is rare.
And I can tell you from the Joyent perspective,
we've got a technologist to technologist relationship
with Telefonica and it's been great.
There's a great share interest there and shared enthusiasm.
The other thing I want to *** out
is Joyent's @brendangregg.
We've put together some DTrace training that's upcoming.
If that's something you or your company is interested in,
this is going to be useful.
This is not going to be just for SmartOS.
This is going to be useful on FreeBSD, on Mac OS X
and damn in, some day on Linux.
So definitely check that out
and Deirdré I know we've got links to that.
Sorry, I don't have a link here.
I think if people go to SmartOS.org they can definitely find it.
Come up to us, Brendan is just outside.
He's the guy who needs a haircut out there
so you can just go find him.
So with that, Bill, why don't you take us away
on the SmartOS zpool architecture.
Thank you. For more information visit www.joyent.com or www.smartos.org.