Slashdot Mirror

The Joyent cloud features a second layer of isolation. Sometimes you see this described as "double-hulled virtualization". The OS performance penalty to achieve this is low to non-existent due to the nature of BSD zones (hardened jails).

Joyent hybrid cloud: Triton Compute Service and Triton DataCenter integration

This is precisely the scenario that Joyent's technology exists to mitigate.

You think you're running Linux containers, but under the hood you've also got zones and ZFS snapshots.

There is a resource penalty involved in using a high-integrity file system like ZFS, (efficient copy-on-write requires extensive write-buffer coalescing) but it's often not a large one compared to the many gains in administrative security and ease.

The following post from Bryan Cantrill is worth a read, if only to know more about Samsung's motivations for buying Joyent.

Samsung acquires Joyent: A CTO’s perspective

Or an Illumos Zone.. Illumos being a fork of OpenSolaris.
Even better, you can run Linux inside an Zone on SmartOS, an Illumos distribution, via system call translation..
And Smartdatacenter+ sdc-docker = SmartOS based IaaS solution with docker support..
Smartdatacenter and SmartOS are made by Joyent, everything is opensource.
That's what they use to power their public cloud..
Truly an hidden gem, as other Joyent opensource stuff like manta..

Let me offer an even better suggestion. Think twice and three times before you open source anything. If there is anything in your code that goes against the ever-present SJW internet groupthink, prepare to have your reputation tarnished and you character impugned.

I would never open source anything in the current PC run amok environment. Or at the least, I would minify it.

You did remember to use the right pronouns, right? Right??

OpenSolaris is old and discontinued. OpenIndiana is a CDDL fork of OpenSolaris, rebased onto what's now called Illumos (http://illumos.org/), and is one of several Illumos "distros".

OpenIndiana was meant to be an answer to desktop Linux. It did not do especially well in terms of uptake, for reasons related to Linux's desktop results. However, there are a variety of other distros which are more server-oriented, and they are fairly popular.

They include for example SmartOS (used by http://joyent.com/ for multitenant hosting and for their own software development), OmniOS (used for mainly single-tenant hosting, and for software development http://omniti.com/), Nexenta (used for building large storage systems), and Delphix (a data storage service).

They all rely on the debuggability of Illumos (mdb, dtrace), virtualization (zones, now including Linux branded zones, crossbow, kvm), services (NFS and iSCSI in particular, also various others like SMB), OpenZFS, and a variety of other useful features, such as even under light use making enormous use of threading for parallelism and concurrency (and the threading systems scale well; OpenZFS alone typically uses a couple thousand threads, hundreds of thousands of mutexes, and many condvars, and all will go higher with load; other kernel subsystems can be similar).

It's fairly common for computer services departments in universities and laboratories and so forth to use e.g. an OmniOS server in front of a large storage pool, offering up iSCSI, NFS and other shares to clients, or alternative SmartOS in front of a large storage pool, offering up lightweight VMs to clients.

Oracle's Solaris has diverged from Illumos (and vice-versa). The key features are similar, but Oracle has been targeting much higher-end applications -- much larger and busier storage pools, especially ones which are very heavily random-acess (big Oracle databases are an application). Like Illumos, it can run very well on hardware with huge numbers of cores (including hyperthread-like cores). Unlike Illumos, it's not developed in the open (and is not open source), but it is well-supported enough that expensive contracts get you fixes and sometimes features quickly. Illumos has been slower until fairly recently, for reasons including the lack of ability to do a fully self-hosted build (it relied on nonstandard build tools), an idiosyncratic source code repository, both of which have now been changed in the past few weeks.

Has anyone considered Joyent's Manta ?

This is a distributed object storage with integrated compute.
Data is stored on a cluster of SmartOS hosts..
And processed directly on each host inside a OS container (SmartOS zone), no data movement.

Lot of APIs available: R, command-line, python, ruby, node.js etc..

Available on their cloud and as a on-premises commercial product, opensourced last November (simulteanously with smartdatacenter).

When you can run Docker inside Solaris Zones..
And the datacenter is viewed as an elastic Docker host.
Things start to become interesting..

Thanks to Joyent Triton

Yes, thanks to Joyent, who opensourced smartdatacenter the software they use to run their public cloud...

The hypervisor is SmartOS, it is based on Illumos (fork of opensolaris)..
It has ZFS, Dtrace, Zones (think virtualisation with bare metal performance), crossbow.. and KVM as they ported it..
You can run SmartOS instances inside Zones, and even Linux instances (by way of ABI translation), or any OS using KVM,
And even present you datacenter as an elastic docker host, as they implemented Docker API in SDC.. (sdc-docker aka Triton)

I'm currently evaluating it, so far I'm impressed..

Here are the github repository and the docs

Yes, thanks to Joyent, who opensourced smartdatacenter the software they use to run their public cloud...

The hypervisor is SmartOS, it is based on Illumos (fork of opensolaris)..
It has ZFS, Dtrace, Zones (think virtualisation with bare metal performance), crossbow.. and KVM as they ported it..
You can run SmartOS instances inside Zones, and even Linux instances (by way of ABI translation), or any OS using KVM,
And even present you datacenter as an elastic docker host, as they implemented Docker API in SDC.. (sdc-docker aka Triton)

I'm currently evaluating it, so far I'm impressed..

Here are the github repository and the docs

But just so you heard it from us: if this were the act of a Joyent employee, we would—to deliberately use a gender-neutral pronoun—fire them.

oh wait, this just a fork, not a gender-neutral pronoun issue...

A company who knows that their shit could be forked will either behave themselves

it depends.

for a company that's the primary contributor, the benefits of accepting external contributions doesn't balance out the gain. in fact there's significant overhead. of course there's a breaking even point, but in my experience it's way below 50-50. companies get a net gain when they are very minor contributors, and not so much after that.

that's not to say there aren't ancillary benefits. community good will, joint effort between companies, and, in the case of Google Android, a concession to business partners (we'll buy into Android, as long as we can part ways w/ Google and keep Android).

i don't have data for the Node.js Joyent relationship, but from reading this blog,
https://www.joyent.com/blog/br...

it sounds like Joyent basically ran the place.

I especially liked the link to "empathy is a core engineering value" though: http://www.listbox.com/member/...

Linked from: https://www.joyent.com/blog/th...

And if so, should not empathy extend throughout all levels of a learning organization, including between managers and subordinates? Everyone is learning stuff all the time, including about cultural changes. Firing someone rather than trying to understand the situation and the individual's motives more first and whether change is needed or possible does not seem "empathic". Perhaps that is the kind of thing you tend to learn after many years of experience being a parent or other long-term caregiver (including a long-term manager or mentor) when you see someone learn and grow and change over a long time?

Plus, as other comments suggest here, there is an assumption in this blog post that may ignore the possibility the issue was about consolidating minor changes rather than having them as individual commits. If this issue was deemed by enough of the community to be important, maybe a more systematic patch would indeed be in order? One tiny change is not much work, but it may set a bad precedent?

Also, it is not empathic to coworkers and the rest of a company and community depending on someone to fire that person without notice without reasonable review or attempts at remediation for a less than egregious offense (contrast with, say, someone accused of physically assaulting a coworker). The issue there is proportion and risk/harm assessment.

So, the response of "we would have fired him" seems too extreme in multiple ways.

I am all for meaningful diversity in workgroups, like discussed in this book:
"The Difference: How the Power of Diversity Creates Better Groups, Firms, Schools, and Societies"
http://www.amazon.com/The-Diff...

However, the problem with some of these "politically correct" initiatives or statements which seem on the surface to be helpful to promote "diversity" is that they can actually make workspaces more stressful for *everyone*. Someone can bully with the rules (or their interpretation) just as much, or more, than with a fist... Here is a website by psychologist Izzy Kalman that explores some issues related to bullying and truly creating happy productive workplaces by *really* emphasizing empathy and forgiveness and growth and free speech:
http://bullies2buddies.com/

Just think about it -- does everyone at Joyent now need to be afraid of getting fired if they check the word "he" into the codebase, even by accident? Or maybe by saying "he" accidentally as a meeting? There are potential unintended consequences of creating a different sort of hostile workplace climate, like many US schools are finding out these days as a result of "zero tolerance" policies (like biting a cracker at lunch to make it shaped like a gun can get you in deep deep trouble).

For reference, here is what makes for happy productive creative workplaces in general (Autonomy, Mastery, Purpose):
"RSA Animate - Drive: The surprising truth about what motivates people"
https://www.youtube.com/watch?...

Anyway, these are all complex issues about language, sex, management, control, gender roles, cultural change, recruitment, productivity, norms, and more. They are tricky to talk about or write about without seeming uncaring or inept because of various assumptions people make about the context or the people involved -- and the fact that none of us are "perfect" (and that perfection can be in the eye of the beholder based on priorities). It is sad to see such great software get mired in them. But I guess they are p

Such minimal corrections are a clutter that makes actual, important changes get lost in the noise. It's not that the change was wrong. It was that the usefulness of the change didn't justify creating the clutter it added on maintenance level.

Also:

  * 1. Read errors are reported only if nsent==0, otherwise we return nsent.
  * The user needs to know that some data has already been sent, to stop
  * him from sending it twice.

Is this comment sexist?

Is this something worthy of firing a talented expert (as the company blog suggests) over the above?

Do you have your priorities shoved so deep up your ass you really believe using correct gender pronouns in comments of your software is more important than having the code written well?

You think? You treat a core contributor like this and then wonder why he steps down and leaves? The best part is that when they announced his departure they're like "yeah, uh we totally respect him and his amazing contributions now please respect our wishes and stop bringing up the fact that we are a bunch of SJW tools who treated a major contributor with less respect than Linus Torvalds treats people who intentionally crap all over his code base."

I've shown this crap to coworkers who were interested in learning Node and their reaction was "W...T...F..." that's how they treat their community?

Unfortunately, there are so some real bigots in the FOSS community.

Node.js had one such thing pop up..

It was a relatively minor thing - a pull request was rejected because the documentation was updated to be more gender neutral. Unfortunately, the reasons given for said rejection were less than... honorable, and when it happened again, it devolved into a giant show of bigotry by the maintainer.

Perhaps that's why women don't want to go into tech - us geeks despite wanting to show we're more "educated" and learned, still have some real bigots among us who justify their positions in dubious ways. Hell, perhaps tech is no more evolved than construction labourers gawking and cat-calling.

See also this bit of related ludicrosity, wherein someone is threatened by being fired for refusing to merge stupid trivial political patches.

https://www.joyent.com/blog/the-power-of-a-pronoun

We use Joyent - very happy with their service. http://joyent.com/

They are now offering a refund or 5 years of hosting.

I'll probably take it and move on.

Are you serious? If they're really going for an IPO, as an early investor, you should get in on that.

Jason's BS math exposed

http://discuss.joyent.com/viewtopic.php?pid=241234#p241234

For those affected, they're now (as of today) offering 5 free years of continued service, or a refund on the original purchase.

They are now offering a refund or equivalent hosting services:
http://discuss.joyent.com/viewtopic.php?id=33682&p=9

They are now offering a refund or 5 years of hosting. I'll probably take it and move on.

Joyent has been using the term for quite a while now.

Absolutely agree that cloud hosting services offer significant economies over traditional hosting. While we're naming vendors, a more complete list of cloud vendors includes the following (and most offer a much fuller range of web hosting services than EC2!):

US: Amazon EC2, MediaTemple, GoGrid, Mosso, Linode, Joyent

UK: ElasticHosts, FlexiScale

The problem is that people keep trying to make web programming easier and more accessible. To do that you have to hide more and more stuff from the developer. More advanced guesswork needs to be made by the application framework.

One of the things Ruby On Rails fans always boast about is how few lines of code it takes to write an application. Assuming that RoR isn't ignoring the instructions that normally are done by other languages, it means that the RoR framework is doing it behind the scenes for the developer. When frameworks do this they usually have to be as generic and flexible as possible which can attribute to some overheard. At least from what I've seen.

If you wanat really good performance you can write your application in C/C++ but you're going to have to write in all the scalability infrastructure yourself.

Java web application servers are designed with scalability in mind so that it can be easy to make your application scale. There are also other frameworks that you can use to make developing your application faster as well.

I use my own home grown webapp framework. Back when I tested it with PHP4/Apache1.3/mod_php I was able to get better performance with Tomcat/Apache2 testing against applications that were doing basically the same thing. Actually the java application was doing slightly more work. At the time there were threading issues with using Apache2 with PHP and a lot of scripts weren't able to use it. I'm not sure what the current state is now if php modules have been updated to be thread aware.

To get PHP apps to run fast the best option is to use mod_php in apache. This is great if you have a VPS or dedicated server. If you're in shared hosting environment, mod_php runs every user's script using the same permissions and if you're not careful you could have some problems. One of my clients were on a shared hosting site and even though they don't use PHP another user's PHP script was able to deface all the pages on that server.

There are ways to try and isolate mod_php to only have write access in a particular directory but a script can be setup to ignore something like that. The best way would be to use FastCGI to run a separate instance of the PHP engine for each user under their own account. Though this will decrease the performance compared to using mod_php. Though PHP files seem to compile faster than JSP files, compilation only happens once and from then on acess time is very fast. When you're running a separate out of process instance of the PHP engine for each user, you're also going to be using more memory.

When RoR first came out, a lot of people were saying that it was a great way for prototyping applications but wouldn't trust the performance for production. I guess not much has changed.

A lot also has to do with the backend data store you're using and how it's scalability integrates with the application.

Joyent.com is one of the few places that are using OpenSolaris as a hosting platform. They provide Solaris Containers (Zones) for users. Similiar to other types of VPS options from other providers. Twitter was originally hosted at Joyent but due to performance problems they moved to a different provider. One of Joyent's claimed benefits is that they are an RoR shop and can help with Rails apps. So I don't think anyone can say that Twitter was just written incorrectly but people that didn't know what they were doing. It's a shame that Joyent got the bad press when Twitter left. Though from what I understand they were having some network issues at one point and a big problem when one of their Thumpers (Sun X4500) got corrupted. Joyent seems like a good platform run by some good people and what I'm basically saying is that I'm glad to see that Twitter's problems weren't Joyent's fault. I don't have any financial interest in Joyent and don't know any of them personally nor do I have an account with them. I was just looking into them but I wound up getting a dedica

We've been using Joyent and are very happy. I've used EC2 for a few things and I think Joyent is more economical for many applications.

There's the begining of this sort of thing in development now - Google have Gears, which provides a Javascript interface to a locally stored SQLite database. Try using Google Reader in offline mode sometime, it's the same application, and will synchronise any changes you make when you take it back online.

Joyent have also developed Slingshot, built on top of Rails, which allows you to provide your web app as an offline desktop application. Again, this all synchs up with the servers once you get back into range of a network.

I provide hosting, support and implementation services for SQL-Ledger (sql-ledger.com) which is also an open source financial application. Although I did not develop sql-ledger, I am developing a number of addons and patches which I am releasing to the community through my website (http://www.ledger123.com/). I selected the best possible platform (joyent accelerator http://www.joyent.com/accelerator/) for my hosting so that my customers could just forget about issues with security, reliability etc. Added round the clock support to customers and free support to community and result is not bad. Making enough to feed a full time staff of two and one part time sysadmin. But I had the luxury of taking a mature open source application and build business around it.

Jason Hoffman's talk at RailsConf also looked at scaling Rails apps. The slides are available in PDF format.

Well, here is one company that use OpenSolaris in production. http://joyent.com/ . Read their blogs at http://joyeur.com/ And they say fsck you if you aren't using ZFS.

"In addition to hosted tiers that offer larger storage capacities, Joyent also offers the Accelerator, an easy-to-install server appliance that allows you to run your own private instance of the Joyent Connector suite of web applications."

It's exactly what he was talking about.

You mean something like this?

An interesting idea. Something like this is already happening, but instead of Office, it's intent is to replace Exchange (although since it's aimed at very small companies, it's probably more along the lines of bringing a plug-and-play Exchange-style experience to those who can't afford it.)