Hacker Newsnew | past | comments | ask | show | jobs | submitlogin
If I was your cloud provider, I'd never let you down (joyent.com)
289 points by chillax on Oct 23, 2012 | hide | past | favorite | 164 comments


With the lifetime account fiasco still ongoing [1], they should probably shut up.

[1] Joyent/Textdrive sold lifetime shared hosting accounts for a one time payment in the beginning. Those accounts would exist "as long as we exists". A couple of months back they sent an email that they are "discontinuing" lifetime accounts. After uproar from the community they offered refunds. Many people (including me) agreed to the refund, but then they decided to make another 180 degree turn, not paying out the refund. Instead they are now spinning off a new "Textdrive" that is supposed to "take over" the lifetime customers. No details about funding/general outlook/etc. of this new company has been provided so far. Questions to this end are shrugged off as "everything is perfect, just trust us, ..."

Incidentally, the new Textdrive Forum (discuss.textdrive.com) seems to be down at the moment...


"lifetime" "unlimited" "forever" etc products should always been looked at sceptically for that reason. Anything that is either finite or requires up-keep cannot be offered on these terms, it is just impossible.

I actually avoid unlimited products on purpose because I think a lie is a bad way to start doing business together. Be it web-hosting, broadband, or anything else. I would prefer they just be up front and then we both know where we stand (no ambiguity).


What Joyent forgot about was that by getting rid of the lifetime account holders (I was one) they were effectively silencing their greatest word of mouth sales. I would always recommend their services to others, and I would purchase many of their other products/services as they came out.

But by pulling out the rug under me[1], it left me distrustful of them. So let's assume that I got my money's worth, but what Joyent has now lost is the word of mouth. Even worse, instead of not telling people about Joyent, I actively turn them away from Joyent.

While Amazon goes down from time to time, at least I have less unknowns with them. With Joyent, I don't know how they are going to fuck me at the last minute when they decide to pivot for the umpteenth time.

[1] Not once, but thrice. First, they cancel the lifetime accounts, forcing me to quickly move to another provider (thank you GAFB and Heroku). Then they say I can get a refund for the service, which when I asked for it they first denied that they gave it out and then told me that the offer was rescinded because the Textdrive service was coming out. Third, they keep calling me telling me that I need to move, even though if they spent 2 seconds checking for account activity they would see that no domain routes to them anymore, nor is any data stored with them.


Same here - I'll never recommend them, and have actively discouraged people from using or investigating them.

They may be great technically, but I always mention their spotty/inconsistent and poor treatment of me as a customer. Even if I'm not the most lucrative customer, how you treat every customer is important. You never know who, dispute being a low-revenue personal user, is a high-revenue business user.


> "lifetime" "unlimited" "forever" etc products should always been looked at sceptically for that reason. Anything that is either finite or requires up-keep cannot be offered on these terms, it is just impossible.

When a company is offering "lifetime", "unlimited", "forever" et al, the most likely reason is it is cash strapped and is hoping to raise some cash with these lucrative schemes. I, as a customer, am well aware about "no free lunches" but we are entering an agreement when I buy your supposedly "lifetime" plan. When you do an 180 later because the initial money you got from me isn't enough to keep me running forever, it isn't something unplanned that happened to you. You always knew that. It's just that you are hell bent on screwing me.


Reminds me of the lifetime free flights that were offered in the past: http://www.economist.com/blogs/gulliver/2012/05/lifetime-air...


They still honor those passes, except in the case of people clearly abusing them.


How can you abuse "lifetime, unlimited"? I want a plane ride, I book a ticket. I cannot book multiple simultaneous tickets. So abuse can only be defined as "They used it more than we'd anticipated."

That school of thought is very much like the oversubscribed phone lines of yore, which made the phone companies highly intolerant of long modem sessions.


This kind of sentiment is why everything nowadays comes with 20 pages of fine print that nobody can wade through. It's not enough just to expect people to be reasonable — everything must be clearly spelled out in excruciating detail, as people will feel free to do any insane thing that comes to their mind if you do not restrict it explicitly, and they will loudly object if you tell them that what they're doing is not reasonable. I'm not sure it is a much more desirable state of affairs.

(Clarification: I'm specifically addressing the idea that there can be no "abusive" use of a lifetime offer. Obviously the TextDrive customers are not in this category.)


A contract does not provide for what is reasonable. It stipulates the relationship between two parties. if you say unlimited personal use, with no right of resale, then I can fly all day every day without worry. I don't see how flying a lot on an unlimited lifetime ticket can be legitimately called abusive.

If I buy a lifetime, fly anytime ticket, and use it all the time, you don't get to call that abuse. If I use the maximum physical quantity of something, for which I have purchased an unlimited supply, that is not abuse. If I am in breach of contract, then I am in breach of contract, but that's not abuse. Abuse implies that you don;t like what I'm doing but I'm not in breach of contract.

Calling something abuse generally means that I've found a use of the service that you didn't forecast for and despite being contractually acceptable, it means you aren't getting the advantage you thought you would from the contract. Expecting me to not use the service to my full advantage is naive at best.


Yes, that's my point exactly. By claiming the absolute right to force upon the counterparty anything they have not explicitly forbidden by contract, you require them to create absurdly lengthy and unnecessarily expansive agreements that enumerate every possible circumstance and give them almost unlimited leeway in the event of a dispute. And then people complain that EULAs are too long and make unnecessary power grabs.


What part of "a contract stipulates the relationship between two parties" did you not understand?

It goes both ways!

That is exactly what contracts are meant to do, and have always been meant to do. And in order to use them successfully, they don't need to be "absurdly lengthy" or anything. The boilerplate you're referring to generally has nothing to do with "abuse" or "being reasonable", and most of it doesn't even need to be there, because if you'd leave it out, it'd be implied.

What you do have to do, however, is stick to the unique terms that you did write in your contract. And if you cannot, don't put them there, or don't offer the contract.

Simple as that. If you offer people unlimited phonecalls, you should be prepared for them to be on the phone 24/7. If you're not, you really shouldn't be offering them unlimited phonecalls, instead maybe offer them 20,000 minutes per month (that's about half of 24/7).

And if "twenty thousand minutes per month" doesn't sound as snazzy as "unlimited minutes", that's too bad because the reality of the matter is that you apparently cannot afford to actually offer "unlimited minutes".


Take a look at the contract that you've entered into with an airline when you buy a regular ticket. These guys are in the pantheon of crossing every t and dotting every i, going do far as to say if we declare something an act of god, then you get no refund, nor recourse.


IIRC, some people were selling their free companion ticket--you could make pretty decent money flying someone along with you whenever you flew. And that was definitely prohibited in the original agreement.


Selling the companion ticket falls outside of my definition of "reasonable use".

The article I read said they were annoyed that some of these guys would actually use the tickets to, you know, fly places, all the time. As if they had unlimited access.

Did they really expect to sell a $400k item so somebody and for the buyer to fly three times a year?


"Rothstein made 3,009 reservations in less than four years, almost always booking two seats, but canceled 2,523 of them."

Sounds like he was not actually using the ticket to, you know, fly places.


So what? How far ahead of the flights did make the reservations, and how far ahead did he cancel?

Are we really saying that the handful of wealthy people who bought unlimited tickets are to blame for AA's problems, and not the fact that AA appears to have been bad math?


People were making booking on 3-4 flights per day, on the same route, and just showing up to the one that suited them best on the day.


Order a ticket plus companion ticket to London for every other day of the month. Sell the companion ticket on craigslist for $200 and ferry people you've never met to Europe.


I never read anything that talked about this as part of the problem. It was always "We didn't expect them to use it so much."

That's an awful lot of strangers to make transatlantic flights with at 200 bucks a pop, and it would take quite a long time to recoup the multi-hundred-thousand dollar investment in the companion ticket. Hardly seems worth the time for somebody who can afford half a million for a couple of lifetime passes.


Click through and read the LA Times article.

When bond broker Willard May of Round Rock, Texas, was forced into retirement after a run-in with federal securities regulators in the early 1990s, he turned to his trusty AAirpass to generate income. Using his companion ticket, he began shuttling a Dallas couple back and forth to Europe for $2,000 a month.

"For years, that was all the flying I did," said May, 81. "It's how I got the bills paid."


OK. so there was a guy. I still contend that contracts must stipulate the entire business relationship. If the contract did not forbid accepting money to fly people in his companion seat then fire the lawyer. If it did, then he's in breach of contract, and they can cancel, and sue for damages.


> "They used it more than we'd anticipated."

If you read the Times article you'll see some pass holders sell tickets, book the same route on different flights (in case they get delayed in traffic) and book two seats in order to have more space. It's not just a case of flying too much.


If the upkeep is smaller than the cost of capital to the company, such an arrangement probably remains mutually beneficial. Given that the cost of computing resources was falling enormously and continues doing so, I can see why TextDrive expected this arrangement to be profitable for them. Note that pensions and annuities, arrangements with terms best described as "pay $X/month for as long as I live", exist.

(Of course, they pivoted out of the hosting business and found that supporting a one-off machine is an expensive headache. But it's not impossible for such an arrangement to work.)


If as a business you decide to offer a lifetime plan, it is assumed that you have accounted for the up-keep of the service. Joyent should be held responsible for not holding up their end of the deal.


Lifetime provision, or its counterpart: flat rate provisioning (unlimited X for $5/mo) puts the customer into an antagonistic relationship with their provider:

http://blog.kozubik.com/john_kozubik/2009/11/flat-rate-stora...

If your needs for service are serious, you should avoid such arrangements.


I actually just posted (before I saw this thread) about how to roll your own online storage using a BRIC (Bunch of Redundant Independent Clouds). By striping data across different clouds, you spread the risk of any one storage provider letting you down.

http://news.ycombinator.com/item?id=4689238


So, what do you think about CrashPlan. A very much known, sort of trustworthy business? They offer unlimited accounts too.


Why not sue them?


As far as I understand one would have to sue in small claims court in California to get anything out of it. Since I'm in Europe, it unfortunately is not a cost-effective option. (I think I paid ~500$ for the original plan, so it's cheaper to write it off)


> so it's cheaper to write it off

That's what they're counting on.


Probably. Unfortunately it's still true: I'm not going to spend thousands of dollars and lots of my time to sue them in California.


Because that is a very poor solution to most problems in the world.


As much as I hate to admit it, the threat of a lawsuit seems to be helpful in keeping many people and corporations in line.


Most people tend a very negative view towards suing. Personally, when you're in the right and have a strong case, as the above seems to have, it works beautifully.

I've had the written threat to sue work, as well as the actual suing. It's not as expensive, or time consuming, as you think and is the primary method of contract enforcement. It's also kinda fun and an interesting intellectual experience.


Red herring.

Besides, as far as I understand, Textdrive is in the process of being relaunched, honoring all the lifetime plans.


> Red herring.

Here is a company with a history of making tall claims and then doing an 180. How is pointing that out a red herring?

> Besides, as far as I understand, Textdrive is in the process of being relaunched, honoring all the lifetime plans.

From what I know of the story, if they do honor lifetime plans, it isn't out of the goodness of their hearts. They had every intention of flipping the lifetime planholders, and they did, and what they are doing now is damage control.


> Here is a company with a history of making tall claims and then doing an 180. How is pointing that out a red herring?

That isn't particularly relevant to service quality and stability, is it? Maybe it's just me, but I try not to pay much attention to PR bullshit, since they have to make bold claims in order to succeed.

p.s. that their blog was down for the last three hours is somewhat ironic, though.


> Here is a company with a history of making tall claims and then doing a 180. How is pointing that out a red herring?

>> That isn't particularly relevant to service quality and stability, is it?

The claim being made is unlike Amazon, they won't let reddit down. It's relevant to point out these guys have made tall claims in the past and failed to deliver. I won't do business with someone who let's sales swindle customers. Even if the engineering is awesome, there is no guarantee sales didn't sell you something they couldn't deliver.

As far as service quality and stability goes, the only way to evaluate it is to actually switch.

> Maybe it's just me, but I try not to pay much attention to PR bullshit, since they have to make bold claims in order to succeed.

It's possible to do PR without bullshitting. You can always publish your actual numbers viz. number of customers, downtime history, avg issue resolution time. If you are resorting to bullshit, that means you don't have actual goodies.


Whether you fundamentally trust your provider to try to do what they agreed 'isn't particularly relevant to service quality and stability'?

Did you really mean that?


The whole lifetime plan debacle was a really big blunder from PR standpoint, but I wouldn't go as far as to call it a rip-off. Deal that started in 2004-2006 and lasted through to 2012 is 6 to 8 years worth of hosting. As far as I understand, they also offer one more year of free hosting as a compensation, which makes it 7 to 9 years. 7 to 9 years of hosting for $500 is okay in my books.

Of course, that doesn't make their PR blunder look any better, they should've just swallowed the costs, but I don't 'fundamentally distrust' them based solely on this case.


> Deal that started in 2004-2006 and lasted through to 2012 is 6 to 8 years worth of hosting. As far as I understand, they also offer one more year of free hosting as a compensation.

You might have to excuse some here who actually understand lifetime and unlimited to mean more than 6 to 8 years.

> but I don't 'fundamentally distrust' them based solely on this case.

But do you see how others would though?


Maybe not a rip-off, but still a lie.


> > Here is a company with a history of making tall claims and then doing an 180. How is pointing that out a red herring?

> That isn't particularly relevant to service quality and stability, is it?

Not from a technical aspect, no. But it is a very important consideration when looking into any service level agreement that they offer to back up the technical claims. If something goes wrong, will they walk away from the promises made in such an SLA as easily as they thought they could walk away from the lifetime accounts? Unlikely I'm sure, but certainly not impossible and the lifetime thing sets a precedent that is difficult to completely ignore especially if you are talking about the sort of money Reddit must spend on hosting related matters.


Well when you make a decision to use a third party, typically the ethics and makeup of that party are evaluated. You can ignore it at your own peril. But IMO it's not very wise.


Failing to live up to your service commitments is exactly what quality and stability are about.


The problem is that I have no more trust into them honoring an agreement.

I'd very much like to keep my lifetime account (I rarely use it for anything, but when I need it, it's there, it's paid for and I can just use it). Problem is, the new Textdrive was essentially launched out of trying to avert this PR debacle, and neither Joyent nor Textdrive have published any information on how this endeavor is funded or supposed to sustain itself (does Joyent pay for the existing customers, as long as Joyent exists? Will it go bankrupt if it can't support the existing customers off of new customers?).

The guy running the new Textdrive seems to have all the right intentions from what one reads on the forums, and I wish him to succeed, but at this point I have no simply no trust left that this will go well (which is why wanted to take the refund).


> I have no simply no trust left that this will go well (which is why wanted to take the refund).

Hindsight is 20/20, but lifetime service claim sounds too good to be true anyway. Besides, was that 5-6 years of hosting for $500?


It seems quite reasonable when you consider that they used the money to bootstrap their operations: Take some money upfront to buy the servers, and in exchange host a bunch of people for free, covering the operating costs from other income.

It was ~6 years in total duration for me (though used on and off). I recognize that buying monthly hosting for an entirety of 6 years is not cheaper, but it's not what I bought.

I paid the money upfront, knowing fully well that I would not use it continuously, to have a hassle free way to throw up some small stuff if I need it. Others often seem to have similar use cases (or hosting small websites for local community organizations and similar).

I think everybody also recognized that there is a risk the company may go bankrupt, and we're out of hosting, but what's actually happened is that they simply wanted to stop offering the service.


Exactly; selling, say, 2000x$500 lifetime accounts with exponentially decreasing hardware costs is a significantly better deal than the equity you'd give up if you took $1m in VC as an early stage hosting startup.

The early adopters didn't just buy a product, they were essentially angel investors.

It's every bit as ridiculous as sending Ron Conway an email thanking him for his help but informing him that his equity has been "discontinued", since the investment was a long time ago and no longer convenient, but warmly offering him a refund for his portion of the round.


> but lifetime service claim sounds too good to be true anyway.

I like how you quickly turned this into the "blame the victim" issue. It is a pretty clear and obvious failure yet you keep defending them and blaming everyone who believed them.


Not honoring all the lifetime plans at all. Those with lifetime accelerator plans were told they would get a refund, are not being picked up by Textdrive, and have since been denied the refund they were promised.


If this is gloating about the EC2 outage, then ugh. I can't find out if it is because the whole of joyent.com is actually down for me (net::ERR_CONNECTION_TIMED_OUT)


Yep, same for me. Delicious irony in my opinion


If a company is actually stupid enough to say "If I was your cloud provider, I'd never let you down" then I immediately lose all interest in working with that company. Childish and they obviously have no appreciation for how hard it is to properly run a hosting company.


They don't want reddit's business. They know it's not real. But they also know reddit's general user wants reddit to be have better uptime. And that the user will upvote a story offering that (with cats at that).

This is just a PR stunt to gain some customers. Simple, but apparently working (even here on HN, which is quite sad).


I'm pretty sure they have a decent idea of what's involved running a hosting company being as how they're a huge hosting company.

I think what they've done is great, they're appealing to the users by showing they are users and they'll probably get a crapload of traffic because of it whether they land Reddit or not. It's already landed them the #1 slot on HN, a site where AWS staff post AWS updates!


In the professional tech world, not ALL exposure is good exposure. With this post they look like immature kids, and then when their post and entire site goes down, they look incompetent in front of the very people they want to convince.

An informative post explaining the hardships of uptime and reliability in simple words, and how they do their best to keep adding 9's, would have been so much better.



Exactly, they should have done more of that.


Down for me too. Rather embarrassing if I do say so.


Try disabling https-everywhere for joyent.com. Worked for me!


Apparently they're having some SSL issues, try disabling "https everywhere" or any other similar browser addons I presume you're running!


> curl http://joyent.com/

  curl: (7) couldn't connect to host
Doesn't look like an SSL issue...

  Tracing the route to joyent.com (165.225.132.33)
[...]

  8  *  *  *

  9 65.125.14.38 [AS 209] 312 msec 316 msec 308 msec

  10

  command timed-out


All I know is I get a time-out when connecting to joyent with https-everywhere running, and a normal webpage when it's disabled!


http://www.downforeveryoneorjustme.com/joylent.com

  It's not just you! http://joylent.com looks down from here.
edit: 11:12 UTC it's back up again.


It's joyent, not joylent.


> http://isitup.org/joyant.com

    joyant.com is up. [...] ip of 208.91.197.128
Although the spelling was wrong, it clearly was having a few issues.


It is not joyant.com, it is joyent.com.


Well that was stupid of me :/


Joylent? No L, Joyent.com


Uh, no, this is how about switching to the main supporters of node.js would immediately remove a problem that the #1 vps provider can't handle.


My new startup, Uptano (https://uptano.com), is a cloud provider (still hurts me to write "cloud").

..and we've had 100% uptime in our short run so far. Our systems are highly redundant and as decoupled as possible. Still, I wouldn't for one second claim that I, or anyone else, could do better than AWS on reliability.

AWS does a great job on reliability. I do think EBS is a terrible technology that should die in a fiery pit, but they do a damn good job of keeping everything up.

If they really do believe their 99.9999% claim, I'd love to see them make a bet. How about Reddit gets $1M if they move and it falls below that?


Could you please explain the tenancy/virtual servers mechanic of your service? The copy on the website isn't clear. (Also, OH GOD that font. OH. GOD.)

It's not clear to me whether I'm getting my own server (effectively dedicated?) for $136/month or not.

Further, what if I don't want multiple-tenants on the hardware? What does it cost me to effectively treat each physical server a single logical unit? Is that even possible?

I mostly want a more elastic alternative to dedicated servers. Is your service along those lines?


You're paying for a dedicated machine. Then you run virtual servers on your machine. You could run just one (treating it much like a regular dedicated server), or you could run multiple, that's up to you.

There is no scenario in which multiple users share the same hardware. Your hardware is your own.

The web site will improve ;-)

Feel free to email me with any questions: jake@uptano.com


Alternatively, instead of an oversimplified bet, how about they draw up an SLA like normal companies would?

Congrats on the 100%, but how long have you been running? And have you been serving any heavyweight clients or fended off persistent attacks?


Since this is a marketing piece, they should have to put some real money where their mouth is. Typical SLAs don't come close to being punitive for downtime. They might offer to refund the hosting fee for time down in excess of the benchmark, which would not be interesting marketing and does not indicate that they have especially high confidence in their service exceeding its uptime (relative to competitors).


It would be great if they offered reddit a 1 year deal with guaranteed %99.9999 uptime or their money back. But then, I suppose hackers of the world just wouldn't stand down to that challenge.


>Still, I wouldn't for one second claim that I, or anyone else, could do better than AWS on reliability. //

That may be true over all customers but for reddit I'm estimating about 1 day of downtime for me in a year (that might even be conservative?) - so that's 99.7% approx.

Joyent are claiming that their customers [who?] have only about 30s of downtime per annum.

Reliability obviously extends beyond downtime but still AWS don't seem to be do incredibly well for reddit.


They must be trying to sweep http://news.ycombinator.com/item?id=2269329 under the rug. It would take six millennia with 100% uptime for that outage of days to average down to to 30 s/y.

Edit: And allegedly three hours without SSL today, which would put them over 30 s/y for a few more centuries.


With a one-line pitch like "HIGH-PERFORMANCE CLOUD INFRASTRUCTURE FOR REAL-TIME WEB AND MOBILE APPLICATIONS", what could possibly go wrong?

No, I don't hate Joyent. I can't even figure out what they offer. And I don't see "Python" or "Postgres" or "Redis" in their tech stack page, so what would reddit (a Python / Postres / Redis site) want with them? They seem to be mostly node.js / Mongo, which is a nice tech stack but isn't the one reddit is built on.

Clicking around their dev docs seems to suggest you can install your own stuff on their "SmartMachines" (are those Linux VMs?), but there's a heavy node.js bias to the docs.


Joyent is some former Sun guys who are still committed to "the OS is a differentiator". They use Illumos, their fork of OpenSolaris...

I'm pretty sure python, postgres, and redis all run on Solaris, but Joyent's value add is NOT at the application layer, but in OS features that facilitate application workloads better. Their SmartMachines (or zones), ZFS, dtrace.. Solaris is an epic environment to develop in.

I probably only know of a few other OS's that are as powerful (IOS XR, OnTap, Linux..), and only corner cases are more enjoyable to work with (BeOS, plan9, lispmachine).

Their difficulty in messaging is probably around not wanting to be labelled "post-Sun", but cloud! internet! not oracle! You should think of them as providing an OS as a service.


OK, so they use "zones" (beefed up multi-user, which should have less overhead than virtualization) and a fork of OpenSolaris (interesting features, reputed to be very stable).

I wish they'd just say that, as it makes their "Faster than VM" claim seem a lot less wild.

Shared hosting is actually more efficient than VMs, but it got a bad rap due to massively oversold hosts.


Sun engineered some amazing stuff but eventually got bought out by a company with weaker engineers but much, much stronger sales team. After being burned in this way I could see how a bunch of former Sun staffers might use their former technology but call someone else in and say 'look, we can't market for shit - you have a go'.

And that's my guess on why their marketing spiel might seem a little vague to a tech person.


Personally I like Joyent - they've been very helpful to me, they support the community, they're a pretty cool company if I may say so.

And I don't see anything wrong with what Joyent did regarding their Reddit-commentary. It just shows that they are playful engineers with a sense of humor, that's all.

PS I think SmartOS is great.


I thought they would be a replacement for EC2, and not a Heruko-competitor. They provide virtual servers, and not application stacks.

I test drove them with regards to Picc.it, during the beginning of the year and they have a solid stack and great support. They are a little pricey though, so I decided to go with bare metal instead.


Solaris not Linux.


Minor nitpick but: s/Redis/Cassandra/


Yeah their Datasheet says it supports Python.


Wow, that web-site is down, this couldn't possibly be more ironic.


and still down 3 hours later... ouch!


Disable https everywhere.


Connect with Gopher, their Gopher server is still open. I found it over WAIS.


Why? The desired information has already been conveyed from the https servers: Joyent can't withstand load.


Well, or someone forgot to turn on HTTPS. How does the plugin know whether to redirect a given site anyway?


After TextDrive's lifetime hosting encounters, I'd never trust Joyent again, cloud or non-cloud.

http://news.ycombinator.com/item?id=4391669


I do not want to be a jerk here, but the only reason I joined was the promise of "lifetime" - they can NOT just cancel our accounts does it does not suit them. Please contact me webmaster @ opendomain ORG if you would like to join the class action lawsuit. We already have quite a few people signed up.


It should be “if I were your cloud provider.” It’s a possibility and a wish of theirs, so it should be expressed in the subjunctive. http://grammar.quickanddirtytips.com/subjunctive-verbs-was-i...


They were referring to a Justin Bieber lyric, "If I was your boyfriend", which was parodied by the girl who became the Overly Attached Girlfriend meme, which is popular on Reddit.


Anyone remember Joyent's Strongspace service? http://www.datacenterknowledge.com/archives/2008/01/21/joyen...


They got a free thumper from Sun through the "try and buy" program, loaded it up with customers, and ... failed to operate it properly. IIRC, it wasn't just a service outage - some data was irrevocably lost.

ZFS on Solaris was a non-beta, prime time product at that time, so it's not fair to blame it on "a ZFS thing". Deploying customers on loaner hardware is scrappy and admirable, in a way, but Sun wasn't giving those away two by two - there was no redundancy.

The joyent blog, in those days, alternated between enterprise cloud bullshit-bingo posts and facebook game development. I thought it was a clown operation back then and I suspect it still is.


ZFS was officially a non-beta, prime time product. In practice, apparently anyone who tested its robustness thoroughly before deploying it found that it tended to crash and burn at the first hint of trouble exactly like it did for them.


That's pretty much what happened if you were foolish enough to try and run a production system on ZFS back then, I think.


"Unable to connect"

well that was rather ironic..


One of those things that is hilarious but not actually funny.


Ha, ha, only serious.


I believe they are having issues with SSL, are you using HTTPS Everywhere by any chance?


I'm using it yes, disabled and it worked, thanks.


> Lindsay Shaw is a member of the Joyent Marketing and Communications Team.

This says it all really, just a bit of PR to boost awareness of their product...and the HN community has taken the bait.


While I agree with you that no press is bad press, in the Internet realm you also have to be aware of keywords. HN comments right now are littered with none-to-flattering words for Joyent, which then get indexed by Bing/Google and can hurt the reputation for Joyent.

Worse yet, Joyent's PR may backfire and make them look like naive idiots for making claims they can't back up...especially when it comes to throwing rocks at the Amazon juggernaut.


Ironically the site's SSL is not configured correctly... Doesn't work when using the "HTTPS Everywhere" extension unless you manually change it to not opt for HTTPS on the site.


Huh? There's no SSL on joyent.com. Their login form, however, has it, on a different domain: https://my.joyentcloud.com.

Not sure how this can be classified as "not configured correctly". It's HTTPS Everywhere who's incorrect there.


I 've been using their vps for years, and they didn't let down (apart from less than a handful of reboots, the uptime was basically around 400 days). The thing is they kept their prices unchanged for 4 years which is odd.


If I were a grammar nazi, I'd be quite irate.


Don't hate the provider, hate the meme:

http://knowyourmeme.com/memes/overly-attached-girlfriend


They're actually referencing the Ryan Gosling meme, not OAG. There's an image of him right on the page.


WordReference usually has good discussions on such topics. E.g. http://forum.wordreference.com/showthread.php?t=50929

Having studied Spanish, "if I was" sounds very strange now. In Spanish it'd always be the subjunctive: e.g. "si fuera..." and never the indicative "si fui...".


Same with Italian, although occasionally people do use the past rather than the subjunctive, but it sounds pretty bad; much worse than "If I was".


both are correct, "were" is more precise and formal


No, "was" implies the past indicative, which means "If I was your cloud provider in the past", whereas "were" implies the subjunctive mood, which expresses "various states of unreality such as wish, emotion, possibility, judgment, opinion, necessity, or action that has not yet occurred" (http://en.wikipedia.org/wiki/Subjunctive_mood)


but the subjunctive is used inconsistently in english, and less so in england than in the usa. it's really not as clear as you assert.


I thought "If I had been" is past indicative, and "If I was" is ungrammatical.

Insofar as hodgepodgy English has real rules anyway...


"Had been" is the http://en.wikipedia.org/wiki/Pluperfect, further into the past than "was".


Uptime comes from how you design and build a system, not which IaaS provider you choose.


So if your IaaS goes down, that's your fault?


Yes. You are responsible for replication, failover, disaster recovery, etc.


The point of using IaaS is to outsource as least some of the responsibilities.


No, you outsource the work, not the responsibility.


Then what is this 'infrastructure' that they're providing you with?


The option to build a full stack with LBs, Databases, Webservers and everything else, without spending millions in hardware that is perhaps running on 50% of it's actual capacity.


See whoownsmyavailability.com for details.


Sure if you're an enterprise I would see that.


To some degree, yes.


Umm The blog post is down. If you can't handle traffic to a blogpost, I don't see how reddit would take your offer about "no downtime" seriously.


It's not a load issue. Disable HTTPS Everywhere.


> Disable HTTPS Everywhere.

It wasn't that either. It didn't work for me, and for many other people. See the thread for simple `curl http://whatever-joyent` failing.


If something "sounds" suspicious, it usually is. I had one of the early "truly unlimited" mobile data plans w/VZ. They actually honored it until I made the blunder of changing my plan. I was grandfathered in, but they kept sending me teasers to get me to switch. Once I did, going back was not an option. They got what they wanted . . .


I read a post quite a while ago from the reddit admins bitching out EC2 after one of the last dramas, I remember they said it would be a huge migration to migrate all data across to a new provider so they would have to be solid.

I wonder how their current EC2 pricing would compare to Joyent?


During the second-to-last Amazon EBS outage (that affected our RDS instances, I started migrating our database over to Joyent. Before RDS was completely restored, I had a MySQL instance up and running and our production servers pointing to the new host via SSH tunneling. As far as pricing goes, they're competitive with EC2, but, you need to factor in the amount of time that it'll take to move off of Linux and onto Solaris/SmartOS (which isn't that big of a deal in the grand scheme of things)...


Joyent, you so cray. Thanks for reminding us that "Never" and "Lifetime", "100% Uptime Guarantee" and "Fanatical Service" are all bullshit indicators. You've got a good cloud, you don't need to make promises that you can't keep.


This from the company that patented the 8-day outage.

http://www.datacenterknowledge.com/archives/2008/01/21/joyen...


Saying you have a more robust cloud than Amazon is laughable. It's really as simple as that. This is a joke.


Sorry, and I really don't mean to be a grammar nazi, but it's "if I were". It's a subjunctive.




"Error 102 (net::ERR_CONNECTION_REFUSED): The server refused the connection."

The page is down.


They do not serve https for the site, what else are you expecting to see?


Notice to those of us running the EFF HTTPS EVERYWHERE... they'll force SSL and fail.

Not running SSL should be considered a bit of a Faux pas.


Re: "We’ve given our other partners 99.9999% uptime."

That's neither technologically possible nor commercially feasible.

Companies need to stop saying that.


were, not was. learn english modal verbs and subjunctive before soliciting business.


Or switch to us-west-2 to immediately solve more than half their problems?


If I were a serious cloud provider, I would check my grammar.


They should send this open letter to Heroku as well :)


Historically, running Ruby on Joyent was an exercise in frustration. Heroku isn't only Ruby these days, but operating as a stable ruby platform is still a requirement I think.

Among other things: Solaris threading libs with crashy bugs when you pop a new thread open. Took quite a while to figure that one out.


It is perhaps a difference in thread creation semantics, or something like that. But, Solaris threading has been rock solid for many, many years.


Not in my direct confirmed experience. pthread lib on Joyent's particularly weird version of solaris was f'd hard.


It's annoying that the two people that try to tell me that "Rails doesn't scale" are also Joyent customers. I tell them, "no, you can't scale Rails".


Am I the only one finding the post a bit sexist as well? Since when are successful 9gag memes like "Overly attached girlfriend" pc?


Those are reddit memes. Not 9gag memes. 9gag is a content stealing site. See oatmeal, etc.


because of all the "success" with twitter back in the days


What, no ent jokes?


There is no way they can achieve that uptime claim, nor should they try to say they have 99.9999% aka "six nines". That's 31.5 seconds a year, and even a well-designed network is going to have that much at some level. I mean, the VPSes might be distributed across hardware, but a failure of some component might mean that it's still "up", but seriously degraded during the transition. How useful is that for "uptime"? It becomes a bragging right. How about if all routes to a large provider, like Level(3) are down for a reset period, or are re-routed through Cogent, which is maxed out on all peering with L3 at 99% packet loss levels? Is that really uptime? Sure, that's out of their control, but the situation with their own website being down for some people right now underscores that lofty claims are meant to be broken.

Now that they're beating that Joyent drum a little more: watch them fall down under a DoS attack that Amazon could bat away at this point. I'm by no means supporting Amazon, but Joyent is still a small company with big claims. I predict they will fall apart when Sun stops making hardware, and when their engineers argue over whatever trendy "wrong way to do it" Node-type technologies that they're jacking off over next.


When you are talking about "X nines, in retrospect, from a small sample" it's meaningless, and they know it. All it means is, they've had 100% uptime, because they've been lucky enough to avoid a crash. And by 100% uptime, I'm sure that's "100% of crashes weren't in our opinion our fault".


Why is this the top post?




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: