Note: in advance, this is not a slam on the company I will mention. I actually agree with their migration concept, even if I disagree with the details.
In the early days of their life, Joyent made a lifetime offer for goods and services. These were pretty reasonable offerings, and the hook of
How long is it good for?
As long as we exist.
As in … forever.
But what does “forever” actually mean? In the case of Joyent and this product, it ceases operation on 31st-October-2012.
That is, due to changes in business and product offerings, forever ends on 31-October-2012.
Needless to say (read through the comments on the link, on Hacker News, on other places), that some people are badly miffed by this. They thought “forever” actually meant “forever” in the sense of … well … forever. Think … until the heat death of the universe type of forever.
It was a great deal. Spend $500-ish, and get something for 6 years. About $83/year for something like this now (assuming buy in around 2006).
So whats the deal with this … does “forever” mean “forever”, or does it mean something else? Reading through the FAQ at the archive, it looks like they are going to have lots of arbitration. This is going to cost them money.
That’s pretty much all their lifetime contracts.
You get 45 days to move to a new service level.
Ok. Some points, some obvious, some maybe not so.
Are people getting bent out of shape over this? Yeah.
Do they have the right to? Possibly, really in the way it was handled.
Should they expect “forever” means “forever” as above? No. No company can offer a pay once support us forever scenario. This wasn’t such a great idea. And people getting into should have realized that forever really was “a long time in the future when we could no longer maintain this liability going forward.” Its … well … silly … to expect that “in perpetuity” or “as long as we exist” will not be canceled/altered later on. Indeed, the FAQ gives Joyent several outs, on technicalities, but they are outs none-the-less.
More to the point, this 45 day window, and migration to a yearly paid service really pissed a number of folks off. And I understand them and their reasoning. Joyent cloud is the new product offering, a virtualized container based system. Its pretty cool (look up SmartOS and related) from a technical point of view. Why not offer a 6-9 month migration to the new platform (say the lowest level of service, which seems to coincide with what people had), and a yearly fee based service for those who do not wish to migrate (which should provide enough impetus to migrate)?
Basically I am saying I think it could have been managed better … offering a VM on the Joyent Cloud. There could be some disruption, but if handled well, most customers could be moved painlessly over, and Joyent would be the toast of the town for keeping its “forever” promise.
Instead, twitter et al bits are blasting them, with people claiming they are going to hire a lawyer, etc.
In a word … Ugh. Turning something that could have been a good PR win into something that right now, is … not a win.
Understand, I like the Joyent folks, we are playing with SmartOS, and looking at doing stuff on the Joyent Cloud. So you know where my biases are.
I think people are over-reacting somewhat. But I also think it was handled incorrectly. Best course of action is move the sunset date way out. Call it June 2013. Set up the lifetime VMs. Help customers migrate over. Find out who didn’t get the T shirt, and send them out (yeah, its in the comments).
Why this is relevant for the cloud. Very simple. No company, Joyent, Google, Amazon, … can really hold your account open forever, hold your data or your VM forever. It is silly to think so. Really.
And this gets to the heart of the cost-benefit analysis for clouds, and the risk equation.
I know cloud vendors really want to lock in their customers, and quite a few will for a while.
In order for the cloud model to work, you need to be able to easily migrate your VM to a new provider. You can’t be locked in.
Because if you are, “forever” and all it implies, both on benefits, and risks, become very important.
In general, if your business must be up, no matter what, and you are going to leverage a cloud, you need to find a second supplier not in any way financially/technically entangled with the first supplier, to mirror your services on.
Don’t think thats important? Lets ask Netflix et al, whom were down earlier this summer when a derecho ran through the US. One cloud provider. One outage. Game over.
Is this balancing the risk equation? No.
I used to have to use FBI data center raids to get the point across that dependency upon one hosting supplier could be catastrophic for your business.
There are many businesses impacted by the Joyent move.
Note: I accept and understand their decision to sunset their service. It is the right thing to do. Its the migration plan that needs work, and the notification duration that’s problematic.
What would you do if Amazon suddenly decided (no they won’t but bear with me), to be a pure book seller again, saying that they were sick and tired of doing the whole cloud thing. Sure, they didn’t do lifetime stuff (I don’t think). The point is, moving your data and VMs will be HARD.
Basically this move by Joyent, proper as it is (albeit in need of some customer relations TLC now) highlights the risk aspect of As A Service. You need to know what you are getting into. If someone promises you something too good to be true (“as long as we exist”, “we are always up”, “we will never go offline”, “we wont lose your data”, “we are completely secure”, …) just remember that forewarned is forearmed. Forever means, in this context, for the life of the product. Which can be ended at any point in time, as business conditions require.
At the end of the day, Joyent (and all other cloud providers) are businesses, and they need to make money. One time payment, with service “forever” is unrealistic. Insisting that it is realistic … isn’t sensible, or reasonable. Had Joyent crashed and burned 6 months after the offer, how many of these users would be saying “oh well, it was a risk, and it was worth it” vs “call the lawyer!”.
Again, we are planning on using Joyent’s SmartOS for a few things (you’ll be hearing about them soon). Good company, good tech, trying to deal with a hard situation. I’d suggest they light up VMs roughly akin to what they have sold to the lifetime folks for migration. That would rapidly diffuse that problem.
But this still leaves the risk question open.
For the HPC world, this is IMO less of an issue, as our applications tend to be more self contained, and aren’t tied to particular machines in many cases. For the rest of the world, carrying your VMs with you is much like an HPC group carrying its apps and data with it. We are already used to using remote resources. If you can’t pack up your VM and go somewhere else, this is a problem. One that needs standardization and a fix.
… and Joyent should amend their TOS such that a) they can amend it in the future, and b) they can terminate it at any time for any reason with sufficient notice. This is all about setting expectations. What does “forever” mean.
[update] I should point out that Joyent did offer their “lifetime” forever customers a free 1 year long VM. Sort of like what I suggested above. But its for one year, versus “forever”. Should be easy to “fix”.