Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Uncensored Everpix metrics, financials and slides

github.com

Post reply on HN
Print view View original

101–110 of 164 posts

Re: Uncensored Everpix metrics, financials and slides

#101

Post by reeses »

rgovind wrote: post #47
superflit wrote: post #36

Earlier quoted context omitted.

So are we ready for more H1B visas? Or should we wait for the jobs move more to other countries?

I am not sure if H1's are paid any lesser than locals...Do you have data points to prove otherwise?

If you're paying H1s less than locals you're violating immigration law. The idea behind the H1 program is to help you when you cannot find local expertise in your specialty and need to recruit globally. It's a very sensitive issue in the USA if you are bringing in less expensive foreigners to "steal" jobs from locals.

That said, it's prone to abuse and I have seen some H1s with really poor compensation due to a weak negotiating position (the difficulty of switching jobs while in-country, sponsored adjustment of status, etc.).

Re: Uncensored Everpix metrics, financials and slides

#102

Post by pkrein »

I did some spreadsheet-ing to look at the "profitability" of their userbase. If the business were sustainable, then Monthly Recurring Revenue [1] would be higher than the cost to service the customers [2]. If not, then growing the userbase just increases total monthly losses (assuming a steady freemium conversion rate).

In Everpix's case, it appears they were growing unsustainably.

Their data shows that all the way up until February 2013, they were spending $3 on AWS for every $1 of MRR. That's scary, because it means that a new customer bringing $1 MRR was actually increasing their monthly losses due to usage costs!

Things got better though. In March and April 2013, a surge in premium users seems to have shifted the balance in the right direction. But still, for May 2013 until the shutdown they were spending $1.20-1.40 on AWS to service every $1 MRR.

So growth in their userbase was actually increasing their total burn rate, which makes growth unsustainable.

You can see the spreadsheet analysis here: https://docs.google.com/spreadsheet/ccc?key=0Ap7fmpANG_0QdHV...

[1] https://github.com/everpix/Everpix-Intelligence/blob/master/... [2] https://github.com/everpix/Everpix-Intelligence/blob/master/...

Re: Uncensored Everpix metrics, financials and slides

#103

Post by rgovind »

coldtea wrote: post #97
carsongross wrote: post #9

What a hugely useful thing for you guys to do. Thank you, truly. The lesson I see from this: AWS hurts bad if you are using it heavily, but salaries... salaries kill. Automate, outsource and self-build as much and as long as you can possibly stand it, and then push on a lot longer than that.

> The lesson I see from this: AWS hurts bad if you are using it heavily, but salaries... salaries kill. Automate, outsource and self-build as much and as long as you can possibly stand it, and then push on a lot longer than that. Going off on a tangent here, but if (a) "software is eating everything" and (b) "salaries kill", wouldn't that mean a society of mass (30%+) unemployment in the future? I wonder who would ha…

My theory is that when everything is automated, so will food production also be, assuming there is enough energy in the world for that. Then everyone will need very little money and Everpix can charge much lesser amount of money...thus still keeping number of users the same.

Re: Uncensored Everpix metrics, financials and slides

#104

Post by EricMayo »

$30K/month and ~100 servers seems like a huge waste. I'm in the business of managing similar services. I handle, not photos, but video streams. My work is for one of the largest companies in the world (Fortune-5) and I deal with things like superbowl traffic spikes, etc... Everything is handled in house but for spiky loads, we look to things like Akami.

I realize managing at Fortune-5 is different than 100% oursourcing but the cost constraints, I have to imagine, are very similar. For example, I have to imagine Amazon gets similar price breaks from Cisco as my company - I know Amazon has custom built servers and all so there may be some advantages but at $30K/month, I think any hosting company would happily take 1/2 or 1/4 of that.

I'd be curious to know what kind of bandwidth loads you were pushing through AWS and how much of the AWS cloud can be made redundant per client.

Re: Uncensored Everpix metrics, financials and slides

#105

Post by latch »

runako wrote: post #56
nashequilibrium wrote: post #30

Earlier quoted context omitted.

Is the AWS cost so high because they primarily stored pics? If so what can be done to reduce this cost in terms how to store the pics?

In rough order of difficulty/startup cost: - AWS reserved instances (doesn't help with storage costs though) - dedicated servers at e.g. SoftLayer - colocation

There are _a lot_ of dedicated options. In terms of price, Softlayer's at the high end. If Softlayer gives you a price/performace improvement over AWS of 2.5x, you can get another 2-3x that by picking a cheap (but not necessarily bad!) provider.

Re: Uncensored Everpix metrics, financials and slides

#106

Post by simonw »

dreamdu5t wrote: post #73
phreeza wrote: post #58

with all the lamenting about how hard running a consumer photo service is, I wonder how imgur does it. their traffic must be insane, and they have no subscribers, so how come they can do it and these relatively upmarket services can't?

because imgur was focused on making a product people valued from day one while Everpix was focused on VCs, offices, hiring, blog writing, etc.

Everpix's product was loved by their users. They absolutely built "something people want". Unfortunately the promise they made their users was very expensive to deliver.

Re: Uncensored Everpix metrics, financials and slides

#107

Post by latch »

swisspol wrote: post #82
sehrope wrote: post #67

Earlier quoted context omitted.

If you have a fixed workload ( i.e. X servers always online ) then take a look at reserved instances. The break even point is about 6 months for the light instances and a bit more for the medium ones. I don't recommend the heavy ones as the cost structure is different ( you pay regardless of whether tt's on so ... if you later change your instance types you still have to pay hourly for the full term of the reserve ).…

At Everpix, we were already extensively using spot instances for the vast majority of our EC2 needs (which were pretty intense considering the type of computation and image science we were doing). That did save quite a bit of money when we did the switch and it was easier to do than we expected, so strongly recommended!

Did you guys look at bare metal at all? (either dedicated or colocated)? Spot gets you in the same ballpark of price on a "unit", which can be mentally deceiving since an AWS unit is pretty weak.

In our experience with CPU or IO intensive workloads, Spots are still 2-5x slower than what you can get at decent dedicated provider..

Re: Uncensored Everpix metrics, financials and slides

#108

Post by jasonwatkinspdx »

nutanc wrote: post #92
jasonwatkinspdx wrote: post #78

Earlier quoted context omitted.

I think some people only see one side of the trade-offs here. A counter example would be a start-up I worked with that started out on a smaller provider. When we launched a beta of the application, we unexpectedly got invitations for national TV coverage with less than 24 hours notice. We asked the provider for everything they could give us, but they only had a limited amount of unused capacity themselves so there wa…

That's why you start with Amazon. And then once you reach scale you slowly start building your own private cloud. Its much more efficient and costs a lot less. We have been doing the same in our startup.

Yes.

Something else I think people fail to consider is what I would call a hybrid approach.

For example, even if you mostly operate on your own infrastructure, keep a replica of your data tier going on EC2, along with freshly configured images for the rest of your moving pieces. This way you have an instant pop up B site for disaster scenarios.

There are a lot of other possible concepts. Keeping a dataset at amazon to use for map reduce jobs is convenient. Or even if you want to keep your infrastructure virtual, being able to price arbitrage across a couple providers could be a powerful cost reduction. The price of entry for all this is devops automation, but I think people are now mostly convinced that's worth it.

Re: Uncensored Everpix metrics, financials and slides

#109

Post by JoachimSchipper »

EricMayo wrote: post #104

$30K/month and ~100 servers seems like a huge waste. I'm in the business of managing similar services. I handle, not photos, but video streams. My work is for one of the largest companies in the world (Fortune-5) and I deal with things like superbowl traffic spikes, etc... Everything is handled in house but for spiky loads, we look to things like Akami. I realize managing at Fortune-5 is different than 100% oursourci…

Note that they were apparently doing very fancy image processing, not just pushing bytes; that does make things more difficult/expensive.

Re: Uncensored Everpix metrics, financials and slides

#110

Post by fauigerzigerk »

swisspol wrote: post #80
jusben1369 wrote: post #69

So one of the things that's interesting here is the Amazon component. And the anonymity of Amazon vs a smaller provider. Let's assume for a minute the founders actually wanted to keep it alive (I'm guessing they didn't but that's a whole other story) You're generating revenue of say $40,000 and your Amazon costs are $30K. If you went to a small or mid tier sized provider and said "Good news/bad news. We're generating…

[Everpix founder here] Interesting idea and no we didn't try negotiating with Amazon. Not sure how much room there is here? However it's important to keep in mind in a situation like this, and at this scale of burn rate, you're not looking for a solution to shave off a few thousands $ left and right, which is kicking the can down the road (if even), but for a long term solution. The key problem here is that I don't t…

Bootstrapping it with the existing team would cost about 1.5 million a year by my extremely rough estimate. So you'd have to make $30 per user per year. Facebook's ARPU is about $6 per year (much higher in the US though).

If you make as much in ad revenue as Facebook, each of your 7000 paying customers would have to foot a yearly bill of $171 to keep the service going (but that number would go down as you grow).

So I think you're right. Bootstrapping it at this point is probably unrealistic. You may need like 10 times the number of users for it to be a viable business.

Thanks for sharing your data!

Post reply on HN