Hacker News new | comments | show | ask | jobs | submit login
PHP Fog Raises $1.8 Million To Be The Heroku Of PHP (techcrunch.com)
41 points by ggordan 2386 days ago | hide | past | web | 38 comments | favorite

Don't really get this one. First off, PHP doesn't have the deploy/run process of Sinatra/Ruby/Node/etc. Second, 1.8 mil isn't much money; that's probably the amount you could bootstrap with some past success. And the business-plan-like graphic doesn't reveal that there's anything interesting in their tech approach than any commodity hosting.

By now, you deploy Ruby applications in the same way you deploy PHP applications, for the most part. We've come a long way since a few years ago. An Apache + passenger setup looks basically the same as Apache + mod_php.

Here's the secret sauce of Heroku: I know how to set up servers. I don't want to.

Paying for convenience is always a good business model.

>Here's the secret sauce of Heroku: I know how to set up servers. I don't want to. >Paying for convenience is always a good business model.


I really wish that the Pantheon folks would scale up and open up their platform to generic PHP apps. It's everything I want setup and ready to go: Varnish, Apache, MariaDB, PHP5 APC, Memcached, Solr, Hudson all on one box, configured for dev, staging and prod environments, and managed via git and Hudson for deployments and automation.


Edit: Forgot to add that the best part about the Pantheon service is that while you can use it as a total SaaS like Heroku, it's also 100% hackable, you have root on the box, and can change anything that you want to, quite an amazing idea (have to see how their support issues pan out).

Just wondering, how much control do you have over configuration settings for Heroku?

I imagine you could save a lot of money by optimizing apache/mysql rather than using the default configuration Heroku sets.

Basically none, and I like it that way. You know why?

> I imagine you could save a lot of money by optimizing apache/mysql rather than using the default configuration Heroku sets.

I really doubt this, that's why. Their entire life is making reliable systems for applications. Mine isn't. They know waaaaay more about this than I do.

You don't get a ton of control. And yeah, you could save some money by setting up your own servers and fine-tuning them just the way you want to.

But I don't want to. Like Steve above, I can set up a server, but I really don't want to deal with that stuff. Heroku's value-add to me is that my time is better spent on being a software engineer than being an ops engineer. That value-add is worth the extra money.

i like the idea but i dislike the execution of it (so far, having tried it).

it isn't as complex to deal with a php app, but it can be stressful to play server admin on an app that gets to be popular. simply not having to worry about setting up a server, optimizing it, scaling it, backing it up, etc. is just nice.

I am also confused. What are the benefits of phpfog if it normally only takes 15 min to create a linux server with php/mysql on amazon ec2? The only things I could see different are db slave replication and http caching, but generally those you don't worry till much later, and are not critical.

If you haven't been paying much attention to EC2 until now, getting your head around the pricing structure, what plan makes sense, and the whole "You have a VM but no permanent disk" thing can take some time. PHPFog's value proposition would seem to be "We've figured out EC2 for you, and have default VMs setup for Drupal and Wordpress and a way to deal with those system's need for permanent disk space. If you can get your code in a git repository we'll handle the rest." Just getting Drupal and Wordpress out of a shared hosting environment is a huge win security/stability wise, and if their VMs have a configuration that deals with a lot of the well known attack vectors that aren't shared host related all the better.

I'm not sure that I'd use it myself, because I'd like to spend to time to get familiar with EC2, but with shared hosting becoming a less reliable solution as internet traffic patterns change (and shared hosts increasingly go he commodity route) there's going to be a lot of PHP professionals (and PHP hacks, and non-programmer hacks using PHP systems) looking for a turn-key cloud solution, so I can see the interest in PHPFog.

TL;DR: There's large swaths of the PHP community that doesn't know how to make EC2 work, PHPFog is aiming to be their middleman.

> the whole "You have a VM but no permanent disk" thing can take some time.

Not to sidetrack the discussion, but EC2 has had persistent disk storage since the fall of 2008 ( http://aws.typepad.com/aws/2008/08/amazon-elastic.html ).

Hey, not a sidetrack at all, and I appreciate the link.

At the risk of putting this back on discussion though, from the outside EBS looks like a separate, but complimentary product for EC2, which will further muddy the waters for someone not familiar with Amazon's platform, which leaves room for a someone like PHPFog to swoop in as a middleman to provide a turnkey solution to the problem of getting PHP developers and Drupal/Wordpress site owners up and running in a scalable cloud environment. It's not that this is hard for someone to solves on their own, it's that for a lot of developers in the PHP Ecosystem there's too much inertia to get around to it.

getting your head around the pricing structure, what plan makes sense, and the whole "You have a VM but no permanent disk" thing can take some time.

Yeah, an hour or so? Seriously, I still don't get it. There is also this: https://hub.turnkeylinux.org/ from http://www.turnkeylinux.org/ guys. Haven't tried it myself yet, but I really like what they did with turnkeylinux images, they're absolutely amazing for what they are (time savers).

When you "don't get it" do you mean for yourself, or for the market? If the former, you can stop reading, because for someone already up on EC2 and devops it doesn't make sense to insert a middle man. If it's the later, here's more to consider.

I think you misunderstanding may step from underestimating the time needed to make the right choice on an EC2 Cloud vs. Slicehost VM vs. Rackspace vs. Shared hosting vs. That guy with the space space in his rack for a PHP application.

For someone not up on EC2 they need to research what the difference between a Standard On-Demand Instance, Micro On-Demand Instance, High-Memory On-Demand Instance, High-CPU On-Demand Instance, Cluster Compute Instance, and Cluster GPU Instance is, and then decide which type of VM they need within that sub-group for their app.

Then they need to figure out where their application touches/stores items on the file system, and decide how they're going to restore that whenever they spawn/restart another EC2 instance. (file uploads, avatar image uploads, etc.) Lots of PHP applications make an assumption about about a readable/writable persistant file system (which, coincidentally, is why shared hosting is such a successful attack vector). The is where PHPFog having the Drupal and Wordpress installs come in handy, the implication there being they've solved that for you.

To get familiar with the EC2 infrastructure to make these choices, they need to make a $63/month investment or pony up the $227.50. Minor, but expensive enough that it creates inertia. If PHPFog gets their pricing right, or they have a compelling marketing pitch that this really is turn-key, they can capture a lot of mindshare right now, and possibly be a disruptive player in a segment of the shared hosting market that's been static for a long time.

None of the above is hard for a technically competent individual, it just takes time. It's plumbing, and (this is the important part), in the world of a PHP developer it's the sort of plumbing that up-till-now can be handled by a reliable shared host, or for larger companies/teams/apps, server operations folks who don't do application programming. That's why there's a lack of EC2 knowledge even among competent profesional PHP folks.

Add in all the people just using Drupal/Wordpress/Joomla/Magento/etc who aren't technically competent and you have a another market segment PHPFog could serve.

Are we looking at a $212 acquisition for PHPFog? Probably not. But that doesn't mean they can't be a successful company, or that they can't use a cash infusion to get the tedious-yet-requiring-a-brain work for getting default installations for all the popular PHP applications running.

From http://www.phpfog.com/pricing: "We will provide simple one-click installations of some of the most popular PHP applications out there."

I'd say that taking a 'Heroku clone' approach might work against them.

Their biggest feature may be the ability to easily install and deploy pre-existing PHP products like Joomla, Drupal, vBulletin, etc.

People that use these commercial PHP projects are likely to appreciate such a service for many reasons. In a way, they are capturing a lot of the value that cPanel offers.

I think most of the people that could do that for the LAMP stack could probably get through setting up a Ruby or node server without too much extra time. Also, once you're done it, how hard is it do repeat? Yet I don't see folks making these same arguments against Heroku. One of the key points I got from talking to Lucas is that he's focusing not just on scalability, but HA, which is something that while not hard to do for the LAMP stack, isn't trivial either (usually it's just a matter of time spent setting it up).

The other neat (or annoying step) is that PHPFog goes the git route for handling your code. Just signing up and creating a site puts a default codebase for the platform you've selected into your git repo, and you just check it out, and commit to get started.

With Phusion Passenger, a Ruby environment is just as easy to setup these days as PHP. Back when Heroku first started, I don't think Phusion Passenger existed.

However, the point of these services isn't necessarily to save the developer from having to setup these environments, but rather it's to save the developer from worrying about having to setup a solid, scalable production setup. You still should have a development setup which lives somewhere other than Heroku.

Does it really take that much time to create a linux server with rails on EC2?

Energy barriers seem to matter.

phpfog right now is offering the "don't worry about until much later" benefits as their main selling point, which is the reason why i'm not currently using them, but they have promise if they can bring their concept/model down to smaller apps

Anyone else tired of hearing about "the Heroku of x?" However, you can be pretty sure you hit the right combo of good timing and implementation when your own startup is followed by services which are referred to as "[insert your startup name] of x."

ETA: Let's have some fun and play with the tagline they have on their pricing page which is "Like Heroku, but better."

Like Heroku, but without ninjas and rockstars.

Like Heroku, but better, but less cool.

Imagine explaining it from first principles:

Q: "Please explain what your application does, without comparing it to anything"

A: "It's the Heroku of PHP".. No!

A: "It's like EC2, but preconfigured for PHP with less control".. No!

A: "It's like shared web hosting, but with better performance characteristics and somewhat more control".. No!


A: "You get a server on the internet, where you can install PHP stuff. Except you don't actually have a server, and it is preconfigured to automatically scale your app across virtual machines. It comes with a set of tools and scripts to make installing stuff easy, and you don't need to administer it. By letting us concentrate on system admin and scaling it gives you more time to concentrate on building your application."

Q: "Oh, you mean like Heroku, except for PHP?"


Tired or not it's not a bad way to make value proposition clear and in hindsight is pretty handy:

- "Youtube is the flickr of Video" - http://slashdot.org/articles/05/08/14/1320217.shtml

- "AdMob is the doubleclick of mobile" http://www.businessweek.com/the_thread/techbeat/archives/200...


Sure, but wouldn't it be better if you could make that proposition from your own presentation without having to refer to other services? If PHPFog is the Heroku of PHP, then what was Heroku the Heroku of? ;)

Analogies carry a lot of meaning, often subtle and hard to get across in a paragraph, let alone a line.

Who else would you explain it? PHP cloud hosting?

Peoples understanding of a thing is not always very high res. Take 'evolution by natural selection'. Some people understand the mechanics, consequences and how A leads to B quite well, some don't. People assciate all sorts of things outside of the mechanics with it: 'aha moment,' 'prototypical scientific theory,' 'gradual process,' 'controversy,' 'cornerstone theory,' etc. etc.

You can get at this fuzzy nebulae of meanings and associations via analogies: "Darwin of climate science," "Natural Selection of Geology," "Intelligent Design of Feminism."

A lot of people don't know exactly what Heroku is, they know it's getting a lot of buzz, have heard the praise, have heard customers staunchly defending the prices, etc. They might be inclined to check out a Heroku (or for example, app engine) for their language.

You do realize that "The Heroku Of PHP" is not a line from PHPFog right?

It's how TechCrunch is explaining what the site is.


Actually, it's right on their pricing page. Maybe they didn't think of the line themselves, but it's on their site and not as a quote.

PHP Fog is like Heroku for PHP. But better. We will provide simple one-click installations of some of the most popular PHP applications out there.

ETA: However, I do agree that "like Heroku for PHP" is a big difference from "is Heroku for PHP." My mistake.

It's a good way of explaining it to people as many more have heard of Heroku. It's not like their directly competing so I can't see a negative side.

It's the equivalent of defining a term. You could offer a synonym, or you could circumlocute.

I agree when a company's mission seems simple, it might be overkill. But this kind of analogy was the only way of getting people to reasonably understand what my first startup did. "We're the EC2 of manufacturing."

The future is heterogeneous Platform as a Service (for example, http://dotcloud.com/) in my opinion. I can't imagine going back to Heroku where I can only host Ruby along with a selection externally hosted add-ons.

Why is Shopify listed under the apps? I always thought Shopify was a hosted rails app.

Like the service, hate the name. Though I personally would like it better if it included plug-and-play framework too. (Heroku has Rails, for example)

i hope the funding will allow you to create a free plan, like heroku. i like the idea of whats provided. i don't like the pricing structure.

We are working on some features I think you will be excited about!

i hope so. been following progress since you guys announced it (here? i don't recall). i'd like to see it fit my needs, but it doesn't yet.

nice work!I hope to use it when I can get an account. I just hope it is as easy as they say.

Congrats Lucas!

Thank you!

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | DMCA | Apply to YC | Contact