Hacker News new | comments | ask | show | jobs | submit login
Rackspace Buys Server Management Platform Cloudkick (YC W09) (techcrunch.com)
202 points by wouterinho on Dec 16, 2010 | hide | past | web | favorite | 59 comments

As an on/off CloudKick user (long story) my concern is that RackSpace has purchased the company to obtain performance and metric data for its competitor's serving solutions. If I'm right, expect to see them reduce the price and/or increase the allowances of the free tier in order to obtain more data.

This is always a problem when a company is purchased by one of the vendors it sits above in the value chain, esp when impartiality and independence is important in a space such as vendor monitoring.

I can't see RS buying CloudKick for internal monitoring of its own customer's servers and instances because they would already have a mature solution by now.

I'm not sure I want to be providing RackSpace with performance data of my servers running on their competitor's instances, the CloudKick agent is placed to send (albeit transparently) all sorts of data back to RS. And don't be surprised if other vendors raise concerns to this effect too.

Hi, I'm Paul and I worked at Cloudkick / am now a Racker.

I just want to respond to one point.

  > I can't see RS buying CloudKick for internal 
  > monitoring of its own customer's servers and 
  > instances because they would already have a 
  > mature solution by now.
The intention is to use Cloudkick to monitor things like this. Rackspace has many existing internal solutions, but they have many issues that Cloudkick helps knock out, even for monitoring their internal infrastructure.

Well if RackSpace, an industry veteran with 12 years of serving experience and $600m+ annual revenues, feels that a two-year old venture funded startup offers superior monitoring functionality than that says a lot about the company's internal processes and development priorities.

(that's not a disrespect to CloudKick, but an examination of scales of economy)

But whatever. If this is about bolstering RS's own monitoring does that mean CloudKick will no longer be monitoring 3rd party vendors?

I'm guessing CloudKick will still continue to be more than happy to monitor Amazon EC2, Linode, GoGrid, etc out of the box.

And so even if this acquisition is going to help RackSpace's own monitoring, the point still stands that RS now has a massive competitive intelligence operation going on here too. If the data is available inside the company then it even has a duty to shareholders, as a publicly listed company, to make the most out of that data to enhance profits.

Strong Disagree. Great products often have to come out of small teams and companies. Often times it's even former employees who create those comapnies and are re-acquired!

This is why Cisco acquires companies started by former Cisco employees on a regular basis that some folks have been acquired into Cisco 3 or 4 times in their careers!

Also, the competitive intelligence is not as useful as it may seem. Data for the sake of data isn't valuable. And when you figure out what they could turn it into, it doesn't seem like it would show much more than economic and market forces already show, even if CK got to scale.

Yeah that wasn't the point I was making - of course innovation rarely comes from big companies.

My observation was just if you are a $600m/year web serving infrastructure company than a core competency should be monitoring. And thus if a two year old scrappy startup can build out better monitoring, what on earth have RackSpace been building and relying on for the past 12 years?

On the competitive intelligence I think it's insanely useful for RackSpace. Down to shit like being able to monitor up-stream connectivity and routing of other providers and then being able to cross-sell their own proposition knowing the weak points of the current 3rd party solution.

"Well if RackSpace, an industry veteran with 12 years of serving experience and $600m+ annual revenues, feels that a two-year old venture funded startup offers superior monitoring functionality than that says a lot about the company's internal processes and development priorities."

For RackSpace, monitoring is a cost center and not a core competency.

"If the data is available inside the company then it even has a duty to shareholders, as a publicly listed company, to make the most out of that data to enhance profits."

This cliche, like most cliches, is not only annoying, it is wrong:


What does economies of scale have to do with creating great software? Are you surprised IBM didn't create the Apple II as well?

Paul -- I use CK to monitor my Linode apps. Should I expect to start seeing "switch to rackspace" pitches when I login? Are you going to do anything non-anonymously with my data?

No, the Cloudkick as a monitoring/management service of all cloud providers will always remain.

wow. that is an excellent point, I never even considered that. It kind of reminds me about an article that I read about fandango - apparently the make the bulk of their money from selling their audience demographic info to the movie studios, and not the $1 fee they tack on to the ticket.

My main complaint with cloudkick has always been the wide price gap between the free account and the first paid version. I would have paid for their agent based service for a few of my projects but at $99 a month for the cheapest version, it would have cost over 5x what the server it was monitoring cost.

I think they would have grown much faster with a smoother price curve. Make a plan with 2 or 3 servers of full monitoring for $30 a month to grab the low end of the market when people are just starting and putting in the foundational pieces of their technology stack.

Overall though, they're really good. They seem to have done plenty well anyway. Congratulations.

That's why I am writing my own: http://crondom.com/

I wanted very basic things (namely application event monitoring for fraud), so I built my own analytics package around cron jobs.

I saw you promoting this before and to be frank I looked at it and didn't understand what it would be good for. I thought, "Oh, ok so it's like a cron task that's easier to manage but you pay for" and then filed it away deep in my brain under "Maybe someday I'll need that"

Maybe I'm mistaken but I don't see how it is even remotely similar to what these monitoring services provide. Can you explain that better?

I laughed at the name, but how will someone justify that line item in their engineering budget at a larger company? Catchy though, I'll definitely remember it.

I saw this and I didn't understand what it is for or how it works.

If it runs on your server then how does it monitor my application? Is there an agent or something? Why is it better than running cron on (say) Google App Engine?

If it runs on my server, then exactly why would I want it? I think I can live with running every 5 minutes or 20 seconds instead of 7 minutes or 17 seconds.

So.. I just don't get it? Where can I read more (without registering)?

If you have a developer plan and use "refer a friend" option, you can earn up to three extra "premium" nodes for free by inviting your friends - https://www.cloudkick.com/blog/2010/nov/24/invite_a_friend/.

Agreed. Hopefully Rackspace will re-price accordingly.

> Make a plan with 2 or 3 servers of full monitoring for $30 a month

In that price range you have Pingdom and AlertFox...

Couldn't have happened to a better bunch of guys. Really great news, congrats gents.

YC is on a roll. Congratulations to all involved.

Now back to work as i wait patiently from my turn. :)

I wonder what this will mean for Cloudkick's support for other clouds...

From the post on the Cloudkick blog, "We are committed to continuing support for multiple clouds within the Cloudkick tools. This means no matter which cloud infrastructure vendor you choose or what data center you have physical servers in, you can continue using Cloudkick."

For some background Jungle Disk still supports multiple clouds as well.

This is nothing against cloudkick. I've never even been to their site. I'm sure their service is amazing. My question is a general question that I ask almost every time I see a company get bought.

The question is, "Why buy, when you can build?"

Let's say I'm Rackspace. I have a cloudkick account that I used when I was evaluating them to see if I wanted to buy them. I know exactly what they offer.

So, let's say I take that and I create an engineering specification. Build a service that does this, this, this, and this, duplicating cloudkick, right down to the screen layout of the dashboards, if you're so inclined.

Now, go hire 10 engineers. No, make it 20. Now hire 5 project managers. Pay each engineer $100K + benefits, say they cost me $150K per year. Pay each project manager $150K + benefits, say they cost me $200K per year. So now I have 20 engineers and 5 project managers whose combined salaries + benefits cost me $4 million a year. Let them work for two years. For $8MM, I have my own cloudkick, right?

Why would I pay more than $8MM? Perhaps they didn't, I don't know. Again, my questions isn't about cloudkick. It's about why you see companies make the buy decision, when it often seems like it would be well within their ability to build it themselves, if they are so willing to spend money.

(1) Building is more expensive than you think it is. For instance, if you're not staffed to build something, you have to recruit. Recruiting is dreadfully slow and painful. "Go hire 10 engineers" could take a year.

(2) Building (obviously) takes more time; the time you spend building is time-to-market that you're conceding to competitors. When you buy a competitor, that gap is erased; you get the benefit of their earlier time-to-market.

(3) Building comes with enormous risks. Deciding to build a cloud monitoring platform is not the same thing as executing on a cloud monitoring platform that the market will embrace. It may in fact be more likely that you will, on some important but perhaps subtle axis, fail utterly. Now you've made a huge investment in your target market and gotten negative returns.

(4) Acquirees are proven in the market. If an acquiree sucks, you buy one of their competitors (unless they suck on an axis you can readily remediate, in which case you probably get a big discount).

(5) Acquirees come with customer traction.

I think this is a fantastic post, not just in response to this question but as a response to "Why can't company G just make something like it and take you guys out?"

Fundamentally, and quite succinctly, you've explained why building a successful business/project/technology is hard. And because it's so hard, those that have made something successful and with traction have something of value.

It's a matter of risk. Think of it like a house... Renting vs buying vs building one. All strategies have their benefits.

Renting you let someone else maintain the property and can walk away at any time. This is what CloudKick's current customers all do when they consume the cloud service.

Buying means gaining control. You build equity and have total control over all the maintenance and improvements. And you have all the knowledge that hindsight offers. Is the house in a thriving neighborhood? Was it built well? Did the previous owners take good care of it?

Building means what? Total control over the entire job, from finding architects and contractors, having say in every detail of the design, the budget, bill of materials, schedule, etc.

But the amount of risk is massive here. If funding takes a hit, or budget blows up, what do you do?

Back to software, more projects die in incubation than mature to CloudKick status, no matter how mature the software companies are. Think really hard about how fragile software development practices are.

And there's further risk in the market. If you set out to build your own CloudKick style infrastructure, CloudKick is still out there furiously innovating and battle testing their software. It can be difficult or impossible to catch up.

You are proposing to spend $8MM to build the cloudkick of today in 2 years time. That might be ok if you're just worried about internal tools, but that's no good if you want to offer the service to other providers.

> Let them work for two years. For $8MM, I have my own cloudkick, right?

Assembling all the pieces that make a product does not make a product. There are lots of risks involved in creating new products and generally as companies get larger they become more risk-averse.

Even assembling the pieces involves jumping hurdles in any significantly large company (ie. various levels of approval, staffing, inertia). Acquisitions avoid product development risks and the acquirer also get a group of people that demonstrated they can create a successful product.

You do realize that two years also costs money? The other thing is that fact that there's a company with a working solution in front of you. What if you build this team and then a year later you realize that you screwed up and you can't build it the way you thought.

The other reason you sometimes pay a premium is to get them off the table for a competitor. So even if you could build this solution in two years for 8MM. What if your big competitor decided to come buy them and then starts expanding their business. In two years you come out and say, "Hey look, we have that product from a couple of years ago!"

If you have cash sitting around, you'd want to spend it to accelerate your business. Buying a company with a successful product, great team, large user base and actual revenues can be seen as a much safer investment (assuming the price isn't inflated) than spending 1 or 2 years trying to catch up with the competition, with no guarantee of success.

It doesn't mean of course that buying a company is always the best decision.. it boils down to whether the acquired product and team ends up being a good fit.

Let them work for two years.

There you answered your question. Cloudkick's already worked on the problem for two years.

Time matters, it's all we've got to sell.

Two years is an incredible length of time in this "industry". Look how fast Amazon are rolling out features compared to Rackspace (their current closest rival). Rackspace are losing the arms race. IMV they simply don't have the cloud engineering resources to build (hence the "tom sawyer's fence" which is OpenStack), so they buy. A good strategy if you have the cash.

I can see Rackspace extracting the monitoring features/code from cloudkick and add it to their cloud servers control panel, and leave CK going as it's own entity. That's pretty much what they did with Slicehost iirc.

Why did eBay buy PayPal, when it had spent a lot of money creating it's own alternative?

Sometimes the outside party just built a better solution, and it makes sense to get a whole company, lots of good staff, good PR and new products in one simple purchase.

As long as the return on investment is as-good or better than the ROI of building it yourself, then it's a win in terms of time factor alone.

There is also the value of Cloudkick's existing customer base.

And Cloudkick's existing engineering team, who know the problem domain as well as anybody at this very moment.

There is also the 2 years of opportunity cost of building out that infrastructure, in which case the acquired company will probably be ahead technically, acquired by another competitor, and/or have acquired an even larger customer base.

I think your making assumptions about how easy it is to build a succesful product. It's easy to look at the winning horse and say "Well that horse only cost X, and trained for Y years. I'll just go buy a horse for the same amount and train him for the same time."

Consider how many different people/organizations were trying to accomplish what cloudkick has? You're cutting a check to get the best version of all their efforts.

Additionally, Rackspace has become fairly large, I'm sure there are sufficient politics in that organizaiton to force any in-house app right off the rails. The marketing manager would want the app to only measure all the places where rackspace wins. Any top shelf talent would be poached by the orgs core business units etc... Even starting with the winner in this area, they may still fall behind.

I would say it is for acquiring their customer base. In fact, customer base is probably the biggest factor in a lot of acquisitions. Like you, i have no idea about cloudkick, and this might or might not have been the reason for the acquisition, but would be surprised if it isn't.

So, let's say I take that and I create an engineering specification. Build a service that does this, this, this, and this, duplicating cloudkick, right down to the screen layout of the dashboards, if you're so inclined.

Isn't that...illegal?

Possibly - but if you move stuff around on the pages a bit then it certainly is legal.

Because it's a war. A scrappy five year old company, Softlayer got bought out for $500m, merged with ThePlanet, and now SoftLayer is the first or second largest dedicated server company by sheer number of servers, and once all of TP is integrated with SL's automation, they'll most likely have the highest margins.

SL gets margins by being efficient whereas Rackspace gets margins by having an incredibly large and efficient sales team. SL is a hacker's hosting company which has made very solid progress selling to the enterprise.

To top that off, Salesforce just bought Heroku, telling the world that they really intend to compete in cloud hosting against Amazon, and any other clouds that matter (do any other clouds matter?). The competition is fierce, the stakes are high, and we're seeing an arms race. I wouldn't be surprised if HP got into the mix as well (there must be some reason they bought loudcloud and EDS beyond just rounding out their EYP division).

Any start-up who can help some of the big dedicated server firms (Softlayer, Rackspace, Peer1) either appear to be a sexy alternative to Amazon (to try and woo new startups, which is where all of the dedicated server companies are sort of lackluster, primarily due to a lack of advertising and a lack of a presence in the valley) has a high potential for acquisition in the next year. Look at the past three years of operations and cloud related start-ups. For example I'll be really surprised if either Puppet or Opscode are still independent entities in 2012.

"To top that off, Salesforce just bought Heroku, telling the world that they really intend to compete in cloud hosting against Amazon"

I don't think so, not directly anyway. Heroku is platform-as-a-service (applications), Amazon is infrastructure.

And what do you think runs on top of that infrastructure?

We can quibble on how you use it, but in the end, the goal is to own as much of the application hosting market as they possibly can. At least, if the cloud pundits are right, and if there can be only 2 or 3 utility computing companies in the end.

Merry Christmas, YC!

Awesome news for both CloudKick and Rackspace. For what it's worth, I hadn't really tried out CloudKick before, but tried it after today's news...I'm quite impressed. I really hope Rackspace rolls this functionality into their own offerings.

Price anyone?

Based on 2.5 mil they have raised, price would be north of $25-30 mil. Great job for a 2 year old company!

Do you think it's that much with amazon offering a monitoring solution now as well.

Yes, and for free :)

Plus there is always Pingdom, AlertFox, Webmetrics, Keynote...

In fact, I think the acquisition would have been in $50 mil range simply because VCs otherwise won't allow the sale. If they raise $2.5 mil at say $5-7.5 mil post money valuation, then for making 10x VCs would only say yes to an offer of $50-75 mil.

VCs may not have had a say - which has become more common, esp after only a single round.

Rackspace don't really do inflated acquisitions - they paid $18M for Slicehost and Jungledisk (combined). Cloudkick raised $2M back in April, so you would think that the price would be more around $15-20M.

Either way, congratulations to the team, RS and to YC. CK is a great product.

I am going to guess 30m, give or take a few bucks.

Even if VCs target at a 10x sale, they definitely take offers to sell at a much lesser multiple.

They target at 10x so that with their spread of total-failures/so-sos/big-exits they end up making a nice return.

Congrats cloudkick! Rackspace is getting an awesome product and even more amazing team.

Score another for YCombinator.

Good for them. Good toolset, good platform. Really needs an iPhone app.

Let's hope the insane CloudKick pricing is looked at by RackSpace.

Rackspace is desperately in need of as much help as they can get. Whereas the rest of the industry (Softlayer, etc) spent the last five years building out automated infrastructure, Rackspace spent it mostly on sales. As a customer with dozens of servers at Rackspace, and hundreds at Softlayer, the difference is night and day.

I feel bad for Cloudkick as a company, though. San Antonio is a million miles away from San Francisco, culturally. Good job on having an exit, good luck on not hating yourselves in a year.

I've been working on mobile apps for Rackspace from San Francisco for a long time, and I have to disagree about the SF/SA cultural difference, at least in this context. While San Antonio is definitely vastly different from San Francisco, the culture within Rackspace is certainly compatible with Cloudkick and most other Bay Area tech companies.

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