Rackspace Unlikely to Withstand Amazon, Google Onslaught(blogs.barrons.com)
blogs.barrons.com
Rackspace Unlikely to Withstand Amazon, Google Onslaught
http://blogs.barrons.com/techtraderdaily/2013/12/18/rackspace-ubs-starts-at-hold-unlikely-to-withstand-amazon-google-onslaught/
18 comments
While this may end up proving true, Rackspace has been counted out over and over again during its long, successful history.
Previously it was claimed that ever cheaper dedicated servers would wipe out their expensive offering. I think they're likely to retain a healthy business in 'cloud' for the same reason a nearly infinite supply of cheap dedicated server competitors didn't kill them previously - assuming they keep doing what got them this far, namely managed offerings and customer service. That's a value proposition Google, for one, will never even attempt to match.
Previously it was claimed that ever cheaper dedicated servers would wipe out their expensive offering. I think they're likely to retain a healthy business in 'cloud' for the same reason a nearly infinite supply of cheap dedicated server competitors didn't kill them previously - assuming they keep doing what got them this far, namely managed offerings and customer service. That's a value proposition Google, for one, will never even attempt to match.
Granted my experience was 5 years ago, but aside from initial setup, I would have no problem handing off daily sysadmin duties to them under a managed offering. I even gave them runbooks back then, which they were great at following in case of the often-out-of-control java app we had built for us. I still had midnight calls from them, but their services eliminated a lot of basics from my routine. Like you, I severely doubt Google and Amazon would ever offer these features and will always live in the shared-hosting tier of quality and performance for me.
"Like you, I severely doubt Google and Amazon would ever offer these features and will always live in the shared-hosting tier of quality and performance for me."
But just think what would happen if Google or Amazon decided to go customer service friendly and charge for managing and hand holding. My feeling is that they could turn on that spigot at any time if they wanted to.
But just think what would happen if Google or Amazon decided to go customer service friendly and charge for managing and hand holding. My feeling is that they could turn on that spigot at any time if they wanted to.
Amazon, perhaps. But AWS is already crazy expensive, adding silver spoon levels of managed hosting would price them right out of the market. The reason Amazon is competitive is because they can offer a different product than other folks. The value of that product comes from software development, not customer service staff.
As for google, they couldn't interface with a real customer if their life depended on it. And the idea that they could suddenly build out a massive managed hosting service on a whim is not reflective of reality. Your claim is basically saying that no company in the world has a competitive advantage because anyone else could just start doing what that company is doing on a moment's notice.
As for google, they couldn't interface with a real customer if their life depended on it. And the idea that they could suddenly build out a massive managed hosting service on a whim is not reflective of reality. Your claim is basically saying that no company in the world has a competitive advantage because anyone else could just start doing what that company is doing on a moment's notice.
That's something I don't really understand about all these offerings. They're all crazily expensive. There's small differences, but not much. Rackspace is a tiny bit cheaper, with significant disadvantages, google and aws are roughly the same (I get the impression google is more expensive, but better for cpu workloads).
Compare it to any "standard" vps offering and they're insanely expensive. Compared to the vast majority of dedicated hosters they're insanely expensive. Plus you're the only one on your cpu/memory/disk.
Anybody with a monthly AWS spend of $100 or more could have 2 dedicated servers, one in .eu, one in the US, for 20-30% less dollars, 5-10x more cpu, 32 Gig of dedicated memory or more, 100x more bandwitdth (and no surprise overuse charges, finding hosters that just "limit" you to 10 mbit is easy), everything non-shared and good support. Hetzner ES40 is EUR 40/month, 32Gig ram, 2Tb disk, unlimited bandwidth, similar offerings are available in the US (e.g. www.ovh.com.ca ).
Total spend for effectively ~ m3.2xlarge instance, one in US, one in EU : $120.
AWS spend for a comparable config : 1 month of m3.2xlarge : $650 5/10 Tb of internet facing traffic: $600/$1200
I think this is being generous to AWS, because an m3/2xlarge does not match the specifications of the dedicated server offerings.
Granted, these services are not as flexible as AWS. At the prices they offer though, is there anyone who really argues AWS is better for anything but their free tier ? As for the "extras" amazon offers : since they're all proprietary businesses, every amazon extra you use (e.g. S3) is something you're betting the business on.
And since none of the aws/google/... offerings are anywhere near what I want (just set this .jar in my homedirectory live, ditto for go binaries, or move-tag-and-push git publishing), I have to script stuff to make that happen anyway (I love docker).
Why does AWS have customers at all ?
Compare it to any "standard" vps offering and they're insanely expensive. Compared to the vast majority of dedicated hosters they're insanely expensive. Plus you're the only one on your cpu/memory/disk.
Anybody with a monthly AWS spend of $100 or more could have 2 dedicated servers, one in .eu, one in the US, for 20-30% less dollars, 5-10x more cpu, 32 Gig of dedicated memory or more, 100x more bandwitdth (and no surprise overuse charges, finding hosters that just "limit" you to 10 mbit is easy), everything non-shared and good support. Hetzner ES40 is EUR 40/month, 32Gig ram, 2Tb disk, unlimited bandwidth, similar offerings are available in the US (e.g. www.ovh.com.ca ).
Total spend for effectively ~ m3.2xlarge instance, one in US, one in EU : $120.
AWS spend for a comparable config : 1 month of m3.2xlarge : $650 5/10 Tb of internet facing traffic: $600/$1200
I think this is being generous to AWS, because an m3/2xlarge does not match the specifications of the dedicated server offerings.
Granted, these services are not as flexible as AWS. At the prices they offer though, is there anyone who really argues AWS is better for anything but their free tier ? As for the "extras" amazon offers : since they're all proprietary businesses, every amazon extra you use (e.g. S3) is something you're betting the business on.
And since none of the aws/google/... offerings are anywhere near what I want (just set this .jar in my homedirectory live, ditto for go binaries, or move-tag-and-push git publishing), I have to script stuff to make that happen anyway (I love docker).
Why does AWS have customers at all ?
Outside of US and EU, it is quite difficult to find a reputable and affordable dedicated hosting provider. We use AWS extensively in ap-southeast-1, and their new C3 Compute Optimized instances are actually very competitive in price-performance ratio. Previously, we mostly used C1 instances, which were just okay. I never like the M1 / M3 instances.
Besides, in this region with messy peering agreement, AWS network also has the lowest ping time and the fewest hops.
Besides, in this region with messy peering agreement, AWS network also has the lowest ping time and the fewest hops.
So inside of EU/US you do use dedicated server providers ?
AWS is effectively (almost) a transit-only network. That has pros and cons. On the plus side, it should have decent connectivity anywhere. On the minus side, if (almost) any link at all on a network is congested, that will almost certainly congest AWS traffic too. They're relatively unlikely to have long-running issues. Also : they're REALLY expensive.
AWS is effectively (almost) a transit-only network. That has pros and cons. On the plus side, it should have decent connectivity anywhere. On the minus side, if (almost) any link at all on a network is congested, that will almost certainly congest AWS traffic too. They're relatively unlikely to have long-running issues. Also : they're REALLY expensive.
Are your prices for AWS on demand or dedicated? I know Amazon is expensive, but they've cut prices for dedicated servers and if you're not comparing that then it's apples and oranges.
Sadly, yes, they're for reserved instances. With 1 year reservation "Light" the amazon price is $1070/month.
Here's a calculator : http://calculator.s3.amazonaws.com/calc5.html
Here's a calculator : http://calculator.s3.amazonaws.com/calc5.html
For context: Rackspace has been active since 1998. I had no idea they were such a (relatively) old company.
Agree with the value proposition.
But the way things like this work a downhill slide (and I'm just pointing this out not claiming I have any more insight than what I've read) can be accelerated and make any business model not work for a company the size of Rackspace.
In other words it takes a certain amount of business and confidence of the market to stay in business.
Lose enough to someone or something else and the model no longer works very well. Especially when word gets out and it's no longer viewed as the thing to do "oh people don't rack their own servers anymore they use the cloud". [1] Keep in mind that switching from Rackspace is not trivial but it can be done it just takes time. [2]
[1] Think of anything where you might say "people will always need". Once things get to a tipping point the acceleration and downhill slide becomes really important to investors and employees who begin to jump ship in record numbers. After all people still use blackberrys but you realize that when the iphone started to take hold you wouldn't think "things will be ok for blackberry there will always be people who will use them".
[2] (I'm a customer and I'm happy with what they've done. I've even had emails returned by the President.) Noting though that they got rid of their very low end $16+- monthly VPS's.
But the way things like this work a downhill slide (and I'm just pointing this out not claiming I have any more insight than what I've read) can be accelerated and make any business model not work for a company the size of Rackspace.
In other words it takes a certain amount of business and confidence of the market to stay in business.
Lose enough to someone or something else and the model no longer works very well. Especially when word gets out and it's no longer viewed as the thing to do "oh people don't rack their own servers anymore they use the cloud". [1] Keep in mind that switching from Rackspace is not trivial but it can be done it just takes time. [2]
[1] Think of anything where you might say "people will always need". Once things get to a tipping point the acceleration and downhill slide becomes really important to investors and employees who begin to jump ship in record numbers. After all people still use blackberrys but you realize that when the iphone started to take hold you wouldn't think "things will be ok for blackberry there will always be people who will use them".
[2] (I'm a customer and I'm happy with what they've done. I've even had emails returned by the President.) Noting though that they got rid of their very low end $16+- monthly VPS's.
As a Rackspace customer, I can confidently say that you feel like a customer at Rackspace (vs. being the product at Google).
As a previous Rackspace customer, I always felt that their primary business was not server hosting. I never quite figured out if it was making donuts, or a doggie daycare, but it sure as hell wasn't server hosting.
Used them at 2 different companies in various capacities, and they were never anything more than mildly incompetent in my experience. I called them the Blockbuster of datacenter.
Used them at 2 different companies in various capacities, and they were never anything more than mildly incompetent in my experience. I called them the Blockbuster of datacenter.
and Rackspace picks up the phone!
We should stop taking such predictions seriously because it is more focused on investors rather than the company's actually ability to create value.
I have worked with Rackspace since the dot com boom, they were one of the rare companies which actually grew when the bubble burst [citation needed]. It is unlikely that Rackspace will beat AWS or Google in short term but I dont think it is a dooms day either. Any large company will not make the mistake of relying on just one cloud service provider such as AWS (and I dont see why anyone would chose Google for a serious business). I have used rackspace for many reasons and spending around 20% of my infra cost on Rackspace (20% softlayer 40% Aws 20% Mixpanel).
I have worked with Rackspace since the dot com boom, they were one of the rare companies which actually grew when the bubble burst [citation needed]. It is unlikely that Rackspace will beat AWS or Google in short term but I dont think it is a dooms day either. Any large company will not make the mistake of relying on just one cloud service provider such as AWS (and I dont see why anyone would chose Google for a serious business). I have used rackspace for many reasons and spending around 20% of my infra cost on Rackspace (20% softlayer 40% Aws 20% Mixpanel).
Here's a reason to take the prediction / analysis seriously: The analyst makes these statements, the market joins in, and then management begins to follow along. Not a happy model, nor one that is always followed, but certainly a component of publicly traded companies.
It would be fun to have an review of how insightful a given analyst has been. It would also be good to have a proper 'full disclosure' policy.
It would be fun to have an review of how insightful a given analyst has been. It would also be good to have a proper 'full disclosure' policy.
From the quarterly report in September, public cloud revenue is up $29M a quarter from last year to $108M for Q3/2013. Dedicated hosting income is up $24M a quarter from last year to $280M for Q3/2013. I hardly see how doing a competitive doom and gloom comparison on what is essentially 27% of a company's revenue is useful.
Making a point that AWS owns 80% of the IaaS market has little to do with existing dedicated hosting provider's revenue!
Making a point that AWS owns 80% of the IaaS market has little to do with existing dedicated hosting provider's revenue!
The fact that this article is even choosing to focus on Rackspace as opposed to say linode, asmallorange, prgmr.com, etc. is already a sign that Rackspace has become big enough to be worth talking about. At least so far, apparently they're on the right track. I think they'll be around for a pretty long time even if they never completely crush the other cloud giants.
It might instead be a sign that Rackspace is a public company, and that Barrons doesn't bother much with privately owned businesses since their audience can't invest.
the problem now is that "merely profitable" companies need to be viciously attacked by companies like Google and Amazon that have flying STOCKS but post no PROFITS. Just "doing their job" and having happy customers isn't good enough if a Wall Street giant needs your business added to it's marketshare so they can avoid some taxes next year.
Ok, true, Amazon has no profits. But Google does. Quite a lot even.
Ten bucks says this is how Oracle gets their foot in the door of the cloud space...by buying Rackspace.
^ this, is very likely to happen
A comparison that this article makes 'Ah yes Amazon has the best most excellent Virtual machines but Rackspace has the best most excellent public cloud'. I make the best tomato sauce, but my competing friend makes the best tomato sauce.
I disagree because Rackspace is the most logical disaster recovery (DR) solution for companies that primarily use AWS. Even if you have DR backups in other AWS data centers, you're not protected from bugs that could affect all data centers. I've seen this happen with Windows Azure, and it could happen to any cloud provider. Therefor, very conservative enterprises will want a second cloud provider and Rackspace complements AWS nicely. Considering that the most conservative firms are the last to adopt the cloud, I think this kind of consideration will gain traction and drive business to Rackspace.
> Rackspace has nine data centers on four continents and more than 100,000 servers. For comparison, AWS also has nine data centers but on five continents and virtual machines in the millions. We believe AWS has more than 5x the IaaS capacity of the next 14 vendors combined.
"Servers" and "virtual machines" are two different things. You could easily run 10 VMs on a single server at the lower end of the VM price range. So without more precise measurements, these numbers mean nothing.
(What they say may still be true; you just can't tell from the numbers we're given here.)
"Servers" and "virtual machines" are two different things. You could easily run 10 VMs on a single server at the lower end of the VM price range. So without more precise measurements, these numbers mean nothing.
(What they say may still be true; you just can't tell from the numbers we're given here.)
Traditional economies of scale don't carry over very easily to the digital age.
Google being a perfect example. They leverage many economies of scale, but they do so through efficiencies that are non-trivial to implement. The datacenter model that google pioneered more than a decade ago is only now seeing penetration in the industry as a whole, and nobody can approach google's capabilities on the same score yet.
But while that helps google's search app it doesn't necessarily give them an edge when it comes to hosting. There are so many other factors such as customer engagement which are important in the hosting business. Amazon gets this right but google is the poster child for aspergers on a corporate scale. There's little reason to imagine that rackspace will be incapable of keeping up as technology advances. There's little reason to imagine that even digitalocean won't be able to keep up either.
Personally I wouldn't think it a forgone conclusion that google's hosting (app engine and what-have-you) will ever mature into an industry leading product. The problem with google is that everything that isn't search or backbone ISP is a side show in terms of revenue, so there is an absence of a culture capable of developing, delivering, marketing, maturing, and monetizing products at google. And definitely an absence of a culture capable of acquiring and dealing with customers.
Google being a perfect example. They leverage many economies of scale, but they do so through efficiencies that are non-trivial to implement. The datacenter model that google pioneered more than a decade ago is only now seeing penetration in the industry as a whole, and nobody can approach google's capabilities on the same score yet.
But while that helps google's search app it doesn't necessarily give them an edge when it comes to hosting. There are so many other factors such as customer engagement which are important in the hosting business. Amazon gets this right but google is the poster child for aspergers on a corporate scale. There's little reason to imagine that rackspace will be incapable of keeping up as technology advances. There's little reason to imagine that even digitalocean won't be able to keep up either.
Personally I wouldn't think it a forgone conclusion that google's hosting (app engine and what-have-you) will ever mature into an industry leading product. The problem with google is that everything that isn't search or backbone ISP is a side show in terms of revenue, so there is an absence of a culture capable of developing, delivering, marketing, maturing, and monetizing products at google. And definitely an absence of a culture capable of acquiring and dealing with customers.
"so there is an absence of a culture capable of developing, delivering, marketing, maturing, and monetizing products at google. And definitely an absence of a culture capable of acquiring and dealing with customers."
This really is the core of it. I'm remembering how long it took AT&T after divestiture (mid 80's) to turn into a company that instead of a monopoly and technology had to actually do marketing. The upper ranks of the company down to the guys on the street never had to do that. It took a long long time.
This really is the core of it. I'm remembering how long it took AT&T after divestiture (mid 80's) to turn into a company that instead of a monopoly and technology had to actually do marketing. The upper ranks of the company down to the guys on the street never had to do that. It took a long long time.
We're in a really weird state with technology right now and a lot of it has to do with the comparatively undeveloped level of models and terminology we have available.
Let me use an analogy. There are lots of different people who work in a "shop", but they use different tools and work on different things. There are wood workers who do carpentry, there are metal workers who work with metal, and so on. Even wood work can range from fine carpentry making use of a lathe to general contracting and building simple structures by nailing together pre-made pieces of lumber. While metal work can run the gamut from machining to iron mongery and forging to soldering and brazing and can include everything from a plumber on the one hand to someone who designs internal combustion engines on the other.
Fortunately we have the terminology to describe these different jobs reasonably well. It's not easy to confuse a plumber with a machinist. But that's not true when it comes to tech. We have very little terminology in that realm. It's non-trivial to explain the difference between a systems programmer, a DevOps automation guru, a mobile app dev, a web dev, etc. A lot of these jobs end up falling under the standard "software developer" heading. Even though that's about as silly as calling taxi drivers and truck drivers "automobile operators".
This problem of terminology extends into the tech industry as well. We have sufficiently developed mental models of things like restaurants to be able to understand the differences, and difficulties of changing, between sit-down, take out, delivery, and catering focused establishments. Yet when it comes to tech our brains go on vacation and suddenly the equivalent of a high-class sit-down restaurant switching to making pre-made meals to be sold in grocery stores seems trivial when it is anything but.
You get the same effect when people ruminate on how Apple should spend its huge cash reserves. With people throwing out crazy ideas of how Apple should buy into some completely different industry that would totally change the nature of the company, which is just as ridiculous as the idea they should switch to building rocket ships or submarines.
Let me use an analogy. There are lots of different people who work in a "shop", but they use different tools and work on different things. There are wood workers who do carpentry, there are metal workers who work with metal, and so on. Even wood work can range from fine carpentry making use of a lathe to general contracting and building simple structures by nailing together pre-made pieces of lumber. While metal work can run the gamut from machining to iron mongery and forging to soldering and brazing and can include everything from a plumber on the one hand to someone who designs internal combustion engines on the other.
Fortunately we have the terminology to describe these different jobs reasonably well. It's not easy to confuse a plumber with a machinist. But that's not true when it comes to tech. We have very little terminology in that realm. It's non-trivial to explain the difference between a systems programmer, a DevOps automation guru, a mobile app dev, a web dev, etc. A lot of these jobs end up falling under the standard "software developer" heading. Even though that's about as silly as calling taxi drivers and truck drivers "automobile operators".
This problem of terminology extends into the tech industry as well. We have sufficiently developed mental models of things like restaurants to be able to understand the differences, and difficulties of changing, between sit-down, take out, delivery, and catering focused establishments. Yet when it comes to tech our brains go on vacation and suddenly the equivalent of a high-class sit-down restaurant switching to making pre-made meals to be sold in grocery stores seems trivial when it is anything but.
You get the same effect when people ruminate on how Apple should spend its huge cash reserves. With people throwing out crazy ideas of how Apple should buy into some completely different industry that would totally change the nature of the company, which is just as ridiculous as the idea they should switch to building rocket ships or submarines.
What's interesting is that tech is very precise and yet plumbing is not (it's analog with some margin for error just like cooking as another example).
But both of those things have been around for much longer than tech so perhaps the roles are clearly defined for that reason. (You probably could find counter examples of course.)
Maybe this stems from the fact that the general public doesn't typically deal with "computer guys" [1] so they have no need to differentiate them. But they do need plumbers, carpenters and electricians and they know what they are. [2] (And those are also union jobs so that probably also makes titles necessary).
[1] How many times have you heard "he does something with computers?".
[2] Maybe this all stems from the fact that there was no central defining authority at any point that made any statement as to what the titles meant. Even in religion there are clearly defined titles, right? In tech (on the net at least) what do we have? RFC's a nice way of asking opinions.
But both of those things have been around for much longer than tech so perhaps the roles are clearly defined for that reason. (You probably could find counter examples of course.)
Maybe this stems from the fact that the general public doesn't typically deal with "computer guys" [1] so they have no need to differentiate them. But they do need plumbers, carpenters and electricians and they know what they are. [2] (And those are also union jobs so that probably also makes titles necessary).
[1] How many times have you heard "he does something with computers?".
[2] Maybe this all stems from the fact that there was no central defining authority at any point that made any statement as to what the titles meant. Even in religion there are clearly defined titles, right? In tech (on the net at least) what do we have? RFC's a nice way of asking opinions.
Which of the 3 allow you to reach an actual human, quickly, on the weekends, even if you are spending less than $5K per month?
Asking for real, my guess is RS does, but I don't know about the other 2.
Asking for real, my guess is RS does, but I don't know about the other 2.
Rackspace has a live chat which is 24/7 with knowledgeable support techs. Bugged them twice yesterday and they responded very quickly. My last rackspace bill was 60$.
$100 (or 10% of your spend whichever is greater) gets you that at AWS (https://aws.amazon.com/premiumsupport/)
It seems to me that, in spite of the advent of "cloud-based" hosting providers such as AWS and GAE, the demand for more traditional hosting, ranging from RamNode to DigitalOcean to Hetzner, is not waning at all, though this is just a personal perception.
On the other hand, it baffles me that providers such as Hetzner have not had a greater impact on the price charged by Rackspace and others. It's also surprising to see most of the best prices for dedicated servers coming from Europe and not the US.
On the other hand, it baffles me that providers such as Hetzner have not had a greater impact on the price charged by Rackspace and others. It's also surprising to see most of the best prices for dedicated servers coming from Europe and not the US.
Excellent service doesn't go out of style.
I've used Pair, Linode, AWS, Rackspace and Bytemark.
Pair, Bytemark and Rackspace pick up the phone. Bytemark has the most chops by far, and is flexible to support custom stuff. If you just love *BSD, Pair of course.
Pair, Bytemark and Rackspace pick up the phone. Bytemark has the most chops by far, and is flexible to support custom stuff. If you just love *BSD, Pair of course.
Separately I'm wondering if "jnoller" is still working at Rackspace. Noting that there was a flurry of activity on HN from his comments that appeared to have stopped about a month ago.
He is; he's part of the Developer Relations Group. I know he's been on vacation recently, that's probably part of the reason he hasn't posted as often.
Yup. Still employed happily at rackspace. Just been on vacation at home the last few weeks. Also, busy busy busy - I come back to work then off to codemash! I don't spend all of my time here :)
Someone will still need to offer putting physical servers in a data center. Not all workloads are well served by cloud/VPS offerings.
But you've offered no particular evidence why that someone will be Rackspace.
Racksapce has offered this service since the late 90s.
Barring some other external force - why would one expect they wouldn't continue as a player in a market they helped create?
But without Rackspace, where will Scoble work?
Anecdotally, I would say Rackspace has the most intuitive interface and good telephone support. They've just released new performance instances which are cheaper than AWS: $0.08/hr for 2 vCPUs, 2GB RAM and plenty of SSD vs $0.12/hr for AWS 2 vCPU m1.medium.
Our local VPS has fantastic customer service and we know each other by name.
Amazon has the real truly global infrastructure (Rackspace UK is a separate entity to US, ASIA) and Route 53 Anycast is outstanding.
We'll be trialling out Google Compute Engine to see how it fares.
For now I see Amazon for global DNS infrastructure, but I can't see Rackspace going anywhere.
Seems like Gartner have previous with Rackspace: http://www.theregister.co.uk/2013/11/20/gartner_openstack_cr...
Also interesting reading: http://gigaom.com/2013/12/20/backbreaking-openstack-migratio...