Blog Post

Bad news for Amazon could be good news for “other” cloud providers

If other cloud providers — the Rackspaces(s rax), Hewlett-Packard’s (s hpq), Joyents, Softlayers, Terremarks(s vz) and GoGrids of the world — don’t take advantage of Amazon Web Services data center snafus over the past year, they should have their heads examined. What competitor would not tactfully point out to corporate prospects that AWS US-East has been ground zero for at least three major sets of outages over the past 12 months?

For all the talk of Amazon’s dominance in public cloud, we’re still very early on in this game. Outside of web startups and SaaS companies, many businesses have barely dipped their toe in the cloud to date. That means there is big opportunity for those vendors that can provide an enterprise-class cloud. At the very least, Amazon’s miscues have definitely prompted these companies to think of using multiple clouds and/or evaluate private vs. public cloud where they feel they have more control. So the other cloud players have to play up their strengths either as a replacement or an adjunct to AWS.

Differentiate your cloud from the rest

Rackspace_Logo_08_07_2012[2]Not surprisingly, Rackspace(s rax) is turning up the volume on its “fanatical support” pledge to woo business users who really really want a phone number to call and a hand to hold during cloud deployment — and as needed thereafter. Many customers also like that Rackspace lets them mix workloads between private and public clouds. That private-public-cloud mix is one advantage HP, Softlayer, Joyent and others can offer over AWS. To be fair, Amazon has not stood still here. It’s pushing its own Virtual Private Cloud, that lets users cordon off a chunk of AWS infrastructure for their own use. It has also struck an alliance with Eucalyptus to ease hybrid cloud deployments between AWS and Eucalyptus-based clouds.

Bestfit Mobile has moved some workloads from Amazon to Rackspace because it hit scaling problems with Amazon’s load balancing, said Ray Williamson, VP of software for Bestfit. Rolling out a mobile app for a big retail customer, Bestfit initially launched all the data on Amazon but ran into scaling issues with its load balancer. “Amazon uses a software load balancer — cloud is great for many things but sometimes you need real hardware.” Williamson said.

Sometimes the best cloud is a mixed cloud

Bestfit ended up deploying a mixture of on-premises hardware load balancers and cloud-based infrastructure linked by a fast Rackspace Connect connection without huge additional cost.

But support was also a huge differentiator. “With Amazon it’s hard to get an engineer on the phone. With Rackspace, we get help. We call, they call back and the Rackspace guys can log in and fix the problem if needed. AWS cannot log into your machine by policy,” he said.

San Francisco-based Joyent has some high-profile customers like LinkedIn(s lnkd) but remains something of a stealth player — albeit one with a solid reputation if you drill down. It, like Softlayer, has made noises in the past about taking on Amazon, but has not been blatant in doing so. Still, Hoffman’s  not shy about saying that AWS US-East outages are not flukes. Instead, he said they result from underlying architectural issues with how its control plane is set up that make some services, notably Elastic Load Balancer, vulnerable, in his view.

Jason Hoffman“For some of these services one failure leads to another to another and the whole thing has to be rebooted to fix. Theser are not ‘oh gosh, oh golly gee, something happened’ but things that will keep on happening and it’s why they keep happening, “Hoffman said.

Joyent CTO Jason Hoffman said the company sees lots of prospects that are AWS users. “Do we sit down and go after [AWS] in our materials? No. But we do talk about our availability from an uptime perspective and in aggregate we do less than 10 seconds of downtime per year. So from an availability standpoint we do very well and our conversations with customers touch on that.”

Joyent, Softlayer, Neogrid, OpSource and NephoCloud are what David Linthicum, founder of Blue Mountain Labs and a GigaOM Pro analyst characterizes as “found clouds.”

“They are typically not on the A list, but are found through evaluation and analysis.   I put them on the radar for my clients, but I always get a ‘who?'”

Beware of AWS bashing

It’s probably smart not to be too blatantly opportunistic in bashing Amazon. “Most cloud providers suffer outages from time-to-time, AWS is just more scrutinized than the rest since they are the leader,” Linthicum cautioned. Any cloud company could blast AWS this week and have an outage tomorrow. “However, they certainly can provide up-time statistics that compare their services with those of AWS, and other players. That’s both fair, and productive.”

 Feature photo courtesy of  Flickr user Editor B

10 Responses to “Bad news for Amazon could be good news for “other” cloud providers”

  1. robert38

    A few points worth making. Firstly, AWS has high churn in their customer base, this is masked by very fast customer uptake in the general IaaS ramp-up period. So these outages do significantly hurt any cloud provider including AWS. In fact at our company we see >50% of new customers coming from AWS and others like Rackspace. In other words many start with AWS and move on as they become more experienced and sophisticated.

    Secondly coming back @Jon Henshaw, he’s right, clearly being larger means they hit barriers others don’t however not all cloud providers have a mass market strategy so don’t need to deal with the truly massive scale of AWS. If you were a customer, is there an advantage to you of choosing a massive cloud provider given the scaling problems they have? Also related to this, as basically the first public cloud, AWS has many legacy issues related to technology choices made back in 2006 that made sense then but make no sense now. Again, newer cloud providers benefit from a more mature state of IaaS technologies and a wider choice. Pioneers get arrows in their backs sometimes etc.

    Most serious users consider multiple clouds nowadays and that will continue. Doing this also mitigates both outage and commercial risks associated with using one provider.

    Robert Jenkins

  2. James Mitchell

    Having several providers that are of a similar size to Amazon will increase enterprise uptake to the benefit of all providers. Most of our enterprise prospects take an Amazon + 1 (or more) approach to building infrastructure in the public cloud. I think this is the year of the multi-cloud deployment.

  3. Cloudguy

    AWS importance to Amazon is a myth, as is its “dominant presence” in the cloud space.

    At present it is estimated that Amazon generates $1.5 Billion in revenues from AWS (they do not disclose this, these are analyst estimates). This is only 3% of Amazon’s total revenues.

    A lot of hype AWS is ….

  4. Lawrence Sica

    The support angle baffles me because you can and do get some good support from them. It just depends on some things.

    The first thing I’ve found is that if you have an AWS rep then your support level is really good. The rep I deal with gets people involved as needed if I call him. We get their engineers on the line, on site even. But it comes down that you get what you pay for.

    Also i’ve found that if you pay for business or better support it’s a lot better overall.

    These other players comments interest me because I want to see how they would handle the same level of load and scale AWS has.

  5. Support is a good differentiator and Rackspace have built a reputation of offering good 24/7 support by phone and tickets through their managed hosting, which has now been transferred to the cloud. But don’t forget that Amazon do offer this as an addon you can purchase, choosing different levels of support depending on what your needs are. This goes with the whole philosophy of AWS in that you choose every component a la carte, compared to Rackspace where you pay more for fewer options because more things are there by default.

    It’s also important to remember there are 2 kinds of support initiators – 1) general tech support questions about local problems, architecture choices, etc. 2) Major outages. For the second, you’ll get the same kind of response regardless of which provider you’re with because you ultimately have no control over when things get fixed. I’ve had occasions when I couldn’t get any further useful info or even failed to get through when Rackspace had major outages.

    I think we’ll see some interesting releases this year along the lines of providing performance and availability guarantees (and not just legal SLAs which are only used after the fact). We’re already starting to see this with Provisioned iOPS from AWS and the SSD option for Rackspace Block Storage. Even Rackspace have an extra Managed level of support you have to pay more for on their cloud.

    • Lawrence Sica

      Good point on #2. Major outage issues are a good indicator of how a service responds. IMHO the official response from AWS was good though I want, and have asked for more information.

      I think SLA will become, besides the big data push, an important factor this year. Cloud computing is now accepted as a concept in general for business. Next step is the other bits.

    • Very good points. It’s important to not that Rackspace has had their issues with up time in the past as well in their hosted solution. I’m hopeful that their support first based support will garner additional market share. The overall industry doesn’t benefit from a single dominate player.

  6. Jon Henshaw

    AWS has a head start on scaling the cloud, which means they will hit technological walls (outages) before anyone else does. They will also come up with a solution for each instance so the same thing doesn’t happen again. I would bet that as others grow, they too will experience similar outages, which will just make AWS look all the better. The fickle, impatient users may leave AWS, but they will back, as that’s what fickle short-sighted users do.