Slashdot Mirror

← Back to Stories (view on slashdot.org)

Amazon, Rackspace Add New Cloud Capabilities

Posted by samzenpus on from the gold-and-silver-lining dept.

miller60 writes "Amazon Web Services has rolled out Elastic Beanstalk, a free feature which automatically handles the deployment details of capacity provisioning, load balancing, auto-scaling, and application health monitoring. AWS execs tell GigaOm that Beanstalk represents a move up to Platform-as-a-Service and is designed 'to address the idea of vendor lock-in and inflexibility that commonly afflicts other platforms for application development.' Meanwhile, Amazon rival Rackspace Hosting has extended its cloud platform to its European data centers, opening the service to customers bound by data protection regulations, and says it now has more than 100,000 cloud customers."

45 comments

  1. So, in other words by Locke2005 · · Score: 3, Insightful

    Amazon claims that a feature that only they offer helps prevent vendor lock-in?!?

    --
    I've abandoned my search for truth; now I'm just looking for some useful delusions.
    1. Re:So, in other words by Anonymous Coward · · Score: 1

      Nope. Amazon claims that by using standard tools (So far, Java + Tomcat server), their offer is easy to migrate from - thereby preventing vendor lock in.

    2. Re:So, in other words by farnsworth · · Score: 4, Insightful

      Amazon claims that a feature that only they offer helps prevent vendor lock-in?!?

      No:

      Today's release of Elastic Beanstalk is built for Java developers using the familiar Apache Tomcat software stack, which ensures easy portability if you ever want to move your applications. Elastic Beanstalk is designed so that it can be extended to support multiple development stacks and programming languages in the future.

      Compare this to, say, writing an app using AppEngine and the lack of lock-in becomes clearer.

      --

      There aint no pancake so thin it doesn't have two sides.

    3. Re:So, in other words by TooMuchToDo · · Score: 0

      Until I can migrate from clustered computing environment to clustered computing environment (first person to say "cloud" gets punched in the chest) with one click (with DNS, IP management, etc all handled my a hypervisor headmaster), there is still a fair amount of lock-in.

    4. Re:So, in other words by DragonWriter · · Score: 1

      Compare this to, say, writing an app using AppEngine and the lack of lock-in becomes clearer.

      Given open source implementations of the AppEngine stack like AppScale and TyphoonAE, I'm not sure the comparison is as clear as you suggest.

    5. Re:So, in other words by dave562 · · Score: 1

      You should go ahead and write the functionality that you want. I'm sure it will only take you a weekend. Just think, it's probably in such demand that you'll be rich by Monday.

      Get back to me when you're done.

    6. Re:So, in other words by cain · · Score: 2

      I don't think he'll do it. I mean check out his nick.

    7. Re:So, in other words by dave562 · · Score: 1

      I think is nick is a bit hyperbole. He's posting on /. with the rest of us after all.

    8. Re:So, in other words by TooMuchToDo · · Score: 1

      So I have to put in the time to demand a feature? I'd happily write a check. I'd support openstack financially with $10K or $20K if I knew it wouldn't help Rackspace out.

      Some of us can make enough cash to buy the things we need, because time (as my nick points out) is quite often more expensive than cash.

    9. Re:So, in other words by dave562 · · Score: 1

      The feature that you're asking for is unheard of and fanciful. I've only been doing IT for 15 years, so maybe there was time before I got into the game when what you asked for was a reality. But from everything I've seen, any sort of migration is a long and involved process, whether or not you are "locked in" to a product.

      To say, "until I can migrate ... with one click" shows a complete lack of understanding about what you are asking for. You're going to need more than $20K to convince an ASP to give you one-click "move to the competition" functionality. In fact, even if you're billing your time at lawyer rates, I doubt that ten years of your time would be enough to convince anyone to develop what you want.

      I take that back. You could convince someone to develop it. You couldn't convince them to put it into production. Just look at how Google and Facebook treat each other when it comes to sharing contact data. Contact data is just a few columns and maybe a few hundred rows. You want a complete environment migration... one click?

      Thanks for the laugh.

    10. Re:So, in other words by tomz16 · · Score: 1

      Amazon claims that a feature that only they offer helps prevent vendor lock-in?!?

      Perhaps, but pragmatically the APIs for Eucalyptus are identical to the corresponding amazon services (e.g. EC2, EBS, S3, Elastic ip, etc.). This means that you can dump amazon any time for another provider deploying a cloud solution based on Eucalyptus or even decide to bring up your own home-grown cloud. You can even run hybrid solutions (eucalyptus in your own data center supplemented with EC2 using a single codebase)

      (You may have noticed the "Install Ubuntu Enterprise Cloud" option on the newest releases... Canonical claims that you can have a basic UEC install up and running in as little as 25 minutes. I've never actually tried it, but the instructions look surprisingly simple)

    11. Re:So, in other words by Anonymous Coward · · Score: 0

      Meh. It's no more vendor lock-in than having physical servers located in a rack you rented at the local data center.

    12. Re:So, in other words by t2t10 · · Score: 1

      You can't do that with other kinds of hosting either, so the amount of "lock in" is not greater than if you run things yourself or use traditional hosting.

    13. Re:So, in other words by TooMuchToDo · · Score: 2

      I'm somewhat disappointed in the moderation on my post. It's not like I'm some PHB who doesn't understand technology.

      It's a matter of thinking hard about your app, how it's designed, and the coding based on that design. Architect your environment properly, and have everything accessible via APIs. Want to move your DNS? API calls to replicate the records to another provider and update the nameserver information at the root (with yet another API). Your webservers should be able to scale horizontally, as well as your app servers (if you have them) and your database servers (or, depending on your app, use something like Cassandra that is easier to replicate between 10s to 100s of nodes than MySQL/PostgreSQL/etc). Any app is simply the function of several "cogs" of codes, some apps have more cogs than others (Amazon.com, for example, hence all the services that drive it can also drive AWS).

      To dismiss my post is to dismiss the advancements in computing over the last two decades. If you do decide to dismiss me, don't say it's because it's hard and because we can't. That just seems to be an untenable conclusion.

      If you don't believe that this is the direction we're going in, I'd think others would disagree: http://twilio.jobscore.com/job_seeker/jobs/job_posting?job_id=bSHjYWq2Sr37U6eJe4aGWH&ref=rss

      About the Job:

      Twilio runs 100% in the cloud. Technically we run in "the clouds"- using servers in multiple clouds based on the price, availability, failover & bandwidth of different offerings.

      We already bring machines up and down and auto-configure them with the push of a button using a first-generation tool we built called boxconfig. (we know - soooooo original & creative)

      We are looking for someone to take our mission-critical cloud infrastructure management system to the next level. This system will manage our telecommunications infrastructure cluster – it will orchestrate the provisioning, load balancing, dynamic configuration/re-configuration, monitoring and spend optimization of 10,000+ servers across providers, data centers, availability zones and myriad other variables we haven't even thought of yet.

      You've been itching for the opportunity "do server management infrastructure right" for a while and are fired up to absolutely go to town on this - building scaling and healing automation that factors in security, failover, and quality/analytic tools to track stuff like packet loss, performance, latency, and more. You know, the stuff you'd build if world class infrastructure was the priority - and your boss wasn’t breathing down your neck about that i18n feature and the other whizbang things the marketing and sales VPs need yesterday.

    14. Re:So, in other words by TooMuchToDo · · Score: 1

      But I'd argue it's inherently easier to accomplish in "cloud computing" environments. It's just a matter of ensuring source/destination hypervisors can run the virtual machine images you're using, store the same amount of data using the same methods, etc.

      Think how Amazon AWS currently works. S3 and EBS for storage, EC2 for machine images, hell, they even have Route 53 for anycasted DNS (although it's not as good as dedicated DNS providers).

      Now, if other providers can provide API-compliant, compatible environments, it's a matter of a) migrating your DNS with APIs, b) copying over your server images and starting them up, and c) migrating your storage. This can't be done easily why now?

    15. Re:So, in other words by Anonymous Coward · · Score: 0

      Standing Cloud (in Boulder CO) is working on something like that -- switching between cluster platforms in a way that is transparent to the customer.

    16. Re:So, in other words by Anonymous Coward · · Score: 0

      Rightscale supports migration between the various clustered computing environments.

    17. Re:So, in other words by TooMuchToDo · · Score: 1

      Laugh if you want, people are building the feature set as we speak:

      http://twilio.jobscore.com/job_seeker/jobs/job_posting?job_id=bSHjYWq2Sr37U6eJe4aGWH&ref=rss

  2. Wikileaks by Tokolosh · · Score: 0

    How's that working for ya?

    --
    Prove anything by multiplying Huge Number times Tiny Number
  3. Undoubtedly... by mschaffer · · Score: 0

    They are undoubtedly trying to cash in on the publicity they received from the WikiLeaks scandal.

  4. Dreadful name by syousef · · Score: 1

    "Elastic Beenstalk" sounds like someone's been smoking too much pot.

    --
    These posts express my own personal views, not those of my employer
    1. Re:Dreadful name by Juba · · Score: 2

      Next services they're going to offer : "rubberized wheelbarrow" to help you migrate from one cloud to another, and "ducttaped showerhose" to ease communication between several cloud-deplyed load-balanced webapps.

    2. Re:Dreadful name by Anonymous Coward · · Score: 0

      Well, "Elastic" part is prefix they use for all kinds of things (Elastic Map/Reduce), so nothing odd about that part. Beanstalk, maybe so; but it's not the name one should pay too much attention to, but rather actions of the vendor in question. Bit of pot might not be bad idea for manager/executive level of AWS, wrt their "support" of their customers.

    3. Re:Dreadful name by gandhi_2 · · Score: 1

      In 1966, I went down to Greenwich Village, New York City to a rock club called Electric Banana. Don't look for it; it's not there anymore.

      --
      THL phish sticks
    4. Re:Dreadful name by Anonymous Coward · · Score: 0

      holy shit! I was there one night in 1966, too. Some dude gave me the best blowjob ever. Small world, eh?

  5. Re:Exucse me? by Anonymous Coward · · Score: 1

    if only they would switch the woot.com servers to amazon servers, no more page timeout during a BOC.

  6. Do I Trust The Cloud? by rueger · · Score: 1

    I'm really torn on the whole "cloud" idea.

    On one hand I do use Gmail and just yesterday came away pretty impressed by http://pixlr.com./ I like the convenience, and I like the way many of of these services keep my desktop and Android phone in sync.


    It's one thing to know that your data lives on a specific server box in a specific geographical location. Are we reaching a point where you can't even nail down a specific country that is home to your information?

    --
    Three Squirrels
    1. Re:Do I Trust The Cloud? by alex67500 · · Score: 0

      Man, I live in London, UK, and I trust the clouds: when I see them, I pack my umbrella!

    2. Re:Do I Trust The Cloud? by Anonymous Coward · · Score: 0

      Just encrypt your data. This is what should be done anyway. With that, it becomes bit less of an issue that someone somewhere has access to your data.

    3. Re:Do I Trust The Cloud? by giles+hogben · · Score: 1

      Until you want to do anything with your data other than have it sit there - remember homomorphic encryption doesn't work and probably never will...

  7. Ode to Slashdot, a haiku by Anonymous Coward · · Score: 0

    slashdot.org
    highest rated comments, print
    little brown ring
    big white ring
    centered
    a log slides out
    a splash of cold water
    bullseye, a pucker
    printout, wipe wipe wipe
    a daily experience

  8. As long as you have a "connection..." by Anonymous Coward · · Score: 1

    This whole idea of "services/software" in the Cloud for your business sounds good. The hype is there. As long as you have an Internet connection. And, a backup internet connection.

    1. Re:As long as you have a "connection..." by rainsford · · Score: 1

      True, but it's not like there aren't points of failure for locally hosted data. "Cloud" services seem like a good choice for many businesses because it greatly decreases the amount of IT a business has to manage. And when it comes right down to it, a reliable Internet connection is almost certainly a requirement for a business IT setup no matter WHERE their data is hosted.

    2. Re:As long as you have a "connection..." by alex67500 · · Score: 0

      Well... Gears is starting to help a bit, but nobody is making the most of it just now.

      (on another note, whenever I lose my connection, I just lay there hoping I'll die soon or get a connection back -- whichever the shortest of the 2)

  9. Yay! by Anonymous Coward · · Score: 0

    This is the perfect solution to all my lock-in and inflexibility problems! Tomcat!

  10. SQL Server has entered the building by wdhowellsr · · Score: 1

    I tested Rackspace cloud services for a client that has seasonal swings in site hits but could not recommend it because MS SQL Server was not available on the virtual servers. I even pestered my brother-in-law at MS.

    They now have MS SQL Server available on a per vserver basis.

    I'm definitely looking at them again because you can programmatically create and destroy new server instances and you only pay for the time they exist.

    Everything is grayscale at night.

    1. Re:SQL Server has entered the building by juuri · · Score: 1

      You are doing it wrong if you want to use MS SQL server on dynamic virtual servers.

      There are far better SQL and NoSQL solutions for such instances, not everything is a hammer.

      --
      --- I do not moderate.
    2. Re:SQL Server has entered the building by wdhowellsr · · Score: 1

      I agree completely but I'm an enterprise .Net developer whose clients refuse to use anything other than Microsoft or Oracle databases. Unfortunately Oracle doesn't currently offer a virtual installation.

      When Microsoft came out with the first .Net products I decided to take the plunge and focus on MS .Net developing exclusively and it's worked out pretty well.

  11. Don't confuse "cloud" with "mesh" by countertrolling · · Score: 1

    Mesh is cool

    The "Cloud" isn't.

    --
    For justice, we must go to Don Corleone
  12. Puppet by bill_mcgonigle · · Score: 1

    Until I can migrate from clustered computing environment to clustered computing environment (first person to say "cloud" gets punched in the chest) with one click (with DNS, IP management, etc all handled my a hypervisor headmaster), there is still a fair amount of lock-in.

    Sounds like you're looking to make Puppet better integrated with the various bits you want to manage. vCloud is Scott Ullrich's proposal to get a puppet definition of your whole environment deployed. vCloudBSD is the alpha-version implementation of that idea.

    Maybe pay Puppet Labs to make Puppet do the bits you need that it doesn't already do (e.g. does it already handle Xen?), port vCloud to linux or whatever you need. Getting Amazon to implement vCloud or something like it might be hard - probably you'd have more luck with Rackspace since they seem to have a better community engagement ethic.

    And then you'll need to hire somebody to build you a GUI to make the 1-button interface. Everybody else will start it from the command line. ;)

    --
    My God, it's Full of Source!
    OUTSIDE_IP=$(dig +short my.ip @outsideip.net)
  13. List of two elements by Anonymous Coward · · Score: 0, Offtopic

    Let me introduce the editors to a marvelous new character, which we call the "ampersand":

    &

    This can be used in place of the word "and" when writing a list of two elements.

    For example,

    Amazon and Rackspace

    becomes

    Amazon & Rackspace.

  14. Every time they talk about the cloud... by Black+Art · · Score: 1

    I hear bong water bubbling in the background.

    --
    "Trademarks are the heraldry of the new feudalism."
  15. So, two government stooges in the cloud by Anonymous Coward · · Score: 1

    Rackspace who shafted their customers and democracy for an illegal demand on Indymedia and Amazon who shafted democracy by pulling the plug on Wikileaks are now adding new cloud capabilities.

    Please excuse me if I don't use it.

    (PS I still find it highly ironic that Rackspace have their "Fanatical Support" strapline after their craven collapse over Indymedia.

  16. Re:Exucse me? by Anonymous Coward · · Score: 0

    It's not only the spammers, but also the people running Zeus. I will have to say that they can probably probe more systems and networks per hour using Rackspace. One thing that I've noticed in the last couple of weeks is that they seem to make no more than 90 probes into a network at a time., so they're in, and out (if you're prepared) in less than 1 second.