Forgot your password?
typodupeerror
The Internet

Sending Excess Load To the Cloud? 153

Posted by kdawson
from the failing-over-into-the-mist dept.
TristanBrotherton writes "Cloud computing seems to be a good choice for startups like ours, looking to scale easily with users. (We're providing a series of Web services, assets, and Web applications to users of our mobile client.) There are the obvious choices of Google, Amazon, and smaller shops like EngineYard. The biggest issue we have in choosing cloud computing to run our applications is trust in their robustness. If the provider goes down, we suffer. In traditional hosting environments we mitigate this with multiple sites / vendors. It's not really feasible to host on multiple compute services, so I wondered if a better option might be to set up a small (perhaps two servers) origin infrastructure in a traditional manner at a datacenter, running our applications, but then send excess load, or in the event of our origin servers failing, all load, to compute services. This would give us the best of both worlds. Has anyone done this, or had experience in designing Web applications to scale seamlessly across both environments? Is there particular load-balancing hardware we can use to do this?"
This discussion has been archived. No new comments can be posted.

Sending Excess Load To the Cloud?

Comments Filter:
  • by Animats (122034) on Tuesday September 30, 2008 @02:52AM (#25202043) Homepage

    Unless your "cloud" provider offers a service level guarantee with teeth, is contractually obligated to continue to provide the service for some period of time, and has sound financial fundamentals, this is risky.

    I think we'll see a big shutdown of money-losing web services over the next year.

    • by bigsteve@dstc (140392) on Tuesday September 30, 2008 @03:20AM (#25202153)
      I don't think you can ever say that an IT company will still be in 5 years time, no matter how good their financial fundamentals look today. And if they want to avoid being sued out of business, they probably won't sign up to a contract with the kind of guarantees and penalties that you really want.

      The answer is to not get tied into a single service provider. You need a cloud computing solution that is standard-based (formal or defacto) and that lots of providers are supporting. And you have to be prepared to migrate your stuff if/when the industry moves on to the next version of the standard ... or the next "big leap forward" after cloud computing.

      And that is all hard to do.

      • Re: (Score:3, Insightful)

        by Hognoxious (631665)

        And if they want to avoid being sued out of business, they probably won't sign up to a contract with the kind of guarantees and penalties that you really want.

        That will certainly prevent them being sued out of business.

        However with no customers, they might go out of business anyway.

        • Re: (Score:3, Insightful)

          by bigsteve@dstc (140392)
          I predict that the "no guarantees" commodity Cloud Computing service providers will get lots customers anyway; plenty to run a business on. The cost of supporting customers with strong guarantees of service is significant, and it will be passed on to the customers. Most customers will not be willing or able to pay for this.
          • Re: (Score:3, Insightful)

            by Hognoxious (631665)

            What kind of customers? Presumably not ones who stand to lose a lot of money if there's any kind of loss of service; they'd be willing to pay the premium for guaranteed relaibility.

            In other words, if you aren't offering that you're aiming at the ones who don't make a lot of money to start with. They're often not worth the effort. It might be OK for a mom and pop novelty cake business, but would you want your bank hosting its apps on some nickel & dime service that try their best but it may or may not

            • What kind of customers?

              The kind that Google and Amazon are aiming to support :-)

            • No, they'll get those customers anyway. That will work the same way as non-guaranteed software gets all enterprize customers because it has "support", while supported software does not.

              And when the clould fails to meet the requirements, people will just say that everybody else fails, so there is no better way of doing things.

            • by mstahl (701501)

              Well, for one, customers like the startup where I used to work before getting laid off in the spring. We used EngineYard for our stuff which was beautiful. Scalability was great and everything worked really well with our ruby on rails application. Only problem was that it was absurdly expensive for the amount of reliability we got.

              At one point I got a panicked phone call in the middle of the night that the site was down, and checked my email and it was EngineYard apparently accidentally switching something

      • Re: (Score:3, Insightful)

        There is no such thing as a standards-based cloud computing solution. The best you can get is some middleware layers that try to abstract the APIs for a few different providers.
        Since these APIs are a moving target, these middleware solutions are not very effective.

    • Re: (Score:3, Insightful)

      by anvay.lonkar (1079879)
      Ha ha, if googles shutting down I am pretty sure the market sucks enough for a company like me (providing mobile content services) will not need their cloud
      • Re: (Score:2, Insightful)

        by MrNaz (730548)

        The Internet would survive Google's demise. It was designed to survive a nuclear attack and it survived Excite's demise. It'd survive without Google.

        Just a reminder, as many people seem to have forgotten:

        Google != The Internet

    • Re: (Score:1, Informative)

      by Anonymous Coward

      The cloud with an SLA...

      http://bluelock.clients.cantaloupe.tv/?ref=twg

    • by JnKor (1089201)
      I completely agree. Last year I was part of a team trying to design and implement seamless integration with cloud services. Early in the planning phase, it become apparent that without contractual service guarantees, there is no way to threshold the decision between local and remote execution.
    • by ooglek (98453)

      Amazon S3 has an 99.9% SLA guarantee, but EC2 has no such SLA guarantee that I could find.

      Even so, Amazon's EC2 offers the closest to portable hosted "cloud" available that I know of. They use xen, you can convert xen images to Amazon's AMI's (their own image type), and technically you should be able to convert an AMI back to a xen image and run that image on your own or another xen server platform.

      With the others you mentioned, you are writing your code specifically for those clouds, and if they go down,

  • The C word (Score:4, Funny)

    by jaxtherat (1165473) on Tuesday September 30, 2008 @02:53AM (#25202053) Homepage

    Please don't use it. Every time you use a buzz-phrase God kills a kitten.

    • I totally expect a new round of old patents - but on the cloud!
    • by Nursie (632944)

      I strongly agree.

      It's a really annoying buzzword that seems to cover everything from VMWare to mainframes, to Beowulf clusters, SAN technology....

      What is the fscking cloud?

      Also, are all the offerings he mentions targeted at people running web applications anyway?

      • by MrNaz (730548) on Tuesday September 30, 2008 @06:39AM (#25202747) Homepage

        It's a term invented by idiot managers who saw all those diagrams where the wider internet is represented by a picture of a cloud and were too stupid to grasp the concept of a representative diagram, so they took the picture of the cloud to be literal, and now there is an entire generation of managers who have an image of electrons flying around the sky. They confusion they suffer is only exacerbated when there's a thunderstorm and they hear the word "torrent" to describe the rain, thinking that the storm is the result of those damned P2P users.

        • Re: (Score:3, Informative)

          by MightyYar (622222)

          Well, you ain't gonna change people, so unless you can come up with something catchier than "cloud" you'll have to endure it :)

          • Re: (Score:3, Insightful)

            by John Hasler (414242)

            "Timesharing service" is available.

      • What is the fscking cloud?

        It's just a shorthand for where all the corporate CIOs' heads are right now: up in the clouds.

        Don't worry, their feet will be brought back to the ground in short order as a result of the current world financial problems.

    • by Eli Gottlieb (917758) <eligottliebNO@SPAMgmail.com> on Tuesday September 30, 2008 @11:54AM (#25205431) Homepage Journal

      Web 2.0! Cloud computing! Thin clients! The network is the computer! Software as a service! Utility computing! WHERE DO YOU WANT TO GO TODAY!?

      I really hate cats.

  • All the apps I have used on "the cloud" are over-rated, if google would have just stuck to a few apps and focused on improving them, they would not have spread themselves so remarkably thin. I think this is where independent smaller software companies can have a big advantage if they are any good at recognizing the users needs.

    A user wants to:

    -Save time
    -Save money
    or both

    Users will gladly part with money if your software adds real value to their lives, they won't if you're just trying to repackage the same

    • by RuBLed (995686) on Tuesday September 30, 2008 @03:23AM (#25202157)

      "jack of all trades, mater of none"

      yup, im in slashdot alright...

    • by dracocat (554744)

      Did you read the question? Just curious...

      • Yes I did, but what does he think he's going to accomplish by shifting the service into the cloud, why not just have backup servers in many datacenters? What exactly is the difference?

        • by vidarh (309115) <vidar@hokstad.com> on Tuesday September 30, 2008 @06:00AM (#25202615) Homepage Journal
          Cost and complexity.

          Have you tried managing racks worth of servers in many locations?

          That said, most cloud services today ARE very expensive. EC2, for example, can be trivially beaten with managed hosting, and in some cases totally crushed by maintaining your own servers.

          What cloud services give you (and you pay through the nose for) is the ability to scale quickly. Trouble is, most people never need to scale that quickly.

          Using clouds for "overflow" from a cheaper base setup is not a new idea, and it's definitively a good one. Particularly since it allows you to cut it a lot closer with your base setup. Without overflow capacity elsewhere, you need enough extra capacity in your base setup to handle reasonable growth plus any spikes. With overflow capacity using a cloud service, you only need to handle enough of your daily traffic that whatever you end up using of the overflow capacity is cheaper than adding more servers to your base. As soon as it isn't, you add more servers.

          • Re: (Score:3, Insightful)

            by Kent Recal (714863)

            That said, most cloud services today ARE very expensive. EC2, for example, can be trivially beaten with managed hosting, and in some cases totally crushed by maintaining your own servers.

            I call myth. Cloud providers benefit heavily from economies of scale which is something that you as a little startup simply can't. On Amazon you can run a "midrange" server (i.e. ec2 large-instance) with plenty of traffic and a few hundred gigs of persistent storage for roughly $350/month. That is pretty close the amount th

          • That said, most cloud services today ARE very expensive. EC2, for example, can be trivially beaten with managed hosting, and in some cases totally crushed by maintaining your own servers.

            Most people who say that the Elastic Compute Cloud is expensive with respect to managed hosting fail to expand the acronym Elastic Compute Cloud and take special note of the first word, Elastic.

            If your computing needs are Static, then yes, you can do better with managed hosting or maintaining your own servers. Maybe you have a load spike at a certain time of day or time of month or time of year. Maybe it happens at unpredictable times (you get slashdotted, you get hard during events of some sort, when you

  • by Anonymous Coward

    There is always a software solution when it comes to good L4-7 problem and not just load balancer but whole application delivery controller. One that works not only in your data center, one that can easily work in virtual environment (VMWare ESX for example) or Clound. Look at www.zeus.com (only software ADC on Gartner's Magic Quadrant).

    • Re: (Score:3, Informative)

      by pimpimpim (811140)
      I thought of this problem myself for a while, when playing around with the idea to try out the "cloud". You could use pound [apsis.ch], a lot of its use for cloud computing has been discussed in the interwebs already. Biggest point of concern will be if the load balancer keeps your ssl data encrypted [amazonwebservices.com].
  • by BBCWatcher (900486) on Tuesday September 30, 2008 @03:35AM (#25202193)

    If your sole consideration is application availability, then your idea might make sense. But since you said you don't trust the application hosting company's "robustness," do you still trust them to protect and secure your data adequately? In other words, if you don't trust your IT service supplier in one dimension, why would you still trust them in other service quality dimensions?

    Have you thought about establishing a contract with a formal Service Level Agreement (SLA), including penalties and escrow (or equivalent) for non-performance? That would seem to be a much more straightforward and comprehensive way to establish "trust."

    • Re: (Score:3, Insightful)

      by wanax (46819)

      I'd tend to second that. These days even a T1 and a good server only costs several hundred a month. There's no magical cloud out there, only many loud, grimy and over-obliged companies. If you want to reduce costs, go to the regions where there is excess bandwidth.

      While there may eventually be a major market for extra processor cycles, it doesn't exist now, and trying to force the issue is early.

  • Overload (Score:3, Informative)

    by debatem1 (1087307) on Tuesday September 30, 2008 @03:36AM (#25202195)
    Break n' bake servers work out really well with Amazon's EC2. I've never had to use it for anything really critical but so long as you maintain a set of closely sync'd liveUSB and AMI images I don't see why you'd have a problem. Just make sure that your existing failover mechanisms automatically initiate the backup plan, notify you, and isolate your local system for forensics or repair, since a security breach that will take down your local system has a high likelihood of succeeding in the cloud.
  • Pragmatic Advice (Score:3, Interesting)

    by Muther (1375133) on Tuesday September 30, 2008 @03:38AM (#25202203)
    It's very probable that none of these offerings will work well if your application integration is not aware of itself as a group of applications and services.

    Think about it: Install Apache on one host. OK. Now, two hosts... Well, do you round robin DNS, or do you run a squid reverse proxy, do you buy something else...

    Next, how are you going to monitor this monster, Nagios or OpenView... or something else.

    How many people are responsible for this puppy?

    Oh, yeah... And I'm just talking about static web hosting, you start having all kinds of fun when you want to track user sessions, etc.

    My advice to you is to look into an Application Service Provider. Make them do all of the integration work.

    If you can afford Internet connectivity to a pair of servers, you can probably afford an Application Service Provider.
  • Given that you're talking about satisfying your base load with 2 servers and you can buy pretty decent 1U rackmount machines for $1-2k each....a)have cash on hand to buy that extra capacity b)have a vendor who can supply the equipment quickly c)tell your colo that you anticipate growing (so please don't put you in a rack where there's no room nearby, or if they can't do that, that you'll need some sort of wiring or VLANs), and d)have some sort of plan to quickly deploy the host O/S and app software. Oh yea
  • by mcrbids (148650) on Tuesday September 30, 2008 @03:52AM (#25202263) Journal

    If you are running a web-based, hosted financial application, outsourcing "to the cloud" is a non-starter. If you are hosting pictures of kitty cats, the cloud can be an excellent resource.

    Throw a server up, upload some files, start up a PG or MySQL database, and integrity is easy. But as soon as you introduce the 2nd system, integrity issues start jumping out of the woodwork. It gets worse with each additional node. Redundancy isn't just fancy-sounding, it's damned hard to do right, and as soon as you introduce it, you have to accept an elevated error rate because the number of things that can go wrong go UP, even as the number of catastrophic system failures drop.

    For a great example of redundancy in action, take a look in the mirror. You have individual cells dying by the millions every minute. Your memory is fuzzy at best, your pattern-recognition in your brain frequently sees things that aren't there, and you make stupid mistakes every single day. And that's fine, because the overall system is pretty damned redundant and resilient. A mash of protein goo and calcium deposits able to sustain one of the most complex information systems around, reliably, 24x7, for an average of 70 years or so apiece.

    Good luck getting any kind of hosting platform to maintain that kind of uptime, no matter the expense! But in biology, minor errors are so commonplace that they are hard to catalogue, let alone count.

    So pick your battle, and realize that high-performance, high-redundancy clustering is very, very difficult to do well.

    In the meantime, spend money on good quality hardware, and use top-notch colo hosting. The cost of doing it right is actually significantly lower than doing it "on the cheap" so spend money where it counts (good quality infrastructure) and save where it matters. (EG: public opinion) It's almost odd - if you look for the very, very best colo, regardless of cost, you'll find that their monetary cost is probably one of the lower ones around. (head scratcher) I've found this to be rather consistent with several reviews under my belt.

    Also, I find it best to use whitebox systems with midrange hardware. These are quality, high-performance hardware developed with everything but the name brand. In my case, I've standardized on 1U multicore X86/64 systems with hot-swap, high performance 15k SCSI drives put out by Tyan and SuperMicro. There are a large number of dealers of such systems, my current favorite is Aberdeen Inc [www.aberdeeninc]. They can sell you an amazing amount of performance and reliability for around $2500.

    This is the stuff that Sun will sell you for $8,000/pop. They will stand up to day-in, day-out heavy use for years, with hundreds or thousands of users every day, millions of website hits per day, etc. They are high performance. This is quality hardware. And with the money you save, you can have an immediate hot backup for less than the cost of the "premium support" of the big guys, and more redundancy in the meantime.

    My $0.02. Since it's free advice, you're free to use it as you see fit!

    • Re: (Score:3, Funny)

      by joe_kull (238178)

      For a great example of redundancy in action, take a look in the mirror. You have individual cells dying by the millions every minute. Your memory is fuzzy at best, your pattern-recognition in your brain frequently sees things that aren't there, and you make stupid mistakes every single day. And that's fine, because the overall system is pretty damned redundant and resilient. A mash of protein goo and calcium deposits able to sustain one of the most complex information systems around, reliably, 24x7, for an

    • by Jay L (74152) *

      Take a look in the mirror. You have individual cells dying by the millions every minute. Your memory is fuzzy at best, your pattern-recognition in your brain frequently sees things that aren't there, and you make stupid mistakes every single day

      I see you've been reading my JDate profile.

    • For a great example of redundancy in action, take a look in the mirror. You have individual cells dying by the millions every minute. Your memory is fuzzy at best, your pattern-recognition in your brain frequently sees things that aren't there, and you make stupid mistakes every single day. And that's fine, because the overall system is pretty damned redundant and resilient. A mash of protein goo and calcium deposits able to sustain one of the most complex information systems around, reliably, 24x7, for an
  • by syousef (465911) on Tuesday September 30, 2008 @04:01AM (#25202283) Journal

    Distributed computing of any kind is complex and not something to be undertaken with no experience or assistance. Hire someone who knows their stuff to help you out. Being with a business case and don't be surprised if running your own cloud turns out not to be the way to go.

    • by xant (99438)

      On the contrary sir. An amateur should attempt this; but a company with assets on the line should only attempt this with someone who understands asynchrony, parallelism and distributed computing.

      An amateur, i.e. someone working on their own time with no money on the line, should be doing this so they can learn how it's done. The feeling when you first understand asynchronous programming or some concept of parallelism is a natural high that I enthusiastically recommend. It's also a good way to get yoursel

      • by syousef (465911)

        Thanks for using phrases from 19th century England and your reminder of the literal interpretation of the word amateur (or at least one literal interpretation). Clearly I was speaking to the OP and my use of the term amateur was in relation to a person who was working professionally but without any experience. This is a common colloquial usage.

        Did you have an actual point or are you just trolling because you have too much time on your hands?

        • by xant (99438)

          Are you a complete ass? If you actually read my post you would know that I agreed with you. I was making an additional point about the value of learning. Considering your sig, you're a bit hasty to throw down the "troll" moniker.

  • Before I signed on to something like that, I'd spend some time meditating about Internet security. It's dubious at best and seems, if anything, to be slowly deteriorating. Is your situation such that you can deal with all your data being captured? How about it being altered? What future security constraints could shut you down for days, or weeks, or permanently? If you are OK on those things, then maybe.

  • by 91degrees (207121) on Tuesday September 30, 2008 @04:43AM (#25202381) Journal
    Work out how to build the servers as cheaply as possible. If peak load starts to get troublesome, add some more servers.

    Cloud computing is a buzzword. Big server farms are may be dull, but it's a tried and tested technology that works. Ask Google.
    • by Thundersnatch (671481) on Tuesday September 30, 2008 @10:31AM (#25204369) Journal

      Work out how to build the servers as cheaply as possible. If peak load starts to get troublesome, add some more servers.

      This is exactly wrong. The acquisition costs of servers is the smallest portion of operating a reliable online service. Redundant bandwidth, generator power, cooling infrastructure, and staff time are much costlier problems to solve, and represent the vast majority of expense over 3+ years. Not to mention developing the software infrastructure needed to manage high-availability scale-out applications (especially at the database tier).

      We just built a new datacenter, and server hardware was less than 6% of the total project budget. The OP should probably be looking at two or more virutal colocation providers, prefereably built on the same virtual machine platform to make failover and redundancy easier. Leave the details of supporting server hardware to the colocation provider.

      If the OP is actually going to build two or more datacenters and host his own servers (not advisable for a startup), paying Dell/HP/whomever a bit extra for 4-hour warranty service will be much cheaper in the long run than managing a bunch of white-box frankenservers whose parts become unavailable after 6 months.

  • There are a number of emerging technologies that enable very smooth scaling between single data centre (trad colo), single data centre + single cloud provider, single data centre + multiple cloud providers, multiple data centre + multiple cloud providers and all using the same base application stack.

    These emerging technologies enable the configuration of your application, using underlying grid-type infrastructure, to deploy in multiple physical and virtual environments.

    By virtualising the virtualisati
  • Best bet for DR web hosting is duplicating the applications in two or more physically different locations.
    In order to realize some return on such an expensive deployment, you need to be able to sweat all your assets. the only way to do that cost effectively is by adding some sort of load balancing across these applications. Most frequantly used solutions for hese types of environments are DNS based load balancing. Both Cisco and Radware offer a GLSB solution. In my experience, and I'll probably be killed fo

  • by PietjeJantje (917584) on Tuesday September 30, 2008 @06:33AM (#25202729)
    I'm surprised \. is posting this without referring to the Stallman interview [guardian.co.uk] that was all over the nerd sites like reddit yesterday. It is very relevant. You missed it? Come on guys, you're not always the fastest and I don't care, but this is a fail.
    • Re: (Score:3, Insightful)

      by MightyYar (622222)

      Wow, he's not taking his meds today. Gmail is a trap? How? It offers email that you can access with IMAP. Keep a copy and you can just switch later down the road - even run IMAP locally.

      I also don't understand his mantra of "keeping information in your own hands". I'd contend that some of these big outfits know more about security and reliability than it might be possible to afford as a little guy. So long as the solution is relatively standard and portable across providers, I don't see an issue.

      If you use a proprietary program or somebody else's web server, you're defenceless. You're putty in the hands of whoever developed that software."

      See, that's

  • (We're providing a series of Web services, assets, and Web applications to users of our mobil

    This guy has the PHB manual talking points inserted rectally so hard it's still coming out of his mouth. I can't even take people seriously when they talk like this. Probably why I live in my Mom's basement instead of having a corporate job and a girlfriend.

  • Funnily enough this is a project I am working on right now.

    I'm coming at it from an HPC (high performance computing) perspective. We'll have a cluster in-house supporting the base load and overflow to a utility computing provider.

    Job scheduling software (currently torque [clusterresources.com] but also trialing slurm [llnl.gov]) is used and once the total load has passed a threshold more remote compute VMs are fired up.

    We should have it in production by - /me checks gantt chart - last month.

    It seems like an idea whose time has come.

  • Did anyone else read the title "Sending Execs To The Cloud"?

  • I've been wondering something similar, but what Im looking for is something that is the equivalent of google docs, calendar, ect that I can host on my own webserver. (dreamhost is pretty awesome) For my work at a newer IT Support company and for my own school and personal files, and since Im always running all over town and use my iron-key, a easy to install cloud computing setup would make my life a lot simpler, and if there is no such thing, there is definitely a market for it. Any ideas anyone?
  • My company is a development partner with http://mor.ph/ [mor.ph] and we're doing exactly this. A website we're working on has high load twice a year, and other than that it can be handled by a single web server. What we're doing is augmenting our year-long server with a few extra app servers running at mor.ph during the two peak seasons.

    The problems you worry about in this case are the same as any distributed HA web serving: having another site handle master-server failover gracefully, etc.

    -Josh Adams
    http://isotop [isotope11.com]

  • They want their timesharing back.

    As long as these "cloud" services remain incompatible there is no "cloud": just a bunch of competing timesharing services. There will be no "cloud" until you can switch your "excess load" from one provider to another with complete transparency.

    I'm not holding my breath.

  • I'm a little surprised this is even a question.

    The cloud computing does not solve every problem going. Nor does map/reduce or XML.

    As a side affair I run a DTP Software as a Service (Service). This makes extensive use of 'Cloud' stuff.

    Use neutral technologies that work on many cloud computing platforms. Postgres, Java, and Linux Image. As a quick startup, run VitualBox on each of the Cloud infrastructures, and then only code for your VirtualBox instance.

    You have to write all the bridging, bootstr
  • Disclaimer: I work as a software engineer at Salesforce.com.

    Depending on your needs and budget, Force.com might be a viable option. It's all proprietary, but they do provide enterprise-class service level agreements and have delivered enterprise-grade levels of uptime to companies big and small ( http://trust.salesforce.com/trust/status/ [salesforce.com] ). Its Apex programming language is Java-like ( http://www.salesforce.com/platform/application-development/apex-programming-language/ [salesforce.com] ), and it's possible to write arbitrar

  • I just wanted to put in my 2 cents that it's cheaper and easier to scale using content deliver networks as a web proxy if you can get a cache hit rate that justifies it.

    The efficient model would look like:

    browser <- CDN <- DNS <- cloud

    Storage and bandwidth of a CDN web proxy will be about 1/2 that of a cloud service.

    Sites like plentyoffish.com get away with just 4 servers by using a CDN. With only four machines then the choice of "cloud" vs. "real machine" becomes a "don't care".

  • by bokmann (323771)

    Yes, my company has done this for a client and we are about to do it for one more. We have 3 servers hosted at 'traditional' hosing providers (one at Rimu Hosting, the other at ServerBeach), and during peak load, we fire up Amazon EC2 instances and throw traffic towards them. The 'real hardware' servers can handle normal traffic and redundancy (any one can fail), we just use the cloud for handling peak traffic. We also use the cloud for creating a multiple machine staging environment for testing. We can

    • by ooglek (98453)

      You use EC2 because you have to pay Rimu $30-40/month, but with EC2 you can launch and stop a server on demand.

      Plus I hate managing hardware, or relying on data center monkeys. I hate shared servers. Virtual servers are OK, but you never know what you'll get when you need it. Or when they'll go down. Granted, I'm looking for an EC2 backup -- can I take an AMI and turn it into a xen image which I can upload to my own servers running xen or a hosting provider running xen.

  • EC2 is great. It allows fast saving, recovery and on-demand provisioning as needed. The cost is on par or maybe a bit more than most service providers, but I like the pay-as-you-need-it self service -- no server monkeys, hardware failures, etc.

    Even though EC2 is great, and has an SLA claiming 99.9% uptime or you don't pay, I need a disaster plan. I'm still looking for an EC2 AMI to xen image process, so I can take my existing AMIs and move them to my own xen server or a hosted xen server if Amazon EC2 or

  • Try reading this blog post about "cloudbursting" [typepad.com] and the pages it links to. It talks about using your fixed infrastructure, but also expanding to the cloud when you need a sudden burst of power/infrastructure.

  • A cloud is usually the first sign I have excess load.

    Oh wait, this isn't the methane gas thread!

The bomb will never go off. I speak as an expert in explosives. -- Admiral William Leahy, U.S. Atomic Bomb Project

Working...