ConsultoBabble Deployment Analysis Report for Cloud Deployment of EtherealMind.com

I don’t use a Cloud for any of my blogs or email services. I’ve looked at three different cloud providers including Amazon, Rackspace and others. I guessed that they all would work, more or less.

Well, after I’ve spent a week or so installing Linux, NGINX, Varnish/Memcache, PHP, WordPress etc. Then spend time hardening the Linux, NGINX, WordPress etc. And configuring the email server. And the DNS server. And patch/update it every week or two.And after all that, my hosting costs spiralled out of control almost immediately. It was a complete failure from start to finish.

I’ve decided to summarise my Cloud Computing experience by preparing a ConsultoBabble ™ report for your enjoyment. I warn you, it’s not happy reading.

 

“ConsultoBabble™” exit statement

This report summarises the exit statement from the closure of the Cloud Computing Project for EtherealMind.com. We agree that this project did not meet the high expectations of our customers and that we failed to meet the necessary service levels.

Implementation of EtherealMind.com on the Cloud

Initial implementation was estimated at twenty hours. However, unexpected software implementation challenges resulted in approx two hundred hours investment. We could have improved this time with better Linux skills however the correct skill set was not available at project commencement.

We agree that our convention for naming this process “DevOps” was misleading.

The initial deployment was completed on a cloud server despite the time pressures. However, the Cloud Provider billing systems was specified in the initial design, and ongoing review of charges was specified. The planned monthly cost of the website was exceeded in the first week with no recourse for refund or reparation from the Cloud Provider.

A deep analysis of the system logs from NGINX was initiated. Our process shows that the web sites are being scraped on the RSS feeds from hundreds of separate sources including “market sentiment analysis” companies, chinese scrapers, internal corporate intranets, vendors feeds and much more that could not be identified.

And none of these “hits” resulted in “page views” for success measurement or impressions. As such, they are classed as failure criteria for service delivery benchmarks specified in the initial service outline.

There are no mechanisms to control who can “hit” the websites, or restrict access. Inbound filtering of client traffic for an open website isn’t practical. We accept that each HTTP request was costing money in terms of CPU/Memory consumption, Storage (logs) and CDN/Caching volumes. each resource request consumes compound resources and compound charges are levied by the Cloud Provider.

Conclusions

Customer Impact Statemet : Costs are increasing without any control mechanism or business reward. Further, no control over the demand side of the business opportunity to restrict costs within a set budget.

Cost Containment: After performing an analysis of possible rectification processes to limit the performance of the systems it was clear that these options also required a significant investment to locate, learn and implement software that would deliver cost management.

Recommendation: This project is agreed closed. It’s our recommendation that use of Cloud solutions is impractical unless you have control over the demand. That is, unless you can control the resource consumption by reaching the demand for services, you may be committing your budget to an unlimited amount of funding. Further, there are many unexpected ways for Cloud to incur costs.

Closing

We apologise for making hyperbolic promises about the suitability of deploying a medium sized web site to a cloud hosting system. We recognise that this solution would never have meet the requirements if we had understood a how a “Cloud Computer” worked.

We understand that you now have a successful deployment using a managed service provider whom is specialised in providing specific solutions on the dedicated systems.

  • Sean Walberg

    TL;DR – You hired a network guy to do a sysadmin’s job. Therefore cloud sucks.

    • http://etherealmind.com Etherealmind

      There is an aspect of that. What I found is that the time spent learning to use Amazon or Rackspace, plus the overhead of mucking around with Linux took far to much time.

      It didn’t replace the work I was doing using a hosting package, or offer any benefits that weren’t offset by increased costs (a LOT of extra cost). It was easier to move to a managed service than use the cloud.

  • http://twitter.com/ioshints ioshints

    Well, you’re not using an __infrastructure__ cloud for your web presence. It could be argued you use PaaS for your web server and SaaS for your email ;))

  • Francesco Bonanno

    Mah… costs aside, what are the advantages of a web site in the Cloud respect to the normal web Hosting?

    • http://etherealmind.com Etherealmind

      The web hosting I was using needed to be increased for the CPU/Memory to improve performance. I figured that some “Cloud” experience would be a “good thing™” and decided to migrate my WordPress to a cloud platform.

      However, I discovered that using Amazon or Rackspace would cost about four times as much as a standard hosting package for http://etherealmind.com and http://packetpushers.net. So I’ve moved to a wordpress hosting package that handles all the Linux, security and caching etc.

      The theoretical advantage for using, say, Amazon is that I could always scale up my resources by adding more CPU/Memory or services like CDN etc. In practice, it didn’t work that way. The administrative overhead of the services took too much time and the interfaces to support the hardware took a lot of time to learn (again, using Amazon or Rackspace interfaces for system management are a whole new skill set to learn and understand).

      Twiddling with Linux isn’t a valuable skill for me so the time spent doing that is wasted time. Similarly, learning how to use Rackspace or Amazon is also time wasted.

      YMMV.

      • Francesco Bonanno

        Uhm… I think the Cloud solutions are for “I don’t use this very much, so I don’t need to deploy an infrastructure for this, so… Cloud, so I can save so much money…” but in the case “I use this very much and I want to Upgrade” Cloud could not be convenient, in effect.
        In this case, perhaps, in more convenient to ship the next level service from the Hosting Provider.
        Thnx.

  • http://twitter.com/BRCDbreams Brook Reams

    Greg,

    A good carpenter has hammers, saws, chisels, carpenter’s squares, planes, etc.  She knows which ones to use depends on the job to be done.  She also keeps a list of sub-contractors to call, and this is her most important tool.  Knowing when to let someone else do the job can greatly improve profit. ;-)

    Onward …

  • Wenjing Chu

    I am shocked to hear you would even think going down this route. What was the purpose? You are already buying shoes from Macy’s, and presumably not-unhappy, why would you consider employing a scalable manufacturing plant in China for just a pair of shoes? IaaS clouds would make sense if you were contemplating to start a new shoe-designer business to compete with Nike. There is another layer of businesses that do tailor towards end consumers, ie SaaS.

  • http://rbenigno.pip.verisignlabs.com/ Ryan B

    It seems like this could be summarized in another way.  The cloud deployment failed, so instead we deployed to the cloud.

  • Pingback: The Cloud Isn’t Cheap – Part four or five I think — My EtherealMind

  • Dogenfrost

    How is a managed service (ie SaaS) not a cloud service?