Archive | Cloud Computing

Thoughts on Chargeback – Part I

I recently finished listening to the lastest VMware Communities Roundtable podcast and it got me thinking more about chargeback models and how to determine the chargeback for a real live production environment. It seems like we are still shooting in the dark when it comes to figuring out how to charge the consumers of our private cloud services.

I would like to offer some ideas and suggestions for how to calculate the per VM cost for storage, compute and network resources. I am very interested in this concept as my company moves toward a chargeable service model for IT. I will break this up into 3 parts so as to (hopefully) get more comments and feedback, as well as to go into more detail about each area of chargeback.

Here goes…

Part I

Storage

Storage chargeback not only consists of storage capacity used but also metrics of the IOPS used over a given period of time for a particular VM. As such, we need to have a base rate that takes into account capacity in GB for a particular VM that is charged per month, without any deviance, since the storage capacity is used as long as the virtual disk exists on the datastore.

In addition, the amount of storage bandwidth should be charged based on a peak IOPS or average IOPS basis. This amount varies per billing period based on the actual metrics from the storage system.

The base rate (consistent rate for storage used) should consist of the following items: Total purchase price of the storage system + support contract + power/cooling + hosting costs + personnel costs / number of years system to be used (i.e. minimum lease period) = base rate per VM (for storage)

The rate factor (conditional based on usage) should consist of the following items: Tier of storage (inverse) x number of IOPS (peak or average) = rate factor per VM (for storage)

Again, I am looking for feedback on this. If I am missed something, please let me know. What metrics are you using to determine realistic base rates and rate factors for your environment?

*Next time, we talk about chargeback for Network utilization…

2,382 total views, no views today

0

On Cloud Security

After attending my first VMworld and having just met up with a practitioner of cloud security Christofer Hoff (rationalsurvivability.com) as well as having some time to reflect on the recent product announcements, I would like to say some things about cloud computing security. This is, and will be, the #1 concern for enterprise customers going forward with their public cloud deployments. This is something that VMware and others may think they have well in hand, but it appears lots of dominoes must fall into place before their vision can become a reality.

Here’s why…

Privacy

Companies spend a lot of time and money keeping their intellectual property and company mission private. To think that VMware will be able to overcome this by offering a certified vCloud provider certification to providers is naïve.  There are NDAs in place for a reason between certain businesses and their customers and by the very use of these cloud services some NDAs will become broken. Significant changes to laws, policies and business agreements will need to happen before any progress will be made in this regard; no easy feat. Even still, some companies will never go beyond the private cloud due to the fact that the need to protect their data far exceeds the flexibility and other benefits that public clouds provide.

Reliability

Reliability and SLAs are definitely a part of the security picture. For example, as a company reselling my services via a cloud services provider I am not only taking the compute, network and storage off-premise, but I am also in effect outsourcing responsibility for the uptime of the services. A wise CEO once made the following statement that I find somewhat relevant here; take with it what you may: “Never outsource your core competency or something that supports the core competency”. Those aren’t his exact words, but a professional, succinct way to illustrate his meaning. If you outsource something like 99.999% uptime for example, then you are giving a 3rd party a stake in the success or failure of the service you are providing and ultimately your company itself. Cloud service providers can and will offer you SLAs and refunds for downtime all day long, but in the end your customers will have a bad taste in their mouth and your reputation will be irreparably damaged. I suppose you give up some sense of security using any offsite service such as a co-location facility, but the effect is magnified in a public cloud scenario. If this sounds like an argument against outsourcing, well YES, it absolutely is. Unless you are the company providing the outsourcing, you might end up on the short end of the stick.

Job Security

Unless you are currently working for one of these public cloud providers, I suggest you learn as much about hybrid cloud integration or cloud architecture if you expect to have a long career in IT in the future. Reason being, people are expensive and if a company can buy services by the hour for which they have a definite SLA and performance metrics, why would they hire a person to do them. I have never heard of or seen a job posting with an SLA in the job description, but perhaps we will in the future. Skill up and join one of these cloud providers if you want a better chance at consistent employment.

1,888 total views, no views today

0

Powered by WordPress. Designed by WooThemes