Healthcare and Cloud Computing
Previous Meetings
Follow us
Slacloud Tweets

 

Tuesday
Jun152010

We're Merging...

Please sign up for our new group-  not for the cloud faint of heart....
http://www.meetup.com/Hub-Cloud-Club/

 

Monday
Jan252010

Upcoming events

Ed Tech '10-  May 6th from 5-8pm

Register:  http://edtech10.eventbrite.com/

 

Patient-Centered IT- May 20th 9am-12 noon

Register:  http://pcit2010.eventbrite.com/

Friday
Dec112009

CloudCamp Boston December 10th, 2009

Wednesday
Nov252009

Big Data - Video in the Cloud

Back on November 10th we had a new company in the works called VideointheCloud present at our monthly meeting. When the whole topic had come up I was thinking - big deal, video stored in the cloud and downloaded on-demand and why was this something a couple of smart guys would go off and do?

The speakers Branko Gerovac and David Carver were a team who have been involved in video/media computing for a combined 45+ years making them experts in the video/media business. They also have several successful ventures under their belt including starting up and launching Blackwave a video storage and delivery system manufacturer.

The presentation was different than most because they went through the process of explaining the scale of what is going on in the video-on-demand business, the market structure (YouTube vs. cable video-on-demand), and the problem statement as it relates to transcoding and delivery the data.

They posted their own blog post with their slides which can be found here but I'd like to share some quick highlights:

- YouTube represents 314M minutes/month of video viewed by 126M people in the US

- 25K/hours of YouTubecontent uploaded every month

- Total video 2.9 TRILLION minutes/month!

Now if you assume that the use of google like low power CPU's (they use 1.2ghz) are crammed into a 20' container and we assume that for 24M titles with 2.5minutes a title - we get about 40B minutes a month. If upload is 3M minutes a month - the cost is over $5B to transcode the data on upload where transcoding "only" costs $2.8B on download.

If you take a different approach and use bigger processors and stick with the download process you can actually get the costs down to the $500K range - making this much more viable as a business.

According to Branko and David the three key issues are transcoding horsepower, bandwidth, and storage making this a great application for a purpose built cloud for Video.

I highly recommend looking at their website and take a look at some of the data they are capturing live like the last 7 trailing days of YouTube video content. Big data problems are going to be great beneficiaries of the cloud.

/wayne

 

 

 

Monday
Nov232009

The Planet - Colo vs Hosting vs Cloud

A few weeks ago we had a couple of great speakers - one of them was Rob Walters - Director of Product Management from The Planet based in Houston Texas. The Planet has over 20,000 customers, manage over 48,500 servers running in 6 SAS70 Type II data centers with over 140+ certified engineers.

  What makes The Planet unique is that they offer colo, hosting and cloud services and they cater to a specific segment - mostly SMB sized customers. Most people that I ask about segmentation in the cloud don't believe that it matters - I think it does - and not just to vendors. The spectrum of convenience outweighs risk to the consumer for free services to the issues of risk to the enterprise and its core property - it's information.

Colo has been around for quite a while with companies providing basic power, pipe, and ping services during the dot.com era. A colo customer usually owns their own system and it is kept in a cage inside the colo provider. The benefit is the colo site usually has heavy redundancy built into the network connections, the AC, and the power is backed up by generator with full time staff monitoring the facilities.

Managed services usually means that the provider is helping manage the environment and in the case of The Planet can be a virtualized environment. Managed services may be as simple as having the provider do the backups/restores to caring/feeding of the OS, user accounts, etc and the customer is only responsible for the portions of the stack that they want to own themselves. At The Planet - they own the systems changing the economic model to OPEX from a CAPEX model.

While I'm not going to debate the cloud definition - in The Planets case the environment is very portal oriented, easy to use, and delivers services rather than infrastructure per se with limited customization allowed and services metering method varies from the other services with scaling up/down based on demand.

Take a look at providers like The Planet, GoGrid, and Rackspace and compare them to Amazon, Google, or Terremark and Savvis. All three are much different in their offers with different levels of customization, SLA's, and the transparency level (to see more about transparency see my blog.

/wayne

The Planet Presentation