Kickstarter - WORLDS SMALLEST DATACENTER
hello fellow icrontians, long time member here.
I would like to introduce to you, the worlds smallest datacenter “Rockhopper” Based on ARM architecture, a Rockhopper can contain up to 8 nodes with a total of 64 cores @ 2.0Ghz which interconnect to our proprietary high-speed and hot pluggable. The nodes also come equipped with a DDR3 sodimm slot that can support up to 8Gb, for a total of 64Gb. Energy efficient, scalable and high performance platform available to be used anywhere in their homes, offices or an IT environment. Encouraging users to be thinkers and experiment with new possibilities.
We are excited about launching our kickstarter campaign. Visit our kickstarter page to get more details.
hmm... I might be interested in this for my MC server.
This is super awesome. I just came back from a conference where Hadoop was uttered practically every other word. I want this so bad.
Given that it's a Kickstarter, I probably can't get work to buy me one until it's already in production, but I'm considering getting one myself anyway. Maybe with just 2 or 4 nodes so I can made the case for a full set later.
pssst.. wanna help out with the hadoop install this week? we might be able to pull some strings and give you shell access.
But how does it do with f@h?
That sounds like fun, but I should explain that I've never actually gone hands-on with Hadoop or installed it before. I'm just looking forward to getting a chance to do so after having some of my skepticism towards it cured last week.
No problem Gargoyle, once its up i'll let you know and post some vids.
i am doing a massive clean up on the cluster now
Everything about this thread is awesome to me. @citrixmeta if you ever need a second extra set of hands and eyes from a relative n00b (after @Gargoyle of course), please let me know. My enterprise is going live with a Hadoop instance soon and the more I can learn the better.
//edit: is the design intended for redundancy to be introduced through additional chassis rather than additional nodes within a chassis? I see there's a single PSU per chassis in your design. So, the multiple nodes within a chassis are for parallel processing or multiple tasks, while HA would be accomplished through duplicate chassis?
here is a dry run of hadoop!
hey Ghoosdum! hope all is well with you.
it comes down to the nodes being connected via a backplane 1Gbe, you decide what to do with the nodes. the switch chip is picked for high throughput and low latency.