Comments on: Objects in Space. Trends and Information on Big Data, New Data Management Technologies, Data Science and Innovation. Tue, 20 Dec 2016 11:47:00 +0000 hourly 1 By: Manash Chaudhuri Thu, 24 Feb 2011 09:25:42 +0000 Thank you for sending this interesting articles. It’s amazing how far
we have come along in terms of handling database to the tune of 10^12!

Thank you again for sharing such good information and making the
internet much more valuable than a mere social network…:)

By: Nik Tue, 22 Feb 2011 23:49:00 +0000 Very interesting article! It would have been great to learn more about other parts of your infrastructure although I understand that Roberto’s main focus is on databases 😉

For example
– how do you ensure non-blocking ingress?
– how do you deal with the load of multiple tenants running queries in parallel on the vast amounts of data?
– how do you deal with application prioritization? Do you over-provision to account for peaks?
– what tools do you use to track performance / application health?

There are probably a lot of parallels between your implementation and a more normal SaaS setup but there are definitely areas that are unique to your use case that would be very interesting to learn from. Where can we learn more about how you have set up your infrastructure?

Leveraging the cloud makes sense (you mention 100s of AWS EC2 instances) to ensure elasticity but I hope you also look at solutions that enable you to do more with your existing infrastructure by increasing your application density – particularly as you mentioned your goal to reduce energy consumption. Linux Containers or Librato Silverline would be a more elegant approach than throwing VMs at the problem.

I hope we get an update on the project further down the road!

By: Kevin Rice Tue, 22 Feb 2011 18:58:02 +0000 Interesting article and use of Java. I wonder if the project thinks it will need 64 bit implementations of the JVM to make make it work?

By: Simone Brunozzi Tue, 22 Feb 2011 09:55:41 +0000 This is a great article!
I think that some interesting options for the Gaia project and Amazon Web Services would be:
1) Reserved EC2 Instances
2) Spot EC2 Instances
There can be significant cost savings if you can use the right combination of Reserved, Spot, or some on-Demand EC2 instances.


By: Alessandra Bagnato Tue, 22 Feb 2011 09:45:58 +0000 really very nice and interesting, thanks :)