Thursday, June 28, 2012

ARM based data center. Inspiring.

In a previous post I wrote ARM based servers. Since then, and thanks to all the comments and responses I got, I looked more into this ARM thing and it's absolutely fascinating...

Look at this beauty (taken from the site of Calxeda, the manufacturer):

What is it? A chip? A server? No, it's a cluster of 4 servers...

And this:

is HP Redstone Server, 288 chips, 1,152 cores (Calxeda quad-core SoC) in a 4U server “Dramatically reducing the cost and complexity of cabling and switching”. Calxeda is talking about: “Cut energy and space by 90%”, and “10x the performance at the same power, the same space” and it's just the beginning...

And this is from the last couple of days... From ISC'12 (International Supercomputing Conference): "ARM in Servers – Taming Big Data with Calxeda":

  • In the case of data intensive computing, re-balancing or ‘right-sizing’ the solution to eliminate bottlenecks can significantly improve overall efficiency
  • By combining a quad-core ARM® Cortex™-A series processor with topology agnostic integrated fabric interconnect (providing up to 50Gbits of bandwidth at latencies less than 200ns per hop), they can eliminate network bottlenecks and increase scalability

You still can't go to the store and buy a 4U ARM-based database server that performs 10x and uses 1/10 of the power (combine them, it order of magnitude of 100x...). It's not now, maybe not tomorrow, but it's not sci-fi. And technologies will have to adapt to this world of "multiple machines, shared nothing, commodity hardware". I think databases will be the hardest tech to adapt, the only way is to distribute the data wisely and then distribute the processing, sometimes parallelize processing and access to harness those thousands of cores.


  1. I'd love to see these combined with Shard-Query

  2. You've got that right Justin, I think tools like shard-queries and other more comprehensive software platforms such ScaleBase ( that give a complete solution for database distribution and scale-out over shard-nothing commodity hardware servers - will play a major role in this amazing new future we're facing!

  3. Full Disclosure: I'm former Intel, former server group. So my instinctive bias runs deep.

    I can't count the number of times I've seen similar "10x, 1/10th" claims made. Like this one, they're usually comparing a specific optimized design against some vaguely defined, unoptimized generic Intel server platform. I've personally been involved in multiple in-depth analyses of these claims, and I've yet to see a single one hold up to scrutiny.

    ARM's certainly got well-established credility in the cell phone and tablet arena. That doesn't make it a suitable processor for server workloads, where 64-bit addressability and SIMD throughput are increasingly important, especially for data-manipulation activities.

    Density-optimized, power-optimized, integrated fabric Intel-based systems exist today. They are competitive with ARM in terms of acquisition and operational cost, and they run circles around ARM when it comes to computational capacity and throughput.

    And then there's that teensy little issue of actual software availability for server workloads....:)

  4. Thank you Mitch for the great input. You actually very much familiar with my eagerness, and how happy I would be to deploy and test my "database grid" principle, on Intel-based fabric system, as well as on ARM-based, for me there's no right or wrong or precedence! :)

    Intel has done tremendously with density and power optimization even with with its 35W i5 and i7 and I'm anxious to experience the Surface tablet with Intel Inside(r)!

    Thanks for the interest!

  5. This comment has been removed by a blog administrator.

  6. Curious if this comes with its own software or if other DCIM software company technology can also used in order to evaluate and run the servers?

  7. This shift to cloud computing is really gaining speed, yet the sales of data cabinets and other data center essentials are increasing. It's an interest contradiction.