sargun

71 ms·

751.
by sargun 12y ago
So, looking at what you guys are doing. At least based on your design from four years ago, have you looked at what the open computer project is doing? At 4U, you guys can fit 45 drives in your case. The equivalent OpenCompute Knox would be
752.
by sargun 12y ago
I love QuickCheck. It might be a little hard to do for Tetris.
753.
by sargun 12y ago
As someone who thoroughly enjoys ice cream, and distributed systems, this post was really enjoyable to me. Is there a name for this kind of prose that intermingles technical facts, math, and narrative? I find that it does resonate with a sp
754.
by sargun 12y ago
Yeah, there was an excellent ACM article with him, and Bailis. I think if the network starts to partition, or fail in a datacenter, that's time to evacuate the datacenter / AZ. If a handful of machines fail, they should disengage
755.
by sargun 12y ago
This site actually has more information: http://www.micronautomata.com/ It sounds a like like the Parallela boards, as in it's a 2-dimensional grid network and it sounds like they broadcast to the edge of the grid. My
756.
by sargun 12y ago
The total toolkit they have is called the Simian Army -- https://github.com/Netflix/SimianArmy They wrote a blog post about it here: http://techblog.netflix.com/2011/07/netflix-simian-army.htm
757.
by sargun 12y ago
I think we touched on that in "Automated Deployment" -- I don't actually think that continuous delivery is core to the PostOps manifesto, but enabling it is part of the core manifesto. It's all about lowering the opportu
758.
by sargun 12y ago
So, I think that the article didn't strongly address what DevOps actually means. DevOps has come to mean a position on a team as opposed to an ideology like it is at the Google's of the world. My team recently wrote a blog post ab
759.
by sargun 12y ago
I'm curious as to the author's choice of Redis as opposed to RabbitMQ, or another queue. Redis is problematic when it fails, which makes me question the use of a centralized queue server altogether.
760.
by sargun 12y ago
FPGAs have finally become used in some compute clouds. Specifically, Microsoft published some work in this space called Catapult: ftp://ftp.cs.utexas.edu/pub/dburger/papers/ISCA14-Catapult.pdf I predict that we
761.
Clock-SI: Achieving Snapshot Isolation in distributed databases [pdf](research.microsoft.com)
1 points by sargun 12y ago|0 comments
762.
by sargun 12y ago
Why would you fsync logs for a high-level service? Are you afraid a power outage is going to cause you to lose service logs?
763.
by sargun 12y ago
A little blurb about this from Bruce Schneier: https://www.schneier.com/essays/archives/1998/04/the_secret_...
764.
by sargun 12y ago
Chip and pin cards use public key cryptography. From my understanding, you would have to provide the PIN code to the device to first unlock it, and then you could verify the authenticity of the card by reading a public key, which identifies
765.
by sargun 12y ago
I was recently reading only a slightly related article about Karaoke. Apparently in other countries, there has been a rage of killings due to certain songs being Karaoked. See the Wikipedia article here: http://en.wikipedia.org&#
766.
by sargun 12y ago
What's the performance overheads of SystemTap? I've looked at it, and used it for poking at things under a microscope on a VM, but I've never used it on a production system. Could I instrument production code with it?
767.
by sargun 12y ago
The provider has a specific set expertise that's probably better aligned with hosting this service. Since it's a revenue center for them, versus a cost center, they're better equipped to make the case to hire specialists. The
768.
Sutro Tower: From Eyesore to Icon(youtube.com)
3 points by sargun 12y ago|0 comments
769.
by sargun 12y ago
It's amazing to see what Ballmer stepping down from Microsoft has done. Microsoft's stock has outperformed Google's in the past year: https://www.google.com/finance?chdnp=1&chfdeh=0&chdet=140847... , m
770.
by sargun 12y ago
But, the data storage can be in a AP system, allowing better stories around data availability.
771.
by sargun 12y ago
I really like the fact that facts are immutable in the Datomic model. It actually makes it easier to reason about eventually consistent, because you can rely on causal consistency between facts. Eventually consistent systems can give you du
772.
by sargun 12y ago
This is super exciting. Highly-reliable, off-grid power is amazing. Today, datacenters have to buy massive diesel generators, and ensure they have diesel supplies in case of grid power failure. This infrastructure is surprisingly difficult
773.
by sargun 12y ago
Excellent! I think that's the right way to go to be honest. Become an arbitrage service, and then use the monthly service fee to "hide" the irregularity in "cost" to the market. Once you have an overall picture of m
774.
by sargun 12y ago
I briefly did some math around the capacity curve for said services. I'm not entirely sure how they're able to make it work effectively, and scale revenue superlinearly to cost. Also, the demand curves are less than awesome, given
775.
by sargun 12y ago
I would be surprised if India, China, or Russia did not have similar sorts of surveillance programs. But, I think you're right, you're unlikely to be caught in dragnet-style surveillance by the Five Eyes nations, but this style of
776.
This World of Ours [pdf](research.microsoft.com)
4 points by sargun 12y ago|0 comments
777.
by sargun 12y ago
At least on most chips today, they use modulo-N hashing, which results in potentially breaking existing connections, see: http://tools.ietf.org/html/rfc2992 The reason that this isn't really a problem on the inter
778.
by sargun 12y ago
Yeah, this is how Facebook does it, and how we did it at Microsoft / Yammer.
779.
by sargun 12y ago
So, it looks like they're using BGP to "program" the adjacent router a la "SDN." I'm curious as to two things: * Byzantine fault-tolerance: How does this system handle failures when a specific node fails in a w
780.
by sargun 12y ago
So, would you propose a dedicated physical wire for every service? You have to begin weighing the differences between physical vs. logical separation in network design at a point. MSR did two really good studies on network reliability, when

More ›