HN user

cwsteinbach

234 karma
Posts3
Comments13
View on HN
Starlite 13 years ago

In 1996, the US government decided that large numbers of its nuclear weapons would require replacement, refurbishing, or decommissioning. Accordingly, the Department of Energy set up a refurbishment program aimed at extending the service lives of older nuclear weapons. In 2000, the National Nuclear Security Administration (NNSA) specified a life-extension program for W76 warheads that would enable them to remain in service until at least 2040.[2]

It was soon realized that the FOGBANK material was a potential source of problems for the program, as few records of its manufacturing process had been retained when it was originally manufactured in the 1980s, and nearly all staff members who had expertise in its production had either retired or left the agency. The NNSA briefly investigated sourcing a substitute for FOGBANK, but eventually decided that since FOGBANK had been produced previously, they would be able to repeat it.[2] Additionally, "Los Alamos computer simulations at that time were not sophisticated enough to determine conclusively that an alternate material would function as effectively as Fogbank," according to a Los Alamos publication.[3]

Several years ago I had the idea of using Markov chain algorithms to auto-generate Thomas Friedman articles. I was going to build a site around it called mechanicalfriedman.com. I just discovered someone else beat me to it:

http://thomasfriedmanopedgenerator.com/

There goes my billion dollar exit...

Author here. Believe it or not I originally had the compression ratio graph rotated 90 degrees, and had manually modified it to run from 0.00 to 1.00. Google docs for some god awful reason insists on starting at 0.2 by default. Anyway, when my colleagues reviewed a draft of this post they requested that I rotate the graph back, and in the process I forgot to reset the scale. Sorry for the confusion. It's fixed now. As for the definition of "compression ratio", I looked this up and went with the definition found here: http://en.wikipedia.org/wiki/Data_compression_ratio

I agree that it's kind of counterintuitive.

> as ZFS is likely caching the reads in the ARC

Each of the seven queries we used in our benchmark required a sequential scan of the 32GB dataset. It's unlikely that the ARC had any impact on the results since the EC2 instance had only 7GiB of memory.

I wasn't aware that Reiser4 supported compression. Thanks for pointing that out. As for why we chose to use ZFS instead of Btrfs, we feel that ZFS is closer to being in a state where an enterprise customer would be comfortable deploying it in production. This is due to the fact that ZFS has been in development for over a decade with many Solaris sites already using it in production, and Btrfs is still marked as "unstable".

While I can't claim that we logged CPU load while running these tests, I can say that I watched the output of top and iotop and that the CPU load was relatively light. It's also worth pointing out that Amazon describes the I/O performance of c1.xlarge instances as "high". We also considered using an hs1.8xlarge "High Storage" instance for these tests, but eventually decided that we were more interested in testing against conventional disks as opposed to SSDs.

old-gregg's point is valid. Since we didn't benchmark ZFS without compression we can't say for sure how much of the performance improvement is attributable to compression vs. just ZFS.

As far as AWS goes, we have noticed ephemeral disks connected to the same instance can exhibit fairly large performance differences, and attempted to control for that in our tests by reusing the same disk for each test run.

Disclaimer: I'm a committer on the Apache Hive project.

A couple points in no particular order:

* EMR Hive is a closed source fork of the upstream Apache Hive code base. The EMR docs imply that the latest version of EMR Hive is based on Apache Hive 0.8.1 (which was released more than a year ago), which means EMR users aren't benefitting from the performance improvements that appeared in the 0.9 and 0.10 releases.

* It is implied (though not explicitly stated) that the Hive queries were run against gzip compressed TSV files stored in S3, while Redshift was allowed to spend 17 hours converting the same data to its own optimized internal format. Hive supports an optimized columnar format too (RCFile). Why wasn't that used in this performance comparison?