bredman

IMO it's not that bad, people are used to paying locksmiths that much when locked out of houses or vehicles.

The one constant that seems to come up in all these founder stories is that you better love what you're doing. There's no guarantee that your new business will be successful or profitable but it seems inevitable you will sink nearly all your time, energy, and money into it. You better be feel rewarded by the journey because that's all you can count on.

It's not as easy as it could be but you can accomplish aggregate pushdown by using executor hooks in your FDW. As an example I'd point you to our (Citus Data) recent work on doing this with our columnar store foreign data wrapper:

https://github.com/citusdata/postgres_vectorization_test

As discussed there it definitely showed promise in improving performance for us.

Thanks, I haven't looked at Udacity much so I'll give that look. I like using Python as it's a bit practical.

To clarify her goal isn't to land a job, just to learn a new skill and be able to relate to her coworkers that are doing more coding better.

Thanks! Reminding myself of her feedback I think that pacing might be a significant issue which is something I hadn't considered too much (oddly).

She's someone that has lots of computer experience but the closest she's gotten to programming before is simple Excel functions (think SUM). I think she'll be much more excited by one where she's working on "real world" stuff so I'm leaving a bit towards Creative Programming for Digital Media & Mobile Apps and hope that it spurs her to learn even more.

"So the government can have a monopoly in a space" - always a good reason to ban a successful private enterprise.

Interesting but the devil is in the details with systems like this and the article doesn't go into much of that.

I do like the decision to separate AtlasDB from the underlying K/V store.

The business implications of this are interesting. In my opinion this is another warning shot that if you don't open source your business someone else may just do it for you. If I was LiveFyre or Disqus I'd definitely be nervous about Mozilla coming into my space and building competing software with a commitment from two of the big players to use it. Leveraging open source by adding value via services or premium offerings has been a very powerful business model in the last few years.

Interesting, I've have had the exact opposite experience for #2. If there's a movie, or TV show that I want to watch I'd rather pay $5 to watch it than just be told it's not available. This has had such a strong effect on me that if there's something in particular I want to watch I go to Amazon before Netflix simply because I know I will more likely have an option to watch what I want (even if it costs money). Personally I feel like Netflix has back themselves into a corner by only having one all-inclusive price level.

That said the part about Amazon that infuriates me is when they have the movies available but not for rent and only offer the pay $15 to own option. Also the Netflix discovery/browse is experience is light years better than Amazon's.

The benchmarks use cstore_fdw which is a columnar store that is accessed via the PostgreSQL foreign data wrapper system.

ICMS is a different implementation of a columnar store for PostgreSQL. I don't know it extremely well but my understanding is that it uses the PostgreSQL shared memory system to allocate memory for a columnar store. This columnar store is not queried via standard SQL commands but rather using a combination of a custom query language and some user defined functions.

If you use cstore_fdw with PostgreSQL you can use the full PostgreSQL SQL dialect for querying your table. However, we don't currently support writing to your cstore_fdw table using SQL commands like INSERT/UPDATE/DELETE. PostgreSQL added foreign data wrapper support for these commands with their 9.3 release and it is something we're considering adding to cstore_fdw in the future.

Actually CitusDB, which we also produce, does distribute data and queries across a PostgreSQL cluster. You can see some performance comparisons of using cstore_fdw on CitusDB in the blog post and the performance improvements are even more significant there.

Ben from CitusData here.

At a high level columnar stores are good for analytical workloads and not great for OLTP workloads. Some reasons for this:

* cstore_fdw currently only supports modifying data via the COPY command. This means that UPDATE and DELETE operations aren't yet possible which limits the utility for OLTP workloads.

* The skiplist indexing system works well for analytical workloads where you are aggregating or retrieving a significant number of rows but wouldn't be great for retrieving single rows.

To clarify, you can already pay to watch past seasons of Game of Thrones on Amazon Video on Demand in the US. It's just not included for free with Prime which as I understand is the change that's happening for a subset of HBO content.

For those curious here's a similar post from ~4 years ago:

https://news.ycombinator.com/item?id=1411937

In the article there's a screenshot that strongly implies the ads will be coming via email ("this consent supersedes any prior email unsubscribe notification I may have with AT&T")

I love the "I LOVE ENGAGING WITH BRANDS" text in the screenshot.

This was released on March 28th:

http://googlecloudplatform.blogspot.com/2014/03/announcing-g...

Guess this is just on the front page because of XKCD about the other Google DNS product?

I feel like they have to stop at some point but I agree this would make much more sense than something like Oculus Rift. I can imagine lots of great ways to integrate this into Facebook.

Am I the only one that's not that impressed by 2 hours of downtime a year? I've worked at other companies where I was responsible for running a service and 2 hours of downtime a year was considered a failure.

I think this means not all outages tracked affected all users equally. Some users would have seen more or less downtime than the 2 hours mentioned.

Could someone explain DJB's point to me:

  Cryptographers are certainly not responsible for this superstitious nonsense.
  Think about this for a moment: whoever wrote the /dev/random manual page seems to 
  simultaneously believe that
   
  (1) we can't figure out how to deterministically expand one 256-bit /dev/random
  output into an endless stream of unpredictable keys (this is what we need from 
  urandom), but
    
  (2) we _can_ figure out how to use a single key to safely encrypt many messages
  (this is what we need from SSL, PGP, etc.).
 
  For a cryptographer this doesn't even pass the laugh test.

Is the argument here simply that a properly encrypted message should look like randomness? And as a result we should be able to turn something non-random into randomness?

IMO this is the most widely trusted standard benchmark: http://www.tpc.org/information/benchmarks.asp. However not everyone publishes benchmarks and there are a number variations for different workloads to consider.

If you are replacing an existing system I would recommend narrowing the field of candidates and trying a few options with your existing workload.

I'm not a CloudFlare expert but my guess is that there are some risks to the "bleeding edge" optimizations. The decision probably comes down to a trade-off of how much you want to minimize risk and support older browsers in exchange for speed.

I think he did this to show how even with a very restricted dictionary you could still build a relatively secure login system. Having a restricted dictionary could also help with the issues other people are discussing with humans being socially programmed to select certain word orders as their password under this scheme.

It came out in theaters first, they acquired the rights to it after it had been released there. The best I can find about availability on Netflix is "early this year".

I see your point; ALIAS records are not a CNAME substitute and should not be treated as one. However, they do have their uses and the risk is minimal if you understand what ALIAS records are doing and how to use them.

I'm having a bit of a hard time following your example but I believe if the authoritative DNS provider simply inherited the TTL of the alias target then the behavior would be as desired. This gets back to donavanm's point that this is more an issue with the way ALIAS records are implemented then a problem with the concept of ALIAS records.

One other issue that the author doesn't touch on is that we now have 3 implementations of ALIAS records that I'm aware of (Route 53, github, and Heroku) and there are differences about how they behave. However we have those 3 providers using the same ALIAS to describe similar but significantly different things. This is clearly confusing and potentially disastrous for users.

I think the article short changes the value of startups that can be described as a specialization of another. There's lots of differentiation that can happen in a single space across verticals. Think about what features are important to people who use Word vs. Latex or Blue Nile vs Amazon.

The other thing I would add is I got some advice (which I now agree with) from a college professor that business should focus on being really innovative/risky in one dimension, not in all. This seems like a good way constrain the areas your business will really be innovating in.

> When doctors or nurses use their knowledge of anatomy in order to torture or conduct medical experiments on helpless subjects, we are rightly outraged. Why doesn't society seem to apply the same standards to engineers?

I think the difference we draw between the two is that an engineer builds things that can be used for many purposes (defense or offense for example) and we as a society generally agree with one of those two uses. When a doctor abuses their knowledge to cause pain they are the ones making a choice we disagree with.

Anyways, yes engineers should think about how their work can (and will) be used but I think it's disingenuous to compare them to patient torturing doctors.

In the context of full web pages how I've seen this done is by embedding a Javascript call that occurs after the page is loaded to your site. I suspect you could do something similar with this resource.

Alternatively you can look at what CDNs are hosting the resources and then just use the existing tools to compare the expected performance of each CDN. Of course this comes with a few caveats:

1/ Performance may be different on a given CDN depending on which "package" the customer has purchased. I know was the case ~1-2 years ago for Akamai.

2/ A CDN may perform well for small objects but not for larger objects, make sure you look at representative benchmarks.

3/ JS CDNs using CDN load balancing services like Cedexis mean more work to find all the CDNs that are being load balanced across (jsDelivr uses Cedexis)

(edit: formatting)