This was a reply to bombcar.
183 karma · joined February 24, 2015
This was a reply to bombcar.
Interestingly, that's probably how most customers are handled by many of the submitted apps, if at all.
I have never encountered someone else that was able to hear whether a tv was on or not. At what distance could you do that?
So it is approached from the consuming end for better or worse.
What got me into Linux was a little later in 1997. I was playing Quake and I found out that my ping times were 60ms lower over a modem 33k6 connection than with Windows. That got me sold. And I just had a knack for the command line.
During my student time I quickly found out that tech paid my bills a lot easier than working in a cafe. It didn't attract many girls though so I kind of tried to combine it ;-)
Anyone comparing VNC to RDP favorably is nuts. RDP is a much more optimized protocol for high latency, low bandwidth. Only Citrix compares favorably from user's point of view. Maybe NoMachine NX as well. Both i cannot get to work over https (nomachine) and kubernetes (citrix).
Not requiring a custom, vague client also helps.
XRDP uses a Xorg driver or is able to use a VNC backend. Even then its faster than VNC. It can use pam for authentication.
Learning what you can't do is as important as to know what you can do. I'm happy that Alphabet is doing these long term research projects and bets. >5 years I think, apart from government, only oil & gas companies project that long term, but they know much better what they will 'find'. What Alphabet does it moves the academic needle cause we learn so much on the way.
Maybe they can do better in publishing the results, cause that is an alternative to just shareholder value: society value.
'A 2006 analysis concluded that “without reductions in smoking, there would have been virtually no reduction in overall cancer mortality in either men or women since the early 1990s.'
You argue that there was rapid progress in the last decade. The article agues the opposite (over decades even). It does this with citations and meta analysis where they find that much of the 'rapid progress' is not replicable. The article doesn't say that immunotherapy is not meaningful, but it does say it is over sold where there is hardly evidence of their benefit.
You also argue that the cost will come down, however there is no evidence for this. To the contrary even: according to the article it is going up significantly without any indication of improving.
Why do you state that the "bizarre claim" of not having improved treatment is "just wrong" without providing any supporting evidence for this?
Jupyter doesn't do scheduling and integrates pretty well with Airflow.
Prefect was started by an Airflow maintainer and friend of mine who also contributed the dask executor to Airflow. Hi @Jeremiah!
If you have idempotent tasks, which is a best practice, it is possible to use Airflow in HA even in active/active. It might occasionally schedule a task twice, which should be caught but in any case mitigated by idempotency.
If you are looking for more Enterprise support you can reach out to Astronomer (disclaimer: I'm a advisor to them) or use a Google cloud hosted version (Cloud Composer). Both are great products.
I wonder what your issue is/was? Notebooks are supported by means of a Papermill operator (equivalent to how Netflix operationalizes notebooks) or PythonOperator/BashOperator which would just wrap around your notebook.
However to parralelize tasks Airflow needs to know a bit more hence you might have found it required to break up your notebook into individual tasks that combine into a DAG. Is that what you meant?
The dataset you are talking about is not just in/out but is a lot more comprehensive than that.
Seriously.
Can anyone explain why this would be good?
So you would need additional defenses against tampering such as OpenID Connect. In the banking apps that I have been working with we implemented additional symmetric encryption on top of the protocol (yes obfuscating the keys) and all other kinds of small things.
I’m glad mastercard does not rely solely on TLS.
1. Collect only what is necessary for providing your service 2. Make clear what you store and for what reason 3. Ask consent and give the opportunity to retract this consent as easily
Deletion:
1. PII means information that makes a person identifiable. This is the type of information that you need to remove 2. So if you are storing PII information for the use of profiling you will need to disconnect the profile from the PII information. E.g. you could use user table where you would overwrite the PII information with generic information. You can still use the now stale profile withou PII information (for example in statistics, aggregations etc), but you cannot tie it to a single person anymore. Ie. You should not be able to reconnect the person to profile you have stored. 3. As technical possibilities evolve you need to improve the disconnection over time.
There are legitimate business reasons to store some PII information. E.g. for security reasons, other laws etc. So IP addresses don’t need to be deleted from your web logs, but if not given consent you cannot use them for ads, sell them etc.
The required clarity that GDPR will bring to your data is actually going to benefit you. Your data scientists will love it, because the tooling that helps with Gdpr also helps with discoverability, data quality etc.
Enjoy GdPR, there is a lot of business opportunity in it.
While your reasons are valid you are missing an important one:
Resource scarcity: the engineers that I need allocate to infrastructure I rather have working on user facing features and improvements. Talent is scarce, being able to out source infrastructure frees up valuable engineering time.
This is one of the main reasons, for example, that Spotify (I’m not working for them) is moving to google cloud.
Spark for Airflow is just one of the engines where a transformation of data can happen.
Shaping DAGs dynamically poses a challenge to the scheduler on how to 'predict' what tasks need to run in the future. The scheduler needs to evaluate which tasks will need to run, without actually executing these tasks themselves. For Airflow in its current state that is a chicken and egg problem.
For the future, I can think of allowing dynamic dags being described through the Rest API, but that is definitely further out and has not really popped up yet on the horizon.
* Scheduler that knows how to handle retries, skipped tasks, failing tasks
* Great UI
* Horizontal scaleable
* Great community
* Extensible; we could make it work in an enterprise context (kerberos, ldap etc)
* No XML
* Testable and debug-able workflows