trelliscoded

I just did a greenfield deployment of Windows Server 2012 R2 for a big virtualization environment, and ReFS was supposed to be ideal for that kind of application. Everyone I talked to about it told me it was a bad idea, mostly because the supposed gains in certain virtualization operations wasn't worth risking being the first sucker to find a corner case in refs.sys. Also, no data recovery tools support it outside of experiments.

Does anyone actually use ReFS in production?

  Location: San Jose, CA
  Remote: Done it before, can do it again.
  Willing to relocate: For the right price.
  Technologies: {Windows, Linux, Cisco, Juniper} administration, bash, C, modern C++, Javascript/node, Linux, Darwin, Win32/64, Python, a bunch of other scripting and frontend languages.
  Résumé/CV: email if needed.
  Email: mackeycam25@gmail.com.

I'm an ideal senior devops, SRE, or cloud engineer. My background is unusual because I started as an embedded engineer, had network and system administration duties added, and finally learned how to do security by winning awards at some high visibility contests. Because I've built every hardware or software part of a technology stack in my career at some point, I'm unusually good at troubleshooting weird cloud problems.

I also have a decade of experience professionally administering complex networks. Even if I'm hired as a programmer, I always end up also being "the network guy" somehow, since I've written so much automation to do the job for me. I've been doing devops since way before anyone called it that.

I don't. The stuff I host on Scaleway is there because 1) I don't want to host it in the US, 2) an ARMv8 experiment, or 3) I need redundant bootstrap capacity in case GCE or AWS goes down.

The dynamic stuff I host for personal use uses a variety of cloud services as backends, and those are anycasted. One app I wrote in a hurry uses Google Sheets as a backend, and that's worked surprisingly well for the effort it took.

I just use Cloudflare's free plan with my Scaleway servers, so Scaleway could be hosting from low earth orbit for all I care.

If I have a particularly complex day tomorrow, especially if there's lots of travel, I write down all the important stuff I need to do in schedule format on a 3x5" card. At the end I put down all the facts I might need, like important contacts or access codes or whatever.

I feel this way about frontend technologies that people talk about, but I concluded a long time ago that trying to keep up with the latest churn in that space is pointless. I have a few technologies which are stable and work well for what I need, so I focus on keeping up with that. Every time I've tried to chase the latest and greatest frontend fad it usually turned out it was an immature reinvention of a wheel someone else already built better.

I suspect the serving vessel for the buffalo milk may have been contaminated. Running water to wash dishes with isn't the easiest thing to find out there.

Mom may have been malnourished as well.

The general goal of the research grants is to create a technology stack that takes sunlight as input and provides hydrogen as output. Such a thing would allow production of storable energy in the open ocean, immediately making the rest of the planet inhabitable by people.

I just have a windows 10 machine plugged into the TV. I have a server farm upstairs with plenty of storage for archives of TV shows if I want them, but these days I can get pretty much anything I want from one of the streaming services online.

I also have a glass table between the couch and the TV, so a dark field mouse is required if I'm doing anything that needs a real mouse. For everything else I just use a wireless keyboard/trackpad combo from Logitech.

Or the ACE.

You'll also see what's left of the drawbridge ghost town if you're quick:

https://en.wikipedia.org/wiki/Drawbridge,_California

I've created several bytecode interpreters, mainly for bootloaders and cross-platform drivers. These environments are sometimes very space constrained, so using the native instructions to do highly repetitive I/O to set up devices can blow your space budget. Instead, I shrink the instruction set down to (usually) 4 bit codes with packed immediate data in a 16 or 32 bit word. In one case, a SH4 based set-top box, this reduced the bootloader from 30+k down to a single flash page. This meant that we could have redundant bootloaders instead of just one copy, and that increased platform reliability and serviceability quite a bit.

I also wrote a BPF-like JIT for network packet matching once, targeting the QorIQ pattern matching engine as a backend.

I forget, and the conference site took the slides down. It was more than 10 years ago.

It's all very incestuous. I suspect that part of what happens is that new companies are comprised of people who have previously worked in the industry, so they have general knowledge of who the available providers are. Without prior experience in the industry, I suspect you're not going to be in business very long.

The other part of getting initial sources is probably calling companies in retail, insurance, etc and pitching them on how much they would make selling customer data to them. If I had to do it, I'd probably get industry reports, sort by annual revenue, start at the top and work my way down as I try to get a hold of the consumer data department at each company.

I've also been approached by these guys when I was part of shutting a company down. They wanted to buy our customer database. I guess when one of these companies shuts down, some new one can buy the database and use that as a seed for new operations.

Every time I see one of these new revisions, I can't help but think that the type system has become the roadrunner and the STL has become wile e coyote, complete with dynamite and cartoon physics.

Seriously though, does anyone actually use any of the post-C++-11 stuff? I can count on one hand the times I've seen nonironic use of a C++-11 extension in a real project, and I've never seen use of -14 or -17 extensions.

I saw a great talk by a private investigator at a security conference on this topic, and did some research after the talk to confirm what he was saying.

The short version is what you're asking for is going to be an uphill battle. Data aggregation companies don't want to disclose their sources because their services are often used by debt collectors and other organizations who want to find people who don't want to be found. If the sources were public knowledge, debtors could avoid them to escape debt collection agencies, for example.

The comprehensive list you're looking for doesn't exist. Each data aggregation company has a different list which is the result of many private agreements they have with their sources. If you want a list of the sources, you can't ask the data aggregators.

You can ask the sources themselves. Because you don't know who they are, you have to guess. Any company that puts a card in your wallet would be a good place to start. Under California Civil Code 1798.83, you can email companies and ask them to provide you with a list of all the direct marketing companies they sold your information to. Try making a request to your insurance agency.

1798.34 also allows you to ask California government agencies to provide you with an accounting of everyone they disclosed your information to. A 1798.34 request to the DMV should be a rich source of data providers.

There's also lists of companies involved in the data aggregation ecosystem in the consumer finance protection board's list of complaints:

https://data.consumerfinance.gov/dataset/Consumer-Complaints...

One surprising thing I found out during the talk is that pizza chains are a rich source of data for these companies. If you think about it, this makes perfect sense. The data includes a guaranteed link between a person, a place, payment information, and a phone number.

> when a court needs to order that someone's information be purged

I don't think this is a thing. A friend had to deal with a DV case and I got to see how all the legal machinery involved works. The court doesn't even bother trying to purge the victim's data when they move out and try to stay away from the abuser. The court simply moved the victim and the DV support organization told them to stay off social media and not to give the new address to anyone (including pizza places.)

That being said, the federal government does have a comprehensive list of all the agencies which are members of the federal privacy council. In theory, these agencies are supposed to have a data integrity board which provides oversight for any data they keep on Americans.

https://www.fpc.gov/federal-agencies/

I've been messing around with Scaleway lately and found out a few interesting things.

Scaleway is part of the Iliad family of companies, which you probably haven't heard about if you're outside of Europe. This includes Online, a networking provider and Scaleway's parent, and Free, which is a huge French ISP with a fairly hackery internal culture.

Online makes their network map public:

http://map.online.net/

Based on the names, it looks like it's 100% Cisco with a mix of Nexus 9000s, Catalyst 4500e edge, and 4900 ToR switches.

Scaleway is one of the only cloud providers I've seen that is using NBASE-T technology. The Intel Avoton SoCs used by their C2 instances support 2.5Gbps link speeds using an appropriate external PHY.

Their C1 instances are Marvell Armada 370 XP parts, datasheet here:

http://www.marvell.com/embedded-processors/armada-xp/assets/...

These are using 1 gig connections. The instances I've spun up connect to a gateway with a Cisco MAC address.

The C2 instances connect to gateways with MAC addresses registered to Freebox SAS. Figuring this out took some digging. It turns out that Scaleway/Online is manufacturing their own 2.5G capable switches. Because Free, the ISP, already has an OUI registered and was the only company in the conglomerate which made hardware, it looks like Online had Free make some custom switches to deploy in their datacenters.

They're also making their own BIOS for the C2 instances. dmidecode reports the following cutesy BIOS data:

Manufacturer: Online Labs Product Name: SR Version: (^_^) Serial Number: 42

The root devices for C1 and C2 instances are both NBD. Performance was better than I expected; I get wire rate I/O throughput on both (>90MB/sec on C1 and >250MB/sec on C2). Of course, the disk bandwidth is shared with the Internet connection on both instances, so there might be contention for some applications.

I believe that the Silver Springs network in california is the largest in terms of nodes.

That would be LexisNexis.

I remember that in university we had a test on a toy 16-bit floating point ALU. We had to walk through each of the steps in the ALU for the usual operations, showing the bit patterns at each step and keeping track of precision loss, etc. Because we had to track state and some aspects of the test were randomized, there were a few cases where identical operations on identical inputs resulted in different outputs due to the residual state.

During the next lecture, someone sticks their hand up and says, "this FPU thing is so nondeterministic, who in their right mind would use floating point for 99% of the kinds of work that we use computers for?"

The professor says, "you should think real hard about that the next time you decide to use a floating point type in your program."

The page claims that CPI is an interpreter, but as this post says it is not. It just uses a regular compiler to build the input, then fiddled with the file descriptors to hook up the new executable's I/O to cpi's existing fds.

Thanks!

How do I answer questions about oil tanker availability? I don't see anything in the manual about that. I guess it might be under BMAP or BI, but nothing obvious stands out to me...

Yes, mostly due to protocol backwards compatibility. If you stopped using it because of ads, check out pi-hole.

Oxo hand tools.

Hermin Miller Aerons.

My Olympus skyhawk. Literally the only piece of hardware that's going to outlive me.

The FN P90 and the 57 pistol.

The flight deck on the Sukhoi SU-34 and the Boeing 787.

Basically, anything where the design department went out and asked the customer what their biggest complaints were, then sat down to eliminate them in the design phase.

I've had to build some relatively simple deep learning systems around tensorflow, and I don't have any special training other than the usual engineering math and statistics.

My observation is that it's much more important to be clever with identifying possible inputs to train on rather than focusing too much on the machine learning itself. A crappy ML implementation that was trained on 20 data sets which are highly relevant and well curated does better than a highly tuned ML system with half the inputs and bad outliers.

The Mathematica book might be what you're looking for.

3D graphics math is all trig and linear algebra, with some signal processing. You don't have to know that stuff to use unity to build VR apps, but it helps. Those skills will never, ever, be obsolete. Even deep learning is rooted in linear algebra.

The technology churn in other spaces like front end development can be mitigated by learning to read really fast. I don't have deep experience with any one front end framework, but I can inhale the docs and source code pretty fast when I need to get my hands dirty with one. Again, speed reading is a skill that will never be obsolete.

Oh, so that's why I've been seeing that weird IPv6 traffic on my collectors.

If you can sniff other users' cookies, you can get their passwords. Good thing no one reuses passwords across sites!

We've gone from a time where entertainment was one size fits all, to a world where all sizes try to fit you.