Schlumberger is building a proprietary OS for its oilfield services | Hacker News Reader

Schlumberger is building a proprietary OS for its oilfield services

fortune.com

32 pointscarlsborg41 comments

I interned there last year. They use a Windows 7 image that is preloaded with drilling software built in-house. Most of the newer software seemed to be written in .NET.

They have some really awesome tech, especially from an electronics point of view. Some of their newer MWD (measurement while drilling) tools have state of the art sensors for every physical phenomenon known to man. The drilling operator can therefore "see" everything. The electronics systems that go downhole can withstand insane temperatures (125C), shocks, and RPM.

So, but that isn't really a "proprietary OS" if it's a Windows 7 image loaded with software.

That internship sounds pretty cool. I bet they do have some neat tech.

I guess that's what they're switching from. It seemed to work fine for them, especially since most reporting can be done using MS Office, which everyone is familiar with. Maybe it's getting expensive to support, given how many licenses they need?

Yeah, the technology is insane. I'll give you an example of what I mean by that.

Think of this situation: all of the parts of the drill string that contain the sensors that provide real-time environment conditions AND control the direction of drilling are, say, 3 km underground. How the hell do you send movement commands or receive sensor data? It's not practical to extend a wire for that distance. Wireless communication is unreliable at best underground. More importantly, how do you provide power to these tools?

Luckily for them, mud is continually pumped through the drill string for many reasons, including cooling the drill bit and breaking down the formation. So they used that mud to communicate and transfer power.

The general idea is to vary the flow rate on the surface to transmit binary information. To send data upwards, the mud flow is blocked and unblocked, and the pressure change is detected on the surface. It's pretty fascinating stuff!

That's pretty amazing. Could you or anyone else elaborate why would providing a conventional electrical conductance based communication channel be difficult ? I am sure it is, otherwise they wouldn't go to such great lengths( or depths, in this case)

I'm no expert on the subject, but according to Wikipedia [1], there is in fact a technique that involves using a wire, but it's still not in widespread use.

[1]: https://en.wikipedia.org/wiki/Measurement_while_drilling

That article still mentions the use of mud pulse telemetry, though.

Considering the low bandwith of mud, pulling a cable all the way down is probably not an option, or else it would already be done. I wonder if it's possible to use a relay of short-range wireless transmitters, with heavy error correction?

Bandwidth of mud telemetry is 'up to' 40 bps or 250 with compression, according to this: http://www.bakerhughes.com/products-and-services/drilling/dr...

For now Mud Pulse telemetry is well researched and pretty reliable technology. But I would not make a hazard guess what will happen tomorrow.

Mud does sound very reliable or having much bandwidth. Plus is the drill not made out of metal? Why not just use the entire drill as a conductive "wire"?

No, mud pulse telemetry doesn't provide much bandwidth - it transmits at around 10 bits/sec. Interesting tidbit: I was told that the flow rate is manually controlled at the surface, as in when a command needs to be sent downhole, a guy goes to the control valve and adjusts it manually.

As for using the drill string itself, since there is no separate return path, no current can flow through it. Further, there is no way I can see them setting up a common ground to get a voltage across the drill string. For this reason, they need to run a cable down the drill string to do this.

That is indeed fascinating stuff. Hydraulic circuits. Crazy innovative.

I worked in a similar field until recently ... a lot of the ICs being used were automotive rated, hence the high temp rating. Also, pressures on the outside of the tool get massive - up to 15,000psi depending on depth.

Is it automotive? I was thinking of military grade..

I left SLB few months ago , and during my stay there I was exposed to most of their MWD tools. I worked both in the maintenance and field engineering organization. Please note that all MWD(D&M) tools are expected to withstand 150 degC however some high temperature tools could get up to 170 degC. But this is just industry standard (Baker, Halliburton, and others have such).Their electronics technology has been around for decades, however SLB started off(recently) making incremental improvement and at same time incorporating new technologies. Please don't use the word "see" because there is nothing radical here or new. MWD tools are just sensors.. magnetometers, accelerators, strain gauges , FPGA and some basic electronics power system. In some cases you may have transmitters and receivers oriented to a particular frequency range. MWD tools have sensors in the same family as what we have in a smart phone.

How about density, porosity, resistivity, and so on? I saw some LWD tools that used different sonic-based techniques, nuclear sources, and others for formation evaluation. I'm thinking of the PeriScope, EcoScope from LWD, and the TeleScope from MWD. The point-the-bit tech in the Xceed is also amazing!

I didn't get to read up on the details, as they prevented interns from accessing technical documents, but from what I saw some of the sensors were quite advanced. Granted, I don't have that much exposure to state of the art sensor tech, so I could be wrong.

I wont like to discuss SLB technology here because that is the duty of their salesmen, however I will pick on "point-the-bit tech in the Xceed" technology to illustrate a bit. Xceed is modeled after an MWD tool called PowerPulse(TeleScope predecessor), it copied about 90% of that technology. However to the uninitiated it may look like something from the outer space. It has resolver(motor), that means you have angular position, and throw in magnetometers, accelerators , and sensors into the play. Presto!

Here's the wording about the OS:

This new operating system will be able to analyze everything happening on a rig in real time, boosting both efficiency and safety. It also serves to lock customers into a single ecosystem that the company hopes will act as an incentive to use Schlumberger services and (now) equipment for all their drilling needs.

So their requirement is real-time. Their real requirement though is vendor lock-in. They're really just looking for a way to lock companies into their platform so that the cost of switching is prohibitive. That's a way of decreasing safety and efficiency. I think the real-time requirement is just a cover for them to increase their customer retention through the worst way possible.

Offering a little "extra" to the client in the form of lock-in is despicable.

Old-school operators in the energy space are all about going with “the best in breed” equipment and service providers. They have deep relationships with salesmen in certain companies and tend to be very loyal to them, so it will be hard to sell many of them on the idea of buying from just one source.

Oh I see, they're exchanging gifts at different levels and if you lock yourself into one vendor you won't be able to get a wide variety of gifts for closing a sale.

I wonder how "best of breed" is measured, if it's measured at all or if it's just based on marketing and sales brochures.

There is nothing new here ..... They already have the technology but fragmented .. They have been streaming real-time data to clients and their support centers for over a decade now. Now they want to package all that as one... pretty business strategy. But I guess they would not fool anyone.

I wonder if this is an operating system, or what a laymen would potentially call an operating system, maybe some sort of full screen application. I would guess the latter.

The way it actually worked in 2004, I would be amazed were it any different now:

* it's a complicated vertical market with buckets of cash

* there are two main providers, Landmark and Schlumberger

* you buy most of your stuff from one or the other

* the software, at $10k/seat/year, is the second cheapest part, the cheapest being the beefy box with top-of-the-line 3D graphics to run it on. the highly-paid geologist or petrochemist or whatever to sit there running it is the third-cheapest part.

* the software is unbelievably shoddy for how much money goes into it. X applications in 2004 that would refuse to run in 24-bit, demanding you restart in 8-bit, that sort of thing. Hence my rule of thumb that "software quality is inversely proportional to ticket price."

* OH MY GOD THERE IS SO MUCH MONEY SLOSHING AROUND. So it was nice to sysadmin somewhere that would just spend money to do stuff properly.

* I daily wanted to stab FlexLM in the face. And it's the least-worst option on dongleware.

So maintaining their own complicated system of stuff that runs on Windows 7 is already what they do, and is relatively cheap compared to how much money is in the vertical.

They could go with Lisp :)

Ada or ATS-Lang would be good for such mission-critical applications with various modules needing to keep it safe. The DoD and aerospace already use Ada for such purposes. Just please don't use C! (Maybe BitC, but too immature). Then run Erlang (Elixir or LFE - Lisp Flavored Erlang) on top of the OS for distributed computing, robustness, hot-swapping, and all else where Erlang excels.

What's wrong with the suggestion of Ada or ATS-Lang? Lisp was used on space missions. Relatively new to the whole karma, upvote and downvote on HN. I would like a rational reason on why this was downvoted to better understand what was off about my reply.

HN in general does not appreciate comments that can be construed (or miss-construed) as language fanboyism. Bear in mind that when using written communication the onus is on the writer to convey their meaning accurately.

Thanks for taking the time to try and explain. Given I mentioned 4 languages with different reasons also puts some onus on the reader to have an aptitude for comprehension as well. How can 4 languages be fanboyism? I will, however, bear this in mind, although, it does come across as a little pretentious or guarded on the downvoter's side.

the phrase "Operating System" here is being used in the sense of "software to manage resources"... not the resources of the computer, but the resources of the whole system. It is not referring to a replacement for Linux, Windows, VxWorks or uCOS, rather incorporating individual subsystems running those OSs into a whole oilfield-management-system.

“It is similar to an operating system used on your mobile phone,”

Well that's the technical bits out of the way :)

We already know that software doesn't need to be run on the same OS in order to work together.

Whats wrong with just helping Linux ?

I always find it funny when people try to create "proprietary" stuff in 2015 - its like people learnt nothing from windows.

I remember reading about how investment bankers tried to create their own language and it looked like a crude version of C.

Its not easy to compete with open source. You just cannot spend 1 billion dollars making an OS. Who is going to fix bugs for the next 20 years ?

Protecting trade secrets and the GPL don't necessarily mix well.

Google manages to do that just fine.

But google doesn't sell software. Most of their sofware is never distributed outside google (apart from the parts they strategically open-source), so the GPL can't force them to open up their source code.

It's also quite probable that a lot of Google's critical software might not even be covered under GPL. For Linux, unless they're modifying the kernel code, it wouldn't really matter right?

It doesn't matter either way. Google doesn't distribute their software (except for android, chrome, etc... which does have sources available).

... Which is exactly what the parent said. I'm saying that even if they DID distribute, it still wouldn't matter, as most likely, Google's main-value stuff isn't a modification to GPL software...

Depends on how you define "main value stuff." One of Google's largest engineering teams is infrastructure working on core services, the Kernel, OS, drivers, hardware etc..

Neither is the oil industry.

But this company is not the one doing the drilling or running the software. They sell hardware and software to the oil industry

The whole industry? There's quite a few companies that sell hardware and software. This company being one of them.

For one, Linux isn't an RTOS and never will be. The many rejections of real-time patches into the Linux kernel tree have shown that the kernel devs favor performance to precision, so it would be a fool's errand to try once again to push a real-time patch just to get it patched. At which point, you'll end up having to do a lot of your own OS development anyways to backport kernel features you need. There are quite a few situations where Linux will delay interrupts for milliseconds, and quite a few situations where milliseconds can mean millions of dollars.

For your last question, 20 years of support will certainly not Linux developers. While MS offers extended support for Windows XP, neither earth nor the heavens will make Red Hat provide extended support for RHEL 3. The Open Source world has a lot to learn from proprietary companies about support still.

According to Cyph0n, they are using a Windows 7 image, which is also not a RTOS. Can just as well use Linux in this case.

If they are writing their software using .NET then windows would seem to be obvious choice (for now).

True, although mono might be an option.

Reply on news.ycombinator.com