7 ms·

All the big players definitely have spend years and millions (hundreds of millions) optimizing these things. But they are optimizing DOS-based systems built in

All the big players definitely have spend years and millions (hundreds of millions) optimizing these things. But they are optimizing DOS-based systems built in the 80s. The tech stack that the old guard is working with is unbelievable.

thedogeye 11y ago
DHL spent $900M with IBM and SAP, recognized a sunk cost and went back to what they were doing before: http://theloadstar.co.uk/we-are-not-to-blame-for-for-dhls-e345m-nfe-new-technology-debacle-says-sap/ http://theloadstar.co.uk/we-are-not-to-blame-for-for-dhls-e3... Note: the 345M euros is just the amount they wrote off, not what they spent on the project. Maybe they did manage to get the EDI integrations done that were worth the other $600M.
magicmu 11y ago
Ah there it is, I was having trouble visualizing Amazon's edge here too. The old "decent-software + shitty-thing = less-shitty-thing".
kyllo 11y ago
I work at a large freight forwarder and we still have a whole department of COBOL developers and several million lines of COBOL code in production. If Amazon is serious about being an ocean carrier, they would be the first major new competitor to enter this space since before the invention of the internet. All the competition is still mired in 80s technology. But the technology issue aside, my point was really that Amazon has a more aggressive growth strategy and "become a monopoly" mentality than any incumbent player in this field and would be willing to take a loss in order to take market share. I believe they also have a small army of operations science PhDs who understand logistics network optimization better than anyone else in the field, and they also invest heavily in automation and robotics. They have self-driving forklifts as well as automated systems that feed instructions into the earpieces of their warehouse workers now. If they apply this strategy toward freight forwarding, that industry is going to be shaken up in a big way.
analognoise 11y ago
You said DOS like its a bad thing. It's well understood, completely paid for, and does the job. Whats wrong with it, in this case? Even if maintenance costs go up, its still cheaper to maintain than to build a new system, especially if the old system fulfills all the requirements and is proven.
- Karunamon 11y ago
  Technical debt. COBOL programmers are a dying breed.
  - analognoise 11y ago
    COBOL has C linkage, has for a while. You don't have to write the new code in COBOL...
    - kyllo 11y ago
      C is not any easier to build a business IT system in than COBOL is. When these companies want to get their business off of a legacy system, they essentially have three practical choices: 1. SAP 2. Oracle 3. A bespoke system coded in Java or .NET.
- seanlinehan 11y ago
  These systems fulfills all the requirements in yesterday's world. And more importantly, they fulfill the requirements of the company's operating organization; not their customer's. Modern customers want real-time data and analysis. They want all of their information available on the web, mobile, sms, email, API. They want a modern technology experience. A DOS program consuming data via EDI and Excel-based input is a good candidate for the worst technology to do these types of things.
  - analognoise 11y ago
    Why would a customer need real-time data and analysis from a bulk cargo shipment? That doesn't make any sense. I think most customer just want to know when their stuff is going to arrive - they don't need 'real-time data and analysis' (or if they do, they don't expect the guy moving the cargo to provide it for them). Also, those DOS\cobol systems already made shipping so cheap you can get a 40 foot container (full of stuff) from China to LA for $1300. They can't scale? They currently handle the whole world's shipping needs. That sounds like it scaled just fine.

thedogeye 11y ago

DHL spent $900M with IBM and SAP, recognized a sunk cost and went back to what they were doing before: http://theloadstar.co.uk/we-are-not-to-blame-for-for-dhls-e345m-nfe-new-technology-debacle-says-sap/ http://theloadstar.co.uk/we-are-not-to-blame-for-for-dhls-e3... Note: the 345M euros is just the amount they wrote off, not what they spent on the project. Maybe they did manage to get the EDI integrations done that were worth the other $600M.

magicmu 11y ago

Ah there it is, I was having trouble visualizing Amazon's edge here too. The old "decent-software + shitty-thing = less-shitty-thing".

kyllo 11y ago

I work at a large freight forwarder and we still have a whole department of COBOL developers and several million lines of COBOL code in production. If Amazon is serious about being an ocean carrier, they would be the first major new competitor to enter this space since before the invention of the internet. All the competition is still mired in 80s technology. But the technology issue aside, my point was really that Amazon has a more aggressive growth strategy and "become a monopoly" mentality than any incumbent player in this field and would be willing to take a loss in order to take market share. I believe they also have a small army of operations science PhDs who understand logistics network optimization better than anyone else in the field, and they also invest heavily in automation and robotics. They have self-driving forklifts as well as automated systems that feed instructions into the earpieces of their warehouse workers now. If they apply this strategy toward freight forwarding, that industry is going to be shaken up in a big way.

analognoise 11y ago

You said DOS like its a bad thing. It's well understood, completely paid for, and does the job. Whats wrong with it, in this case? Even if maintenance costs go up, its still cheaper to maintain than to build a new system, especially if the old system fulfills all the requirements and is proven.

Karunamon 11y ago
Technical debt. COBOL programmers are a dying breed.
- analognoise 11y ago
  COBOL has C linkage, has for a while. You don't have to write the new code in COBOL...
  - kyllo 11y ago
    C is not any easier to build a business IT system in than COBOL is. When these companies want to get their business off of a legacy system, they essentially have three practical choices: 1. SAP 2. Oracle 3. A bespoke system coded in Java or .NET.
seanlinehan 11y ago
These systems fulfills all the requirements in yesterday's world. And more importantly, they fulfill the requirements of the company's operating organization; not their customer's. Modern customers want real-time data and analysis. They want all of their information available on the web, mobile, sms, email, API. They want a modern technology experience. A DOS program consuming data via EDI and Excel-based input is a good candidate for the worst technology to do these types of things.
- analognoise 11y ago
  Why would a customer need real-time data and analysis from a bulk cargo shipment? That doesn't make any sense. I think most customer just want to know when their stuff is going to arrive - they don't need 'real-time data and analysis' (or if they do, they don't expect the guy moving the cargo to provide it for them). Also, those DOS\cobol systems already made shipping so cheap you can get a 40 foot container (full of stuff) from China to LA for $1300. They can't scale? They currently handle the whole world's shipping needs. That sounds like it scaled just fine.