FedEx signals 'zero mainframe, zero datacenter' operations by 2024 The datacenter is dead – at least according to FedEx, which announced plans to close its server farms and transition completely to the cloud, where it hopes to save an estimated $400 million annually. At FedEx's investor relations day held last week, CIO Rob Carter said FedEx had long been a leader in technology, claiming the …
So, despite it being already openly stated that cloud does not savo you money, you expect to save not only millions, but hundreds of millions, by going to The CloudTM.
I await the follow-up on how you are desperately trying to waive an enormous bill of $700 million.
Good luck.
And what, exactly, does FedEx think "the cloud" is beyond other people's datacentres...
And they still manage to lose packages and have others delayed "awaiting information for customs" which you've already given them.... multiple times.
A datacenter where it can use opensource distributed database rather than paying DB2 licenses on top of the IBM OS licence.
A datacenter where it can buy extra capacity for the christmas rush without having to start pouring concrete in January to be ready for the December delivery of another mainframe.
And if they did their design correctly a datacenter where they can switch between Amazon.Microsoft/Goolge easily to get the best price rather than IBM looking at Fedex's profits and decided what next year's maintenance contract will be.
A datacenter where it can use opensource distributed database rather than paying DB2 licenses on top of the IBM OS licence.
I am pretty sure you can run any distributed RDBMS you want on your own servers, open source included.
A datacenter where it can buy extra capacity for the christmas rush without having to start pouring concrete in January to be ready for the December delivery of another mainframe.
Microsoft's Azure cloud is having difficulty providing enough capacity to meet demand
You are just banking your company on that the cloud provider is going to have required capacity when you need it and if they don't have it, you can't just order a server and tell them to stick it in. Also cloud infrastructure has a ton of overhead - you would have to provision much more resources to match the perfomance of natively running infrastructure.
And if they did their design correctly a datacenter where they can switch between Amazon.Microsoft/Goolge easily to get the best price rather than IBM looking at Fedex's profits and decided what next year's maintenance contract will be.
Correct design would have them running on at least two cloud providers simultaneously, so when AWS goes down, they could still service the workloads on the second provider without having customers to experience downtime.
"Microsoft's Azure cloud is having difficulty providing enough capacity to meet demand"
I miss the good old days. When calling in for some service, I'd get the reply, "Sorry. Our computers are down." With the air of finality that there is nothing to be done about it, as it's some fundamental law of nature or something like that.
"Also cloud infrastructure has a ton of overhead - you would have to provision much more resources to match the performance of natively running infrastructure."
Been there, done that. I was the chief cook and bottle washer for an internal corporate system that consisted of some very lightweight web services. We were given the option of "buying" in-house server support from our IT department. At some outlandishly high rate. About $50,000 per month per server* IIRC. But why worry? It's just inter-organizational "funny money". Not real cash. The alternative (which we chose) was to rescue a couple (for redundancy) of Sun "pizza box" workstations and reconfigure them as web servers. They were sufficiently powerful enough to support the demand of some dozens of concurrent users.
*I came to find out that this figure was high due to corporate maneuvering to sell off our IT division. Big numbers like the above made that business unit look quite profitable, justifying a very high sales price. Which came back to bite the eventual buyers in the *ss when contracts were renegotiated with them for much lower, competitive services. The moral of the story is that mere peons like myself can never understand the devious machinations of the BOD or accounting department. There is often far more money to be made reorganizing business units than actually rolling product out the door.
"You are just banking your company on that the cloud provider is going to have required capacity when you need it"
Yes, "cloud" may have started out with those lofty ambitions and ideals, then the accountants took over and asked why all this spare hardware was just sitting there hoping for a paying customer. "Do we really need 20% over capacity just in case? Why not cut it to 15%? 10%? 5%?. Why not just optimise for 100% usage and slow down the less important customers and sell their resource to the higher paying customers? Just put that in the Ts&Cs of the cheapskate contracts. After all, they sold all their on prem kit and are trapped here now.!
Why not run an open source OS and an open source database on your own hardware? Even IBM hardware can run your choice of OS and database.
Is it cheaper to pay the premium for having that ability to increase peak capacity when needed rather than to just have it year round?
That is assuming the capacity is available when you need it: https://www.theregister.com/2022/07/04/azure_capacity_issues/?td=rt-3a
As for being able to easily switch, cloud providers will do their best to make it as hard as possible.
>As for being able to easily switch, cloud providers will do their best to make it as hard as possible.
The majors will do this, just as we saw with on-prem systems and their proprietary OS's etc.
What we actually need, is to encourage the smaller "white label" cloud providers;, these see a commercial benefit in having cloud platforms to which loads from other cloud platforms can be readily transferred to.
Fedex's mistake is not to sell/lease its datacentres to such a cloud provider...
>Why not run an open source OS and an open source database on your own hardware
Mainframe licenses are immensely lucrative, you want to drop DB2 and run your own software then the OS license fee doubles, or you don't want to upgrade to the new OS then the service cost doubles.
Whether migrating their applications to the cloud saves money or not depends on how good they are at managing their own datacenter and servers.
If they are really bad at it, are out of datacenter space, and know that they would have big business continuity problems if they lost a datacenter then it could be a very good idea that will save them a lot of money. I don't know if FedEx are good at running their own datacenters but the management probably considers their business to be about delivering parcels better than many small bit players do and not running datacenters where they are bit players compared to the likes of Amazon etc.
Success at delivering parcels depends entirely on having the right systems to control it. Whether they reslise it or not or whether they like it or not they are an IT business.
Amazon are also in the parcel delivery business wnd realised they were an IT business. That's something FedEx seem not to have noticed.
>they are fully aware of how much they rely on IT
Don't believe you.
Anyone fully aware of how much they rely on IT would know IT is very fickle about $savings and so held the press release until they could say: we moved to "zero mainframe, zero datacenter" and <u>saved</u> $x00m capex and $y00m opex.
Mind you the only time I've seen this level of detail and understanding was back in the 1980's when Lucas implemented EDI: they knew exactly what they were saving year-on-year.
I'm sure the C-Suite will be getting nice fat bonuses deposited into their accounts, and probably the CIO was wined and dined quite extensively. And what do they care? In 2-3 years, tops, C-Suite executives will play musical chairs again and they'll be off at some other company when shit hits the fan at FedEx... probably dealing with the fallout of someone else's short-term decision from 2-3 years ago.
I always cringe at the idea of giving up control of your servers like that. If there's some kind of service outage, you're at the whims of someone else's employees to fix it, you're trusting that they won't exfiltrate sensitive data, and that they aren't just outsourcing the job to some poor sap from India who's effectively held hostage by a visa and likely is thrown into the deep end with training that amounts to being told "don't drown."
Sure, some of those are risks even with your own employees, but you generally have more legal options open to you if you're the employer vs someone else. Plus, if you are hosting things yourself, presumably you have a few admins keeping an eye on the underlings. Even if the contracting company has a few admins supposedly watching the grunts, they're having to keep track of probably hundreds of different clients and it'd be pretty easy for something to slip through the cracks.
People tend to think cloud doesn't go down. But it happens even multiple times a year. You can then just raise a ticket and pray it is something simple and whoever is looking into it, unlikely cares about your business, but rather to close the ticket as soon as possible (and depending on pay - nobody is going to play a hero for a poor wage).
...spend more OpEx.
Buying hardware is capital expenditure. Paying a monthly subscription to AWS is operational expenditure.
Everything else depends on which department you are in, and how you massage the numbers.
Usually - though in a longer time period than the bonus cycle - eventually you pay AWS more than it would have cost to buy and run the servers you need. Maybe FedEx isn't planning to be around for more than 5 years?
...that still uses an honest to God answering machine for one of their depots in the Greater Toronto Area.
I had a package stuck in customs for ages, and then it was stuck at the depot for ages, because the driver couldn't figure out which unit in an industrial building to deliver it to.
The depot would call me to tell me this every time they tried to deliver it, and I swear on the FSM that it was an answering machine every time I called them back. I wouldn't have been surprised if it recorded to a micro-cassette.
I had already had a very low opinion of fedex prior to that incident, and today I have an even lower one. I didn't think that was possible.
Want to track your package? Contact your local black hat hacker. Just ask the Shanghai National Police, British Army, Geographic Solutions (GSI), OpenSea, Uber, and dozens (hundreds?) of other companies. Don't you just love it when the bean counter over-rule the people who are supposedly the IT security pros?
And when Fedex will move everything to Cloud's providers, they can expect nice prices rise when they'll be captive in the hands of their providers. But like @aerogems said it well, the ones in charge won't be there when this time will come, having collected their fat bonuses and departed. Robert B. Carter, Fedex's CIO is close to retirement at 63.
In Q1/22, Fedex made net income of $1,100,000,000, with operating margin 6.4% . So why doing this? For those greedy shareholders, even if it means loyal workers throw to the bin.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
