What Happens To IBM i In A “Zero Datacenter, Zero Mainframe” FedEx?
July 18, 2022 Timothy Prickett Morgan
Federal Express, the pioneer in overnight document delivery for business founded in Little Rock, Arkansas, in 1971 by Frederick Smith, did not start out in the ground shipping business. But it did start its datacenter operations on IBM mainframes, as any company needing complex and intense computing would at the time, and it chose Memphis, Tennessee, in the heart of America, as its headquarters and base of operations in 1973 when the company moved from a PhD thesis at Yale University to a true business.
In 1983, FedEx made history by being the first company in the United States to reach $1 billion in sales within a decade of the beginning of its operations, and it started intercontinental operations in Europe and Asia. In 1985, as the company’s growth was phenomenal – and United Parcel Service was fighting back with its own document delivery. (And by the way, the US Post Office started Priority Mail for documents in 1968, five years before Smith tried to figure out a better and faster way to do it.)
In 1997, when FedEx shelled out $2.45 billion to buy Caliber System, a package and freight delivery company founded in 1985 in Pittsburgh, Pennsylvania, FedEx acquired the RPS shipping business and its Roberts Express, Viking Freight, Caribbean Transportation Services, and all of the related logistics and technology it had developed. This business, which was founded in 1985 and which you know as the FedEx Ground Service, had a triplet of IBM System/38 minicomputers as its scheduling brain when it started its operations, and since that time the RPS and FedEx Ground operations have relied on System/38, AS/400, and IBM i platforms.
What you may not know is that over time, FedEx has retired over 80 percent of its “mainframe footprint,” as Rob Carter, the company’s chief information officer, put it in a recent meeting with investors. We did not attend the event, and can’t seem to find the presentation or the recording or transcript for it online, but our friends over at Data Center Dynamics listened to the call and got the gist of what Carter told Wall Street.
“We’ve shifted to cloud,” Carter explained. “We’ve been eliminating monolithic applications one after the other after the other . . . we’re moving to a zero data center, zero mainframe environment that’s more flexible, secure, and cost-effective. Within the next two years, we’ll close the last few remaining data centers that we have, we’ll eliminate the final 20 percent of the mainframe footprint, and we’ll move the remaining applications to cloud-native structures that allow them to be flexibly deployed and used in the marketplace and business. While we’re doing this, we’ll achieve $400 million of annual savings.”
FedEx operates a huge datacenter in Colorado Springs, Colorado, and it looks like it was thinking about building another datacenter in its home state in the city of Nashville, but apparently that did not happen. Three years ago, FedEx started decommissioning some of its 11 regional datacenters, moving its workloads to a massive datacenter operated by Switch in Las Vegas. Just because systems move to a co-location facility does not mean the applications are not ending up on more modern and hosted versions of the platforms. So don’t jump to the wrong conclusions. FedEx created a tool called Surround analytics and put it on Microsoft Azure, and FedEx also uses Oracle Cloud for some workloads, too.
Wording matters here. When a lot of people talk, they lump IBM System z and IBM Power running IBM i into the same word: “Mainframe.” We all know that the AS/400 and its progeny are not a mainframe, but an improvement over the mainframe – and so by design. Even Big Blue knows that. So, we have no problem believing FedEx is cutting loose from the IBM System z for its FedEx Express and FedEx Freight lines of business. It needs that money to help for the innovation it needs to bring to bear to compete with Amazon, the retailer that is also building its own shipping business, and UPS, which is stuck being addicted to Amazon and at the same time wishing it would stay out of shipping. The US Post Office is just a drive-by observer caught in the crossfire.
We did have a hard time believing that that the company would move off the Power Systems platform and the IBM i operating system and database, however. At least not all at once. It seems far more likely that FedEx would build new applications that bring all three of its core businesses – Express, Ground, and Freight – into one holistic system but at the same time allow it to preserve their unique tasks and identity while sharing resources to stretch from the first mile to the last mile in any delivery. That is the game these days, which was explained well by Supply Chain Dive.
We reached out to some people about the IBM i and Power Systems machinery in the FedEx Ground organization – these are people who would absolutely know about what is going on there – and they said that since the coronavirus pandemic hit, FedEx has pumped a huge number of transactions across its network of Power machines and as far as they know, FedEx is very happy with the IBM i platform.
Still, FedEx might put Power machines in a Switch datacenter in Las Vegas or elsewhere, or use utility pricing from IBM to get a cloud-like budget, or use a mix of virtual Power instances on IBM Cloud, Microsoft Azure, Connectria, or Google Cloud as well as co-lo iron to run its workloads and support its applications. There is even an outside chance that those remaining IBM System z applications will be ported to Power Systems, probably running a mix of IBM i and Linux, possibly AIX. Who knows?
We know two things. One, FedEx Ground seems keen on keeping IBM i. And two, it is very hard to move off of any platform where you have legacy code that works and you need to keep the business running. Microsoft famously had a very hard time moving off the AS/400 platform, about a decade after it became an enterprise platform supplier itself. And then, when it did, it moved its AS/400 applications from datacenters in the United States (Washington) and Ireland (Dublin) to a co-location and hosting partner in Ireland. Technically, it no longer ran AS/400s, which it said when asked, but in reality Microsoft’s manufacturing operations were on AS/400s – just those owned by someone else. At some point, Microsoft ported its code to a Windows Server platform, but it almost had to given that this is what it was telling users of Unix and proprietary systems to do. Dell had a similar difficulty getting off Tandem fault tolerant SQL servers after Compaq bought DEC and Tandem before it was itself bought by Hewlett Packard. There were several attempts at porting applications before Dell succeeded. And in both cases, this was a decade-long journey.
We would love to know what is really going on, of course, and how FedEx is handling the technical and economic challengers of modernizing its platforms for running businesses. This is about as hard as anything on Earth gets.
RELATED STORIES
Beware The Hype Of Modern Tech
AWS Offers Mainframe Migration Service. Is IBM i Next?
Inside Jack Henry’s Long-Term Modernization Roadmap
Modernization Trumps Migration for IBM i and Mainframe, IDC Says
The State Of The IBM i Base 2022, Part Two: Upgrade Plans
Microsoft Wants to Migrate Your IBM i Code to Azure
Is The IBM i Skills Shortage Accelerating Platform Migrations?
SAP on IBM i to S/4 HANA Migration: ‘No Need to Rush’
Settling In With IBM i For The Long Haul
I sometimes find myself amused…. What does actually “Zero Datacenter” mean? Somebody external that does run your ITC loads?
I personally live on a rent, does that make me “Zero Home”? Home-less (that doesn’t sound so cool)?
When did the number zero get so much seducing appeal in our era?
Joking apart, “server-less” actually does carry in the common vulgate a computation model / architecture valence, being based on referential transparency and lambdas, that allows to see the computation model as a black box itself, thus transferable between machines).
( and in some sense *PGMs on “i” called dynamically can be seen in this way, if one wishes, or as a microservice, without the hassle of unneeded distributed systems).
So it would be nice if the author would follow up the story especially on the software/platform architecture parts, Fedex is indeed an excellent case study for all of us.
I could not agree more.
As for following up, don’t hold your breath. I would love nothing better than to do a profile in compute of Fedex. There is no way in hell their lawyers will let me talk to their IT people. No Fortune 500 company, no Global 2000 company, will do this. I have spent decades trying to get such stories, and they happen so rarely. Almost despite PR. They hyperscalers and the cloud builders are a little better — but still crazy secretive about some things more than makes any sense — because they use their IT design and operations savvy as a recruiting tool. But AWS and Google talk when it suits them.
Small quibble with wording on this sentence near the end: There were several attempts at porting applications between (before) Dell succeeded.
Perfectly good quibble. Thank you.