TL;DR: You use the cloud every time an app on your screen reaches a program running on someone else’s machines: browser mail, photo backup, Netflix streams out of Amazon data centers, and games rendered on remote servers all follow one arrangement, a thin client on your side and a shared back end on theirs.
- Software as a service means the provider’s application runs on their infrastructure and you reach it through a browser; the NIST definition’s own example is web-based email, and Gmail alone has 1.8 billion users.
- Netflix has run on Amazon Web Services since a 2008 database corruption stopped its DVD shipments for three days; cloud games go further and render on the server entirely.
- Of the 2010s product shelf, Mailchimp, QuickBooks Online, Box, Toggl Track and Evernote are alive; Mozy shut down in 2019 and Outright in 2022, while their categories carried on.
- On the phone, you use the cloud through mobile cloud computing: heavy computation moves off the device, paid for in frequent transmissions and wide-area latency.
You use the cloud before breakfast and again on the couch, and almost none of it looks like “cloud computing”: mail loads in a browser tab, a photo you took yesterday is already on your tablet, a show is queued for tonight. This article is the written half of the Whiteboard Architect episode You Already Use the Cloud Every Day (Even If You Never Noticed); it follows the same chapters with the same fact-checks, and it checks the 2010s product lists that most tutorials still repeat.
What it means to use the cloud
Strip the marketing off “the cloud” and one arrangement is left: the provider’s application runs on the provider’s infrastructure, and you reach it through a thin client, usually a browser. That is software as a service, and it is the shape almost every example in this article takes. The NIST definition of cloud computing makes the point with its own worked example: web-based email, the case where the mail program, the storage and the search index all live somewhere else and your browser holds only the front end.
Gmail is that example at full scale. It opened on April 1, 2004, and is now the largest email service worldwide, with 1.8 billion users. Nothing about your inbox runs on your machine: the spam filter, the archive and every attachment sit in Google’s data centers, and any browser that reaches them gets the same inbox. Some services cannot exist otherwise. Apple’s Siri and Google’s Gemini assistant need cloud computing to answer at all, and the browser editors, Microsoft 365 for the web and Google Docs, run their editing surfaces on top of cloud storage, so two people can type in one document from two continents.
It is cloud computing under the definition the field uses. NIST’s service model for SaaS puts the consumer inside applications the provider runs on cloud infrastructure, reached from a client device through a thin client, with the browser as the usual one, and it names web-based email as the example on purpose: mail was the first application most people met that kept its state entirely off their own hardware. Before webmail, “checking your email” meant a program on your disk, a mailbox file it downloaded into, and an ISP server that only held mail until you collected it. Webmail moved the mailbox, the program and the index onto the provider’s machines at once, which is why the definition’s authors reached for it. A modern webmail account also shows the trade you make: your mail is searchable from any device because it is not on any of them, and the provider can read it at rest as easily as you can read it in a browser.Is webmail really cloud computing, or just a website?
Your day is a cloud tour
Run through an ordinary day and the back ends stack up, mostly unseen. In the morning, photos: Google Photos, launched in 2015, backs your photos up to its cloud service, which makes them accessible from all of your devices. The picture you took on your phone at breakfast is on your laptop before you sit down, with no cable and no copy step, because the library lives on Google’s machines and your devices are all windows onto it.
Board 3 of 13: the day tour, back ends invisible.
In the evening, streaming. Netflix looks like a delivery service for video, and it is, but it runs on Amazon Web Services, and the reason is dated: corruption took out Netflix’s own database in August 2008, DVD shipments stopped for three days, and the company responded by moving its data to AWS and announcing plans to migrate all of its infrastructure there. The app on your TV and the recommendation engine that picked the show are clients of someone else’s data center.
Music and games finish the tour. Spotify has streamed since 2008 on a freemium model, and the catalog it searches holds over 100 million songs; none of them are stored on your phone. Cloud gaming takes the idea one step further, because the work that moves is the game itself: Xbox Cloud Gaming opened for Game Pass Ultimate subscribers in 2020; Nvidia’s GeForce Now, from the same year, does the identical job. Both render the game on distant servers, and your screen receives nothing but video.
The video stream looks the same; the input path is what makes it a different service. When you stream a film, the server plays a finished file at you. When you play a game through Xbox Cloud Gaming or GeForce Now, every button press leaves your device, travels to a remote server, gets applied to a running game, and comes back as changed video, so the round trip sits between your thumb and the next frame. Xbox describes the service as a link between your device and remote server hardware, and the servers themselves are Xbox-based blades, real console hardware stacked in a data center. GeForce Now does the same job from data-center servers of its own and delivers the game through streaming video. That round trip is also why the arrangement lives or dies on network quality: the compute is never the bottleneck, and the radio link usually is.How does cloud gaming differ from streaming a film?
The incident was small and the decision built on it was not. In August 2008, corruption in Netflix’s database left the company unable to ship DVDs for three days, which at the time was the core of the business. The fix the company chose was to move its data to Amazon Web Services, and it then announced plans to migrate all of its infrastructure there. First, the migration went to a direct competitor in streaming video, which is the kind of decision you only make when the engineering argument is overwhelming. Second, the pattern it produced is now the default: a video service whose own machines hold none of the product, running its whole operation on rented infrastructure, exactly the arrangement this article keeps finding in every category. When a service you use daily turns out to live on somebody else’s computers, Netflix’s three-day outage in 2008 is one of the reasons that arrangement became normal.What did Netflix’s 2008 database corruption actually change?
The 2010s product shelf, checked
Cloud product lists age badly, and the list most tutorials still quote comes from the 2010s. Checked item by item, it splits into survivors, renames and closures.
| Product | What it does | Where it is now |
|---|---|---|
| Mailchimp | Email campaign platform | Alive; Intuit has owned it since 2021 |
| QuickBooks Online | Cloud accounting covering invoicing, expenses, bank reconciliation, reporting and bookkeeping | Alive; Intuit’s cloud accounting service |
| Box | File storage for businesses, reachable from any connected device | Alive |
| Toggl Track | Time tracking in the browser, from Tallinn, Estonia | Alive |
| Evernote | Notes sync against a master copy kept on Evernote’s servers; reads text inside images | Alive |
| Posting text, photos and multimedia | Alive | |
| Google Apps for Business | Gmail, Drive and the Docs editors, edited together in real time | Renamed G Suite in September 2016, then Google Workspace in October 2020 |
| Following accounts, posts landing in your feed | Renamed X, announced July 22, 2023; following and feeds work as before | |
| Mozy | Online backup pioneer | Closed in 2019 after passing EMC, Dell, then Carbonite |
| Outright | Small-business bookkeeping | Bought by GoDaddy in 2012, discontinued June 18, 2022 |
In that table the products expired and the categories did not. Mail became a browser window, files became a synced folder, notes became a master copy with devices attached, feeds and music became streams. Every one of those categories settled into one shape, a phone or browser front end over a shared back end in a data center, and when one product closed, another product in the same category took the traffic. When a new app arrives, ask which back end its window opens onto.
The two closures on the list, Mozy in 2019 and Outright in 2022, both arrived through acquisition rather than as independent failures. Mozy, founded in 2005 as an online backup pioneer, passed through EMC in 2007, Dell in 2016 and Carbonite in 2018, and Carbonite shut it down in 2019, folding its customers into Carbonite’s own backup service. Outright, a small-business bookkeeping site, went to GoDaddy in 2012 in a deal worth around twenty million dollars, was rebranded as GoDaddy Online Bookkeeping, and discontinued on June 18, 2022. Neither closure hurt the category it served: online backup is now a large market with several established providers, and cloud accounting is dominated by the very products still listed as survivors above, QuickBooks Online among them. The renames tell the milder half of the story. Google Apps for Business turned into G Suite on September 29, 2016, and then Google Workspace on October 6, 2020, with the same real-time collaborative editors underneath both renames. The rebrand to X was announced on July 22, 2023, and the mechanics its users cared about, following accounts and reading a feed, carried straight through it.Which 2010s cloud products closed, and what replaced them?
Inside the phone: mobile cloud computing
The phone in your pocket is the cloud client you notice least, because the heavy work it shows you often happens somewhere else. Mobile cloud computing, usually shortened to MCC, is the named field behind that: Wikipedia’s lead sentence says it combines cloud computing with mobile computing, with the computing work running on cloud hardware, not on the handset. A phone with tight limits on power, storage and energy borrows a server’s resources and still runs applications that would swamp it.
The mechanism has a name too: computation offloading. Heavy, resource-hungry tasks get transferred to the cloud, which sidesteps the device’s ceilings on compute, storage and energy, so jobs finish faster and the battery lasts longer. Your phone does not need a fast chip for a task it never runs; it needs a radio and a screen.
MCC runs on frequent transmissions between device and cloud, and long wide-area network latency is one of the field’s named challenges: push the work off the phone, and the radio link is what decides how the app feels. The same challenge list continues past latency, and it is worth reading in the deep dive below, because the entries named there, lock-in among them, are the ones your own apps live with.
Board 8 of 13: the phone borrows cloud muscle.
The field’s own challenge list is longer than the latency entry. The sourced list names computation offloading itself, mobility management, seamless connectivity, long WAN latency, the device’s energy constraint, vendor lock-in and data lock-in, then security and privacy, and it notes the field still lacks a reference architecture for heterogeneous environments, meaning no settled blueprint for mixing devices, radios and clouds that all differ. One entry deserves unpacking, because tutorials garble it. What some articles call live VM migration is, in the sourced literature, VM-migration-based offloading: the application gets packaged inside a virtual machine, and that VM is migrated to the cloud, where it runs. When researchers diagram MCC’s resources, the taxonomy the field cites lists three resource shapes, hybrids on top: distant immobile clouds, the conventional cloud; proximate immobile entities, machines near you that stay put; and proximate mobile entities, the devices that move. A fourth bucket, distant mobile entities, appears in some summaries but does not verify, so this article leaves it out. For everyday use, two entries from the list matter most: lock-in, because your offloaded work runs on infrastructure you neither own nor can easily leave, and security and privacy, because the data crosses a network you do not control.What problems does mobile cloud computing still have?
Gotchas: dead platforms and overclaims
Three checks apply before you repeat any of it. First, date any product or platform list before you repeat it. The episode’s own source lists five smartphone platforms; three no longer exist: Symbian, whose support ended on January 1, 2014, after the last Nokia Symbian phone shipped in 2012; Windows Mobile, which Windows Phone replaced in 2010 before its own deprecation; and BlackBerry 10, end of life on January 4, 2022, after BlackBerry pivoted to Android. The survivors are Android, the most used operating system worldwide for phones since 2011, and Apple’s iOS. Repeat that list without dates and three of your five platform recommendations point at dead products.
Second, do not repeat the claim that “the phone is just a display.” What the sources support is narrower than that: processing moves off the device, and the device keeps its storage, its apps and its data. Offloading is a decision each app makes, and the borrowed computation is repaid in transmissions and wide-area latency, which is why a phone that renders nothing locally can still feel slow on a weak radio.
Third, know what you are locked into. The MCC challenge list names vendor and data lock-in alongside security and privacy, and by now the reason is concrete: you use the cloud for mail, photos, notes, feeds, music and games, and a handful of other companies hold the back ends. Knowing which company holds which category is the practical version of that caution, and this article’s day tour is the inventory.
In short
You use the cloud every day, from mail in a browser at breakfast to a game rendered on a remote server at night, and the arrangement repeats: your device is a thin client, the application and its data live on a provider’s machines, and the provider’s data center does the work your device displays. The 2010s product shelf expired while its categories carried on, so the question that travels is which back end a new app is a window onto. On the phone, mobile cloud computing trades local computation for transmissions, a good deal until the radio link is the bottleneck.
Board 13 of 13: the recap, sticky note and all.
Continue learning
- Containers vs Virtual Machines: The Problem Docker Solves: the sibling Whiteboard Architect companion, on the unit of software that makes those provider back ends reproducible.
- How Spotify Works: the music stop of the day tour, opened up, from its audio CDN to the nightly run that refreshes recommendations.
- Cloud Computing Interview Questions for 2026: the definitions from this episode in the form interviewers actually ask them.
Disclosure: this article was drafted with AI from the script of my video lesson, then checked against its sources before publishing.
This lesson first appeared on iqraa.tech.
More whiteboard lessons on the Whiteboard Architect YouTube channel.
Top comments (0)