Monday, June 23, 2008

Apple Juice

Since rumors began in May to the effect Apple planned a major online services initiative, I've been revisiting my own experience with Apple's online services. Apple's recently-announced MobileMe will offer mail, contacts, online storage, multi-machine synchronization through a push-style system licensed from Microsoft, and will support third parties who want to get notifications sent to Apple products without their needing to continuously poll for data. It sounds great.

When I first signed up for Apple's ".Mac Services" it was free. It wasn't much more than an email account, but it offered a great promise to folks who'd been stuck with ISPs they wanted to ditch merely because they had no way to get a fired ISP to forward email to a new address, and they didn't want to miss all the email they (you may well laugh) so treasured. Apple's promise was simple: you will get an email address you can keep When Apple deployed .mac, it was publicized with the slogan “free email for life.” This meant that whoever you picked as an ISP (or regardless when you graduate from the school offering you your current ".edu" account) you would never be held hostage by an ISP (which ISP might also, to Apple's irritation, offer support only for products made by Apple's competitors; this used to be a bigger deal than now, so eliminating email-lock was good business).

Folks signed up in droves, and me with them. And when folks bitched about the quality, I chuckled: what did you expect? It's free. Anyone wanting higher performance could go buy it.

Later, when Apple began offering online storage and a backup tool, I set it up to copy stuff I considered important enough I wanted it available offsite. Given my experience with HFS+ filesystem corruption under MacOS 9.0.4 (which self-fragmented until the directory structures were too fragmented to allow the machine to boot itself; this was a short-lived bug fixed in later versions of Apple's HFS+ drivers), I was highly motivated to use the service. Besides -- everyone needs a good backup plan, and this automated scheme seemed a good idea.

I wasn't altogether happy with the service, but it'd been free. When Apple announced it'd charge money for the service, I assumed this was intended to allow Apple to invest more in the service and to improve throughput (which had been mediocre) and reliability (which hadn't impressed me). Alas. After paying for one discounted year, and failing to instruct Apple at the right time to ensure I didn't get auto-charged for the next year, I ended up with two years of experience that Apple's service was not worth the $99 Apple demanded.

Mind you, by then I was an Apple shareholder, and I was all in favor of Apple fleecing anyone who couldn't tell $99 was a bad deal, but I sure didn't mean Apple should be getting my money in exchange for an email service that frequently fell victim to outages, or an online storage service whose throughput (amount of data for the time spent moving it) and latency (the amount of time between asking Apple's servers to do something, and getting any evidence Apple heard or cared) were so abysmal I thought they might be running on someone's hand-me-down 386. So I wasn't about to pay $99 for stuff that was easily had elsewhere for less (and at better quality).

Worst was the Backup application. It gave a progress bar that bore no relationship to the likely time until done. It didn't admit failure all the time, and would sit there chewing up processing power while the progress bar pretended readiness to advance one more pixel ... just ... any ... minute ... now .... The progress bar was a real treat this way: Apple eventually adopted a progress bar whose line was animated so it took some attention to realize it wasn't actually moving forward, it was just showing you a flowing animation that forever stopped at some arbitrary pixel ... that might be tantalizingly close to finished.

The (it was called) used a compressed file format that should have made things quick and fast, but it seemed to fail to finish its work quite a bit. Unlike a solution such as rsync, a free Unix tool that copies only the changed bits of files, seemed to want to make an enormous file of all the stuff you wanted backed up, and to shove this enchilada grande through the network connection to the servers Apple tasked with serving these customer requests. Apple's servers can't be said to have moved with stately grace, because they fell flat on their face to often to qualify for such a compliment: they were slow and they were unreliable. When the failed, it sometimes did it without any warning or notice. I never had an adea what was and was not backed up. I didn't know what I'd get if I tried restoring from the backup, and when I tried as a test it failed, too.

I submitted feedback patiently, expecting fixes. I submitted screen shots and made sure Apple understood how reliably failed and how utterly useless it had become to be as a user. I gave up and canceled the service. I got an email address from somebody else ( and went about my merry way.

Now, Apple offers a new online service that will cost users $99. Apple has added some multi-machine synchronization stuff to its offerings, and I've seen them and felt their bad latency while I had a post-operating-system-upgrade temporary "free trial" of dotMac. And I keep hearing from folks using Apple's consumer email service that they're having trouble getting or sending email. I have some family in another state who've repeatedly instructed me not to mail their dotMac email addresses when I need to send them things in a hurry -- directions as they race for the door, recipes before they dash out for ingredients, and so on -- because their ISP's webmail is so much more reliable than Apple's that you can actually expect to get an email the same day.

Imagine paying $99 for an online service that's so unpredictable that you have folks email your other addresses because they're faster. The mail forwarding service I use to pretend I have a one-email-forever account can't be adding any speed to the process of getting emails, but I've tried it in "Steve Jobs style" bake-offs and found it faster than Apple's offerings.

So when I heard apple had bought a massive data center and was set to offer broadened services, including support for push services expected to be used to market Apple hardware to business enterprises, I thought Apple had finally gotten serious. After all, Apple's MobileMe logo is a cloud and it stands to reason Apple wants folks to trust services provided way out in the ether if Apple wants to build steady revenue from them.

Immediately, I began hearing from friends requests via IM that I send test emails to their addresses so they could make sure the mail problems they faced weren't on their end, but Apple's. I've seen several waves of this stuff since the announcement of MobileMe. The idea that Apple has gotten serious about online services and is about to prove it has pretty much been exploded: Apple's disdain for its existing $99/year customers pretty much ensures that it hasn't made the kind of quality improvements across the board to demonstrate its commitment to service customers. Apple's current service is likely, if anything, to lose business it's previously won.

Sure, Apple will get new customers as they buy Apple hardware but don't realize Apple's quality commitment doesn't extend to online services. And if Apple doesn't change its behavior, it'll educate these folks too.

There's one bright hope on the horizon: Apple is carefully building its next-gen online service using serious high-quality methods and high-availability technollogies, and has written off its old system as so hopeless it's not worth trying to save. In this scenario, Apple will flip a switch and everyone's online requests will begin to be served by the new system, and the clouds will part over Cupertino to reveal a blissful, divine light -- and customers will get what they expected, when they expect it, and won't need to pray any longer in order to get it.

I can't say I think much of Apple's historic commitment to its dotMac customers, or its effort to make the period between the announcement and the launch trouble-free for those who are migrating from the old service to the new, but I hope for the sake of Apple's reputation that as it pushes broad use of MobileMe that Apple has finally delivered something whose quality will be a credit rather than an embarrassment to the Apple brand.

No comments: